Screenwriting

 2.0  

What  are  the  possibilities  of  screenplay  ‘datafication’?  

How  the  screenplay  as  data  can  impact  creating  and  

managing,  presenting  and  sharing,  analyzing  and  visualizing  
textual  screenplay  content.  

Stewart  McKie  

PhD  Degree  

Media  Arts  

Royal  Holloway  College,  London  

Stewart  McKie   -­‐  1  -­‐   As  of  15/08/2014  

Declaration  of  Authorship

I,  Stewart  McKie,  hereby  declare  that  this  thesis  and  the  work  presented  in  it  
is  entirely  my  own.  Where  I  have  referenced  or  consulted  the  work  of  others,  
this  is  always  clearly  stated  whether  in  the  text  of  the  thesis  or  on  the  credits  
page  of  the  Scenepad  online  application  software  deliverable.  

Signed:      

Date:   09  August  2013  

Stewart  McKie   -­‐  2  -­‐   As  of  15/08/2014  

Abstract  

Screenwriting  software  applications  have  traditionally  focused  on  the  

screenplay  as  document,  essentially  enabling  writers  to  create  screenplays  
in  an  industry  standard  format  that  can  be  printed  or  distributed  
electronically.  This  thesis  explores  the  possibilities  of  technology  delivering  
more  value  if  the  screenplay  content  is  stored  at  a  more  granular  level,  as  
data  and  metadata.  This  'datafication'  of  the  screenplay  content  enables  new  
and  interesting  possibilities  for  screenwriting  practice.    

Managing  screenplay  content  as  data  rather  than  'document'  better  reflects  
the  dynamic  nature  of  the  screenplay  and  collaborative  nature  of  
screenwriting  as  a  social  and  business  process.  Screenplay  as  data  also  
facilitates  automated  analysis  and  visualization  (‘analytics’)  of  screenplay  
content  that  can  benefit  both  screenwriters  and  other  stakeholders  in  the  
screenwriting  process.  This  thesis  focuses  specifically  on  the  topics  of  
screenplay  as  data,  the  screenplay  as  social  network  and  screenplay  
analytics.    

The  practical  submissions  that  complement  this  thesis  comprise  a  number  of  
online  application  prototypes  that  explore  some  of  the  propositions  
discussed  in  the  thesis  argument.  My  final  practical  submission  -­‐  Scenepad  -­‐  
is  a  complete  online  screenwriting  application  that  is  informed  by  the  
various  prototypes  that  preceded  it.  Scenepad  puts  many  of  my  thesis  
propositions  into  practice  to  deliver  an  evolutionary  development  in  
screenwriting  technology  I  call  'Screenwriting  2.0'.  I  conclude  this  thesis  by  
outlining  some  possible  future  developments  of  the  screenplay  as  artefact,  
and  screenwriting  as  practice.  

Stewart  McKie   -­‐  3  -­‐   As  of  15/08/2014  

Stewart  McKie   -­‐  4  -­‐   As  of  15/08/2014  
Contents  
DECLARATION  OF  AUTHORSHIP  .......................................................................................................  2  

ABSTRACT  ................................................................................................................................................  3  

CONTENTS  ................................................................................................................................................  5  

FIGURES  AND  TABLES  ...........................................................................................................................  8  

ACKNOWLEDGEMENTS  .....................................................................................................................  13  

SOFTWARE  DELIVERABLES  .............................................................................................................  15  

INTRODUCTION  ...................................................................................................................................  16  

FOCUS:  THE  FEATURE  FILM  SCREENPLAY  .........................................................................................  18  
DATAFICATION  ........................................................................................................................................  21  
RESEARCH  QUESTION  ............................................................................................................................  23  
SCREENWRITING  2.0  ..............................................................................................................................  25  
SOURCES  ...................................................................................................................................................  26  
METHODOLOGY  .......................................................................................................................................  28  

1  SCREENPLAY  AS  DATA  ...................................................................................................................  52  

1.1  THE  DATA-­‐CENTRIC  SCREENPLAY  ...............................................................................................  54  
1.2  DATA  AND  METADATA  ...................................................................................................................  55  
1.3  UNSTRUCTURED  VS.  STRUCTURED  DATA  ...................................................................................  57  
1.4  TABULATED  SCREENPLAYS  ............................................................................................................  64  
1.5  DOCUMENTED  SCREENPLAYS  ........................................................................................................  66  
1.6  TAGGED  SCREENPLAYS  ...................................................................................................................  68  
1.7  XBRL  AND  CORPORATE  REPORTING  ...........................................................................................  72  
1.8  ANALYTIC  METADATA  FOR  SCREENWRITERS  ............................................................................  76  
1.9  SCREENPLAY  ELEMENTS  ................................................................................................................  78  
1.10  SCREENPLAY  DECOMPOSITION  ..................................................................................................  84  
1.11  SCREENPLAYS  AS  ‘BIG  DATA’  ......................................................................................................  90  
1.12  SCREENPLAY  APIS  ........................................................................................................................  93  
1.13  SCREENPLAY  AS  DATA  WALL  OR  STREAM  ...............................................................................  95  
1.14  SCREENPLAY  AS  INTERFACE  .......................................................................................................  95  
SUMMARY  .................................................................................................................................................  96  

2  SCREENPLAY  ANALYTICS  ..............................................................................................................  98  

Stewart  McKie   -­‐  5  -­‐   As  of  15/08/2014  

 2.1  WHY  BOTHER  ANALYZING  SCREENPLAY  CONTENT?  ...............................................................  106  
2.2  WHAT  IS  WORTH  ANALYZING?  ..................................................................................................  116  
2.3  CHARACTER  ....................................................................................................................................  119  
2.4  NARRATIVE  .....................................................................................................................................  133  
2.5  STRUCTURE  .....................................................................................................................................  140  
2.6  ANALYTIC  ARCHETYPES  ...............................................................................................................  144  
2.7  GENERIC  DATA  ANALYSIS  AND  VISUALIZATION  TOOLS  .........................................................  153  
2.8  USING  ‘BIG  DATA’  TOOLS  FOR  SCREENPLAY  ANALYSIS  .........................................................  165  
SUMMARY  ...............................................................................................................................................  173  

3  SCREENPLAY  AS  SOCIAL  NETWORK  ........................................................................................  175  

3.1  FUNCTIONALITY  OF  ONLINE  SOCIAL  NETWORKING  ...............................................................  178  
3.2  SCREENPLAY  STAKEHOLDERS  .....................................................................................................  182  
3.3  SCREENWRITING  AS  BUSINESS  PROCESS  ..................................................................................  187  
3.4  SCREENPLAY  AS  SOCIAL  NETWORK  ANCHOR  ..........................................................................  196  
SUMMARY  ...............................................................................................................................................  198  

4  SCREENPLAY  AS  DESIGN  ..............................................................................................................  199  

4.1  DATA-­‐DRIVEN  SCREENPLAY  MANUFACTURING  ......................................................................  202  
4.2  SCREENPLAY  MODELS  ..................................................................................................................  208  
4.3  SCREENPLAY  PATTERN  LANGUAGE  ...........................................................................................  210  
4.4  REVERSE-­‐ENGINEERING  SCREENPLAY  DESIGN  .......................................................................  219  
4.5  EVENT  INDICATORS  ......................................................................................................................  224  
SUMMARY  ...............................................................................................................................................  227  

5.  THE  ROAD  TO  SCENEPAD  ...........................................................................................................  228  

5.1  SCREENPLAY  ANALYTICS  TODAY  ................................................................................................  231  
5.2  SOCIAL  SCREENWRITING  TODAY  ................................................................................................  243  
5.3  SCREENWRITING  2.0  PROTOTYPES  ...........................................................................................  244  
5.4  SCRIPTCLOUD  [2006]  ..................................................................................................................  246  
5.5  CONTENTCLOUD  [2008]  .............................................................................................................  253  
5.6  SCRIPTGEIST  [2007]  ...................................................................................................................  256  
5.7  SCENECLASS  [2010]  ....................................................................................................................  261  
SUMMARY  ...............................................................................................................................................  266  

6  SCENEPAD  ........................................................................................................................................  267  

6.1  SCENEPAD  –  SCREENPLAY  AS  DATA  ..........................................................................................  268  
6.2  SCENEPAD  -­‐  SCENEWRITING  .......................................................................................................  271  

Stewart  McKie   -­‐  6  -­‐   As  of  15/08/2014  

 6.3  SCENEPAD  AS  SOCIAL  NETWORK  ...............................................................................................  278  
6.4  SCENEPAD  ANALYTICS  ..................................................................................................................  289  
6.5  SCENEPAD  FOR  BUSINESS  ............................................................................................................  298  
6.6  SCREENPLAY  PRODUCTION  ..........................................................................................................  299  
SUMMARY  ...............................................................................................................................................  300  

7.  WHAT  HAPPENS  NEXT?  ..............................................................................................................  302  

BIBLIOGRAPHY  ..................................................................................................................................  309  

APPENDIX  A:  HOFFMAN’S  HELLO  WORLD  XBRL  EXAMPLE  ................................................  335  

APPENDIX  B:  SELECTIVE  SCENEPAD  FEATURE  LIST  .............................................................  338  

APPENDIX  C:  SCREENPLAY  ANALYTICS  –  A  USE  CASE  ...........................................................  340  

1.  ON-­‐SCREEN  TIME  ............................................................................................................................  343  
2.  DIALOG  AND  ACTION  .......................................................................................................................  346  
3.  WHAT  THE  PROTAGONIST  SAYS  AND  DOES  ...............................................................................  347  
4.  PROTAGONIST  &  ANTAGONIST  ROLE  METADATA  .....................................................................  349  
5.  PROTAGONIST  SOCIAL  NETWORK  .................................................................................................  351  
6.  PROTAGONIST  SENTIMENT  ............................................................................................................  352  
7.  PROTAGONIST  VS.  ANTAGONIST  ....................................................................................................  353  
CONCLUSION  ..........................................................................................................................................  356  

Stewart  McKie   -­‐  7  -­‐   As  of  15/08/2014  

Figures  and  Tables  
 
FIGURE  1.1  -­‐  GOOGLE  NGRAM  VIEWER  PLOTS  SPECIFIC  NAME  MENTIONS  OVER  TIME  ......................................  22  
FIGURE  1.2  –  COMMENT  POSTED  BY  ALEX  EPSTEIN  ON  HIS  COMPLICATIONS  ENSUE  BLOG  ...............................  38  
FIGURE  1.3  –  COMMENT  POSTED  BY  A  STAGE32.COM  USER  ON  THEIR  BLOG  AND  SENT  TO  ME  BY  EMAIL  ON  
31  DECEMBER  2013  .........................................................................................................................................  39  
TABLE  1.1  –  FUNCTIONAL  EVOLUTION  OF  PROTOTYPE  DELIVERABLES  ................................................................  51  
TABLE  1.2  –  TEXT  FILE  REPRESENTATIONS  ..............................................................................................................  59  
TABLE  1.3    -­‐  ROW/COLUMN  INTERSECTION  ............................................................................................................  62  
FIGURE  1.2  -­‐  SCRIPT  AS  TABLES  (SIMPLIFIED)  .........................................................................................................  65  
FIGURE  1.3  -­‐  SCRIPT  AS  DOCUMENTS  (SIMPLIFIED)  .................................................................................................  68  
FIGURE  1.4  -­‐  AN  EXAMPLE  OF  A  SCREENPLAY  HIERARCHY  MODEL  .....................................................................  71  
FIGURE  1.5  –  TYPICAL  SCREENPLAY  ELEMENTS  ......................................................................................................  80  
FIGURE  1.6  -­‐  CHATMAN’S  NARRATIVE  ENTITY  CLASSES  ........................................................................................  87  
FIGURE  1.7  –  QUALITY  PYRAMID  OF  CROWDSOURCED  DATA  ................................................................................  93  
FIGURE  2.1  -­‐  FREYTAG’S  PYRAMID  ...........................................................................................................................  101  
FIGURE  2.3  -­‐  WATCHING  THE  SCRIPT  ......................................................................................................................  102  
FIGURE  2.4  -­‐  BRANDWATCH  SOCIAL  OSCARS  DATAVIZ  ........................................................................................  107  
FIGURE  2.5A  –  STORY  CHART  FOR  CASABLANCA  ...................................................................................................  115  
FIGURE  2.5B  –  MCKEE’S  STORY  CHART  OF  PULP  FICTION.  ..................................................................................  116  
TABLE  2.1  -­‐  POTENTIAL  SCREENPLAY  ANALYSIS  TOPICS  .....................................................................................  117  
FIGURE  2.6  -­‐  THE  REFORMER  ENNEAGRAM  STYLE  ...............................................................................................  123  
FIGURE  2.7  -­‐  PROTAGONIST-­‐DRIVEN  CAST  DESIGN  (MCKEE  1998:380)  .......................................................  125  
FIGURE  2.8  -­‐  CHARACTER  CONFLICT  AS  STORY  (ADAPTED  FROM  BABOULENE  2012:20)  .............................  125  
FIGURE  2.9  –  SOCIAL  NETWORK  OF  HAMLET  (SCENE  VI)  ...................................................................................  127  
FIGURE  2.10  –  CHARACTER  WEIGHTS  IN  BUFFY  ....................................................................................................  128  
FIGURE  2.11  -­‐  DYNAMIC  CHARACTER  MATRIX  ......................................................................................................  133  
FIGURE  2.12  -­‐  PULP  FICTION  NARRATIVE  LINES  ...................................................................................................  138  
FIGURE  2.13  -­‐  FLASHBACK  AS  THWARTED  DREAM:  SHINE  .................................................................................  139  
FIGURE  2.14  -­‐  FIELD’S  THREE-­‐ACT  STRUCTURE  ...................................................................................................  146  
FIGURE  2.15  -­‐  CONVERGING  PLOT  LINES  IN  THE  DAY  OF  THE  JACKAL  (POPE,  1998:173)  ..........................  147  
FIGURE  2.16  -­‐  VOGLER’S  HERO’S  JOURNEY  (VOGLER,  1994:194)  ...................................................................  148  
FIGURE  2.17  -­‐  MURDOCK’S  HEROINE’S  JOURNEY  (MURDOCK,  1990:5)  ..........................................................  148  
FIGURE  2.18  -­‐  DIE  HARD  ANALYSIS  (VOYTILLA,  1999:41)  ...............................................................................  149  
FIGURE  2.19A  -­‐  ROLLERCOASTERS  OF  ESCALATING  TENSION  (COOPER,  1997:71)  ......................................  150  
FIGURE  2.19B  -­‐  PAUL  AND  JACK’S  FRACTURED  TANDEM  IN  21  GRAMS  (ARONSON,  2010:293)  ................  151  
FIGURE  2.20  -­‐  SCREENPLAY  STRUCTURE  AS  NODE  TREE  (KITCHEN,  2006:162)  ..........................................  152  

Stewart  McKie   -­‐  8  -­‐   As  of  15/08/2014  

FIGURE  2.21  -­‐  CHARACTERS  IN  THE  THIRD  MAN  (MACKENDRICK,  2005:19)  ...............................................  153  
FIGURE  2.22  -­‐  ARC  DIAGRAM  APPLIED  TO  MY  SCREENPLAY  CARPINI  ................................................................  155  
FIGURE  2.23-­‐  A  TOPIC  FLOWER  APPLIED  TO  A  SCREENPLAY  ..............................................................................  157  
FIGURE  2.24  -­‐  A  SHARED  WORD  DIAGRAM  APPLIED  TO  TWO  DRAFTS  OF  A  SCREENPLAY  ............................  158  
FIGURE  2.25  -­‐  NAMING  NAMES  .................................................................................................................................  159  
FIGURE  2.26  -­‐  TRANSCRIPT  ANALYZER  ...................................................................................................................  160  
FIGURE  2.27  –  TRANSCRIPT  ANALYSIS  ....................................................................................................................  161  
FIGURE  2.28  –  ARGO  WORD  CLOUD  ........................................................................................................................  163  
FIGURE  2.29A  -­‐  STARSKY  AND  HUTCH  STORYLINES  ..............................................................................................  164  
FIGURE  2.29B  -­‐  HILL  STREET  BLUES  STORYLINES  ................................................................................................  164  
FIGURE  2.29.C  -­‐  THE  SOPRANOS  STORYLINES  ........................................................................................................  165  
FIGURE  2.30  –  SPLUNK  SNIPPET  ANALYSIS  ............................................................................................................  166  
FIGURE  2.31  –  SPLUNK  WORD  ANALYSIS  ...............................................................................................................  166  
FIGURE  2.32  –  SPLUNK  ROLE  ANALYSIS  ..................................................................................................................  167  
FIGURE  2.33  –  SPLUNK  LOCATION  ANALYSIS  .........................................................................................................  168  
FIGURE  2.34  –  BIGML  SCRIPT  ANALYSIS  ................................................................................................................  169  
FIGURE  2.35  –  BIGML  SCENE  PREDICTION  ............................................................................................................  170  
FIGURE  2.36  –  BIGML  GENDER  PREDICTION  .........................................................................................................  171  
FIGURE  2.37  –  DATASEED  ANALYSIS  .......................................................................................................................  172  
FIGURE  2.38  –  SAMPLE  OF  UPLOADED  DATA  FILE  ................................................................................................  173  
TABLE  3.1  –  SOCIAL  NETWORK  CHARACTERISTICS  ...............................................................................................  177  
FIGURE  3.1  -­‐  THE  3  PRINCIPAL  SCREENPLAY-­‐ANCHORED  SOCIAL  NETWORKS  ..................................................  178  
FIGURE  3.2A  -­‐  WRITE  –  REFLECT  -­‐  REWRITE  .........................................................................................................  191  
FIGURE  3.2B  -­‐  THE  ENRICHED  SCREENWRITING  PROCESS  ...................................................................................  192  
FIGURE  4.1  –  IAN  BARRY’S  SCRIPT  BOOK  ................................................................................................................  200  
FIGURE  4.2  -­‐  A  LINED  SCRIPT  IN  SCRIPT  SUPERVISOR  ..........................................................................................  201  
FIGURE  4.3  -­‐  SCREENPLAY  AS  ADDITIVE  LAYERS  ...................................................................................................  206  
FIGURE  4.4  –  GAULD’S  STRUCTURAL  JOURNEY  .......................................................................................................  206  
FIGURE  4.5  –  YORKE’S  PARADIGM  SUMMARY  .........................................................................................................  209  
TABLE  4.1  –  PATTERN  ATTRIBUTES  ........................................................................................................................  212  
TABLE  4.2  –  3-­‐ACT  STRUCTURE  PATTERN  .............................................................................................................  213  
TABLE  4.3  –  PROTAGONIST  PATTERN  ......................................................................................................................  215  
FIGURE  4.6  –  ALIEN  AS  VISUALIZED  BY  MILESI  &  MATTEO  (2012:9)  ..............................................................  218  
FIGURE  4.7  -­‐  INSPECTOR  MONTALBANO  INFOGRAPHIC  ........................................................................................  219  
FIGURE  4.8  -­‐  STOVIZ  STORYLINE  VISUALIZER  ........................................................................................................  221  
FIGURE  4.9  –  CHUN’S  STORY  VISUALIZER  ...............................................................................................................  223  
FIGURE  5.1  -­‐  WORKING  WITH  A  FORMATTED  SCRIPT  IN  CELTX  ..........................................................................  232  
FIGURE  5.2  -­‐  NAVIGATING  A  SCRIPT  HIERARCHY  IN  SOPHOCLES  ........................................................................  233  
FIGURE  5.3  -­‐  MANAGING  INDEX  CARDS  IN  SUPERNOTECARD  ..............................................................................  234  

Stewart  McKie   -­‐  9  -­‐   As  of  15/08/2014  

FIGURE  5.4  -­‐  SCRIPT  ANALYSIS  CHARTS  IN  SOPHOCLES  .......................................................................................  235  
FIGURE  5.5  -­‐  THE  SOCIAL  NETWORK  OF  CASABLANCA  IN  SOPHOCLES  ..............................................................  237  
FIGURE  5.6  –  PLOTBUILDER  EXCITEMENT  GRAPH  ................................................................................................  238  
FIGURE  5.7  -­‐  THE  STORYVIEW  STRUCTURAL  VIEW  OF  PULP  FICTION  ...............................................................  239  
FIGURE  5.8  -­‐  THE  STORYVIEW  SCRIPT  WALL  POSTER  .........................................................................................  239  
FIGURE  5.9  -­‐  A  POSSIBLE  TIMELINE,  STRUCTURE  &  THROUGHLINE  VISUALIZATION  ........................................  240  
FIGURE  5.10  -­‐  SCRIPT  STORYBOARD  IN  CELTX  ......................................................................................................  241  
FIGURE  5.11  –  AMAZON  STORYTELLER  STORYBOARD  ..........................................................................................  242  
TABLE  5.1  –  LIST  OF  PROTOTYPES  ...........................................................................................................................  246  
FIGURE  5.12  –  SCRIPTCLOUD  STATISTICS  JULY  6,  2008  .....................................................................................  248  
FIGURE  5.13  –  SCRIPTCLOUD.COM  HOME  PAGE  ....................................................................................................  248  
FIGURE  5.14  -­‐  SCRIPTCLOUD  OF  MY  OWN  CARPINI  SCRIPT  .................................................................................  250  
FIGURE  5.15  -­‐  THE  GODFATHER  SCRIPTCLOUD  ......................................................................................................  252  
FIGURE  5.16  -­‐  THE  GODFATHER  KEYWORDS  ..........................................................................................................  252  
FIGURE  5.17  -­‐  THE  FRENCH  CONNECTION  SCRIPTCLOUD  .....................................................................................  253  
FIGURE  5.18  -­‐  CONTENTCLOUDS  HOME  PAGE  .......................................................................................................  254  
FIGURE  5.19  -­‐  COMPARING  GODFATHER  SCRIPTS  .................................................................................................  255  
FIGURE  5.20  -­‐  COMPARING  ACTS  IN  SHAKESPEARE’S  MACBETH  ........................................................................  256  
FIGURE  5.21  -­‐  SCRIPTGEIST  HOME  PAGE  ................................................................................................................  257  
FIGURE  5.22  -­‐  SCRIPTGEIST  STATISTICS  ..................................................................................................................  258  
FIGURE  5.23  -­‐  THE  GODFATHER  DIALOG  CLOUD  ....................................................................................................  259  
FIGURE  5.24  -­‐  THE  GODFATHER  CHARACTER  THROUGHLINE  ..............................................................................  259  
FIGURE  5.25  -­‐  THE  GODFATHER  INT/EXT  SCENES  ..............................................................................................  260  
FIGURE  5.26  -­‐  THE  GODFATHER  CHARACTERS  AND  PART  SIZE  ...........................................................................  260  
FIGURE  5.27  -­‐  THE  GODFATHER  CHARACTER  DIALOG  ..........................................................................................  261  
FIGURE  5.28  -­‐  SCENECLASS  HOME  PAGE  .................................................................................................................  262  
FIGURE  5.29  –  SCENECLASS  CLASS  ENTITY  ............................................................................................................  263  
FIGURE  5.30    -­‐  SCENECLASS  SCENE  ENTITIES  ........................................................................................................  263  
FIGURE  5.31  –  SCENECLASS  SCENE  COMMENTS  ....................................................................................................  264  
FIGURE  5.32  –  SCENECLASS  STUDENT  RESPONSES  ...............................................................................................  265  
FIGURE  6.1  -­‐  ORIGINAL  SCENEPAD1  BETA  HOME  PAGE  .......................................................................................  268  
FIGURE  6.2  –  CORE  SCENEPAD  DATA  ENTITIES  (TABLES)  ..................................................................................  270  
FIGURE  6.3  -­‐  MARINER  STORYMILL  SCENE  SNAPSHOTS  ........................................................................................  273  
FIGURE  6.4  -­‐  SCENEPAD1  SCENE  VERSIONING  ........................................................................................................  274  
FIGURE  6.5  –  SCENEPAD3  –  SCENE  CONTENT  TYPES  AND  ANALYSIS  .................................................................  275  
FIGURE  6.6  –  SCENEPAD4  MANAGING  SCENE  GROUPING  .....................................................................................  275  
FIGURE  6.7  -­‐  SCENEPAD3  HERO’S  JOURNEY  TEMPLATE  .......................................................................................  276  
FIGURE  6.8  –  SCENEPAD1  ACT  AND  SEQUENCE  NAVIGATOR  ...............................................................................  277  
FIGURE  6.9  –  SCENEPAD4  SCENE  STORYLINE  MANAGER  .....................................................................................  277  

Stewart  McKie   -­‐  10  -­‐   As  of  15/08/2014  

FIGURE  6.10  –  SCENEPAD4  SCENE  STORYBOARD  ..................................................................................................  278  
FIGURE  6.11  –  SCENEPAD4  GROUPS  .........................................................................................................................  279  
FIGURE  6.12  -­‐  SCENEPAD4  GROUP  USERS  ...............................................................................................................  280  
FIGURE  6.13    -­‐  SCENEPAD4  SCRIPT  MANAGER  CONTENT  HUB  ...........................................................................  282  
FIGURE  6.14    -­‐  SCENEPAD4  ROLE  GALLERY  ............................................................................................................  282  
FIGURE  6.15  –  SCENEPAD4  SCENE  COMMENTING  .................................................................................................  284  
FIGURE  6.16  -­‐  SCENEPAD4  SCRIPT  QR  CODE  ..........................................................................................................  285  
FIGURE  6.17  -­‐  SCENEPAD4  MOBILE  SITE  HOME  PAGE  ...........................................................................................  286  
FIGURE  6.18  -­‐  SCENEPAD2    -­‐  SHOWING  TUMBLR-­‐SOURCED  IMAGES  DISPLAYED  ON  SCENEPAD  ROLE  WALL287  
FIGURE  6.19  -­‐  SCENEPAD4  -­‐  SHOWING  IMAGES  SOURCED  FROM  KEE.PS  DISPLAYED  UNDER  SCRIPT  KEEPS  
TAB  ......................................................................................................................................................................  288  

FIGURE  6.20  -­‐  SCENEPAD4    -­‐  SHOWING  TWEETS  SOURCED  FROM  TWITTER  DISPLAYED  UNDER  SCRIPT  
TWEETS  TAB  ......................................................................................................................................................  288  
FIGURE  6.21  -­‐  SCENEPAD1  SCRIPT  DASHBOARD  ....................................................................................................  290  
FIGURE  6.22  -­‐  SCENEPAD1  SCENE  DASHBOARD  .....................................................................................................  290  
FIGURE  6.23  -­‐  SCENEPAD4  ROLE-­‐DIALOG  ...............................................................................................................  291  
FIGURE  6.24  -­‐  SCENEPAD1  TOP  10  ROLES  .............................................................................................................  292  
FIGURE  6.25  -­‐  SCENEPAD1  PROTAGONIST  VS.  ANTAGONIST  ................................................................................  292  
FIGURE  6.26  -­‐  SCENEPAD1  TOP  10  LOCATIONS  .....................................................................................................  293  
FIGURE  6.27  -­‐  SCENEPAD1  SCRIPT-­‐LEVEL  WORD  CLOUDS  ..................................................................................  294  
FIGURE  6.28  -­‐  SCENEPAD3  -­‐  ALIEN  KEYROLE  THRULINE  .....................................................................................  296  
FIGURE  6.29  -­‐  SCENEPAD3  ALIEN  END  OF  SCRIPT  THRULINE  .............................................................................  296  
FIGURE  6.30  -­‐  SCENEPAD4  DIALOG  SENTIMENT  ANALYSIS  .................................................................................  297  
FIGURE  6.31  –  SCENEPAD4  MANAGING  SCRIPT  COVERAGE  .................................................................................  298  
FIGURE  6.32  -­‐  SCENEPAD4  SCRIPT  OWNERSHIP  DASHBOARD  .............................................................................  299  
FIGURE  6.33  -­‐  SCENEPAD4  SCRIPT  DAY  CONTENT  HUB  .......................................................................................  300  
FIGURE  C1  –  SCRIPT  HOME  STATISTICS  ...................................................................................................................  341  
FIGURE  C2  –  SCRIPT  ROLES  .......................................................................................................................................  342  
FIGURE  C3  –  ROLE  HOME  STATISTICS  .....................................................................................................................  342  
FIGURE  C4  –  ROLE  FAQS  ...........................................................................................................................................  343  
FIGURE  C5  –  TOP  5  ROLES  BY  SCENE  (NOTE:  EACH  DOT  REPRESENTS  A  UNIQUE  SCENE)  ..............................  344  
FIGURE  C7  –  TOP  5  ROLES  BY  EXIT  SCENE  (NOTE:  EACH  DOT  REPRESENTS  A  UNIQUE  SCENE)  .....................  346  
FIGURE  C8  –  TOP  5  ROLES  BY  DIALOG  SNIPPETS  (NOTE:  EACH  DOT  REPRESENTS  A  UNIQUE  DIALOG  
SNIPPET)  ............................................................................................................................................................  347  

FIGURE  C9  –  TOP  5  ROLE  CUES  MENTIONED  IN  ACTION  SNIPPETS  ....................................................................  347  
FIGURE  C10  -­‐  MOST  FREQUENT  NOUNS/WORDS  IN  DIALOG  SNIPPETS  SPOKEN  BY  ROLE  ............................  349  
FIGURE  C11  -­‐  MOST  FREQUENT  NOUNS/WORDS  IN  ACTION  SNIPPETS  OF  ROLE  ...........................................  349  
FIGURE  C12  –  BASIC  ROLE  METADATA  ...................................................................................................................  350  
FIGURE  C13  –  KEY  ROLE  FAQS  ................................................................................................................................  351  

Stewart  McKie   -­‐  11  -­‐   As  of  15/08/2014  

FIGURE  C15  –  MAGGIE’S  SENTIMENT  ANALYSIS  ....................................................................................................  352  
FIGURE  C16  -­‐  SCRIPT  BALANCE:  PROTAGONIST  VS.  ANTAGONIST  DIALOG  SCENES  ........................................  353  
FIGURE  C17  -­‐  SCRIPT  BALANCE:  PROTAGONIST  VS.  ANTAGONIST  DIALOG  SNIPPETS  .....................................  354  
FIGURE  C18  -­‐  SCRIPT  BALANCE:  PROTAGONIST  VS.  OTHER  CAST  SCENES  ........................................................  355  
FIGURE  C19  –  DOREEN  (ANTAGONIST)  AS  SOCIAL  NETWORK  HUB  ..................................................................  355  
FIGURE  C20  –  DIALOG  GENDER  BALANCE  ..............................................................................................................  356  

Stewart  McKie   -­‐  12  -­‐   As  of  15/08/2014  

Acknowledgements  

I  would  like  to  acknowledge  the  contribution  of  the  following  people  in  
relation  to  my  PhD  research.  

Department  of  Media  Arts,  Royal  Holloway    

The  Department  of  Media  Arts  provided  a  tuition  fee  waiver  for  the  duration  
of  my  studies  at  Royal  Holloway.    

Supervisors  

My  Royal  Holloway,  London  supervisors  Adam  Ganz  and  Fionn  Murtagh  for  
their  support  and  help  over  the  duration  of  my  research.    

Coders  and  Application  Component  Providers  

All  the  individuals  and  companies  listed  in  the  Scenepad  Credits  page  (see  
http://scenepad4.tripos.biz/credits.php)  

Scenepad  Beta-­‐Testers  

The  beta-­‐test  group  of  academics  and  students,  especially  Carmen  Sofia  
Brenes  of  the  University  of  the  Andes,  who  agreed  to  try  out  the  Scenepad  
prototype  and  the  hundreds  of  other  users  who  signed  up  to  try  out  my  
other  online  application  prototypes.  

LCACE  Funding  

London  Centre  for  Arts  and  Cultural  Exchange  (LCACE)  for  funding  the  
Scriptgeist  prototype  delivered  in  2008  based  on  an  application  made  by  
Lydia  Daniels  of  the  Royal  Holloway  Enterprise  center  and  supported  by  
Adam  Ganz.    

Conferences  

The  Semantic  Media  and  Adaptation  and  Personalization  (SMAP)  conference  

Stewart  McKie   -­‐  13  -­‐   As  of  15/08/2014  
for  allowing  me  to  present  my  paper  Scriptcloud.com:  Content  Clouds  for  
Screenplays  at  their  doctoral  consortium  in  2007.    
 
The  Screenwriting  Research  Network  (SRN)  for  allowing  me  to  make  
presentations  at  both  their  Leeds  (2008)  and  Copenhagen  (2010)  
conferences.    

Bournemouth  University  and  the  Southern  Script  Festival  for  allowing  me  to  
present  my  Sceneclass  and  Scenewrite  software  prototypes  at  their  2011  
conference.  

Others  

Sue  Clayton  and  Adam  Ganz  of  Royal  Holloway  and  the  MA  students  who  
made  use  of  the  initial  prototype  of  my  Wordpress  Sceneclass  plugin  during  
one  of  their  residential  program  classes.    

The  Faculty  and  students  of  the  department  of  Media  Arts  at  Royal  Holloway  
for  their  interest  in  my  work  at  the  annual  PhD  ‘Awayday’  presentations.  

Students  of  Shaftesbury  School  in  Dorset,  who  worked  with  me  to  produce  a  
series  of  short  films  from  an  after-­‐school  scriptwriting  class  I  ran,  funded  by  
the  Shaftesbury  School  Foundation.    

John  Foster  of  Bournemouth  University  who  allowed  me  to  present  my  
Scenewrite  prototype  to  his  MA  Screenwriting  students  and  discuss  it  with  
them.  

Karl  Phillips  of  University  College  Falmouth  who  allowed  me  to  present  to  
his  BA  Filmmaking  students  on  the  subject  of  screenplay  analytics  and  
provided  £500  in  seed  funding  for  the  development  of  Scenepad.  

Stewart  McKie   -­‐  14  -­‐   As  of  15/08/2014  

Software  Deliverables  

The  assessment  deliverable,  Scenepad,  is  an  online  software  application  for  
screenwriters  and  screenplay  management  that  applies  many  of  the  
propositions  of  this  thesis  into  practice.  

Scenepad  is  located  at  http://scenepad4.tripos.biz  

A  second  deliverable,  ScriptFAQ,  focused  purely  on  screenplay  analytics  was  

completed  during  my  PhD  ‘modification’  period  and  can  be  found  here:  

http://scriptfaq.tripos.biz  

The  generic  assessment  login  for  both  applications  is:  

Username   [email protected]  

Password   mckiephd2013  

To  review  other  software  prototypes  see    ‘Road  to  Scenepad’  section.  

Stewart  McKie   -­‐  15  -­‐   As  of  15/08/2014  

Introduction  

In  What  Happens  Next.  A  History  of  American  Screenwriting,  Marc  Norman  

ends  with  a  statement  emphasizing  the  uncertain  impact  of  new  
technologies  on  the  practice  of  screenwriting:  

It’s  pointless  to  spend  much  time  speculating  on  what  lies  in  the  future  
for  Hollywood  and  its  screenwriters,  the  result  of  new  technologies  
and  technologies  not  yet  imagined  on  the  narratives...  It’s  a  question  of  
what  happens  next.  (2008:485)  

In  a  similar  vein,  Steven  Maras  in  Screenwriting:  History,  Theory  and  Practice  
also  recognizes  challenging  times  ahead  for  the  practice  of  screenwriting:  

Today,  screenwriting  seems  on  the  brink  of  a  significant  

transformation.  While  digital  technology  and  changes  in  film-­‐making  
techniques  may  prompt  reflection  about  what  writing  for  the  screen  
means,  established  ways  of  thinking  about  screenwriting  continue  to  
structure  or  ‘haunt’  debates  about  screenwriting,  and  perhaps  limit  a  
more  pluralistic  understanding  of  writing.  (2009:186):  

And  in  The  screenplay  as  prototype,  Millard  recognizes  that:  

The  rise  of  new  technologies  and  networks  means  that  writing  now  
happens  primarily  in  digital  environments”  and  “new  
conceptualizations  of  writing  suggest  a  more  fluid  set  of  processes  than  
traditional  models  of  script  development  employ  or  allow.    
(2011:143-­‐144)  

These  and  other  recent  commentators  on  the  screenplay  and  the  practice  of  
screenwriting  recognize  that  technology  can  and  is  having  an  impact  both  on  
our  concept  of  the  screenplay  as  an  artefact  and  the  process  of  screenwriting  
as  a  practice.  But  in  reality,  while  new  technologies  are  revolutionizing  the  
production,  editing,  special  effects  and  supply-­‐chain  distribution  of  digital  
movies  for  example  –  screenplays  and  screenwriting  seem  to  be  stuck  in  a  
technology  time  warp.      

I  propose  that  a  key  reason  for  this  ‘stuckness’  is  the  failure  of  screenwriting  
practice  to  move  away  from  screenplay  as  document  and  towards  
screenplay  as  data,  a  transition  that  is  happening  in  other  areas  of  business  
Stewart  McKie   -­‐  16  -­‐   As  of  15/08/2014  
content  management.  For  example  in  section  1.7  below,  I  refer  specifically  to  
corporate  annual  reporting,  where  the  transition  from  submission  of  
digitized  documents  (e.g.  in  HTML  or  PDF  formats)  to  documents  
decomposed  into  machine-­‐readable  data  points  is  gaining  acceptance  and  
delivering  benefits  both  to  producers  and  consumers  of  this  content.      

The  document-­‐centric  bias  of  screenwriting  is  also  evident  in  screenwriting  
technology  (i.e.  screenwriting  software  applications  used  by  practicing  
screenwriters  for  writing  screenplays).  Over  the  last  couple  of  decades,  since  
its  emergence  in  the  1980s,  the  functional  focus  of  screenwriting  technology  
has  been  on  a  specialist  form  of  word  processing  for  screenwriters.  By  which  
I  mean  providing  the  functionality  for  a  screenwriter  to  use  a  software  
application  specifically  to  write  and  revise  a  script,  to  format  it  for  printing  
according  to  industry-­‐standard  formatting  conventions,  to  view  and  
navigate  the  content,  and  to  print  and  export  that  content.  Or  put  another  
way,  a  focus  on  the  ‘digitization’  of  screenplays  into  digital  documents.  

Over  time,  this  initial  functional  focus  of  screenwriting  applications  on  
writing,  formatting,  viewing  and  publishing  screenplay  content,  has  been  
supplemented  by  new  application  capabilities  to  support  both  ‘upstream’  
and  ‘downstream’  screenwriting  process  activities.  Upstream  activities  
include  story  research,  planning  and  outlining;  downstream  activities  
include  linking  storyboards  to  the  script,  and  generating  movie  pre-­‐
visualizations  or  production  breakdown  reporting  directly  from  the  script  
content.    

However  there  has  been  minimal  attention  paid  to  treating  and  storing  
screenplay  content  as  data  even  with  the  release  of  many  new  screenwriting  
applications  during  the  last  decade.  Over  the  7  year  period  I  have  been  
researching,  the  number  of  screenwriting  applications  available  to  buy  or  
available  for  free  has  at  least  doubled  and  many  screenwriting  applications  
are  now  delivered  in  new  ways,  that  leverage  new  form  factors  and  devices  
(i.e.  the  smartphone  and  tablet).    

Stewart  McKie   -­‐  17  -­‐   As  of  15/08/2014  

Certainly  the  widespread  availability  of  a  digital  screenplay  form  (notably  
Adobe’s  PDF  file  format)  has  already  changed  the  way  screenplays  are  
viewed,  navigated,  searched,  submitted  to  agents  and  competitions  and  
distributed  to  workgroups.  And  the  availability  of  screenwriting  software  on  
new  devices  and  form  factors  has  changed  how,  when  and  where  
screenplays  can  be  written  and  managed.  But  fundamentally  we  are  still  far  
from  any  kind  of  paradigm-­‐shift  to  something  that  might  truly  be  considered  
‘Screenwriting  2.0’  (to  adapt  the  ‘Web  2.0’  term  popularized  by  content  
publisher  Tim  O’Reilly).    

Focus:  The  Feature  Film  Screenplay  

My  primary  content  focus  is  the  feature  film  screenplay,  rather  than  
screenplays  intended  for  television  (TV)  or  computer  gaming  or  animation  
for  example.  It  should  also  be  noted  here  that  any  reference  made  to  a  
screenplay  or  script  generally  means  the  “author’s”  or  “selling”  or  “spec”  
(speculative)  script  and  not  the  “shooting”  script  used  in  production,  which  
may  be  substantially  different  in  content  and  format  than  the  original  
screenplay  purchased  from  the  screenwriter,  nor  any  final  transcript(s)  
produced  from  the  released  movie.    
 
One  of  the  earliest  definitions  of  a  screenplay  (made  by  Barker  in  1914  as  
quoted  in  Karetnikova  1990:5)  is  a  ‘Photoplay  or  Movie  Picture  play  is  a  
story  told  in  pictures.’    The  Palmer  Plan  Handbook  (1919:17),  describes  
photoplay  writing  as  a  ‘new  art’  that  demands  a  ‘new  language,  a  new  
method  of  expressing  thought  and  communicating  emotion’.  These  early  
commentators  clearly  recognized  the  need  to  write  visual  stories  to  suit  a  
visual  medium.  Karetnikova  also  highlights  that  the  earliest  scripts,  such  as  
George  Méliès  A  Trip  to  the  Moon  (1902),  were  little  more  than  a  list  of  
outline  story  headings  (1990:2).  The  role  of  photoplay-­‐  or  scenario-­‐writer  
(now  screenwriter)  was  already  well-­‐established  in  the  second  decade  of  the  
20th  century  when  Frank  E.  Woods,  Captain  Leslie  T.  Peacocke,  Anita  Loos  
and  others  were  making  a  name  for  themselves  and  the  fledgling  Hollywood  

Stewart  McKie   -­‐  18  -­‐   As  of  15/08/2014  

movie  studios  were  already  receiving  some  50,000  submissions  a  year  
according  to  Edwards  and  Skerbelis  in  I  Liked  It,  Didn’t  Love  It  (2009:  12-­‐15).      

Today,  a  feature  film  screenplay  is  a  structured  document  representing  

dramatic  content  intended  for  realization  as  an  audio-­‐visual,  cinematic  
experience,  typically  of  some  90-­‐120  minutes  in  duration.  Feature  film  
screenplays  for  the  cinema  are  different  in  both  structure  and  format  from  
screenplays  intended  for  one-­‐off  TV  dramas  or  a  series  of  TV  episodes,  not  
just  because  they  are  formatted  differently  and  may  contain  some  different  
structural  elements,  but  also  because  they  are  not  subject  to  certain  TV  
screenwriting  conventions  such  as  the  need  to  write  to  ongoing  series  
themes  (defined  by  a  ‘bible’  and  overseen  by  the  ‘showrunner’  role)  or  to  
structure  the  screenplay  in  a  way  that  recognizes  the  key  role  of  
advertisement  breaks  (in  commercial/cable  television)  as  an  influence  on  
story  pacing  and  the  placement  of  plot  points  (e.g.  cliff-­‐hangers  inserted  to  
encourage  viewers  to  return  after  a  bathroom  break  or  to  tune  in  to  the  next  
episode).    Movies  destined  for  TV  are  also  structured  differently  -­‐  into  7  or  8  
acts  according  to  Edwards  and  Skerbelis  (2009:144)  -­‐  as  opposed  to  the  
conventional  3-­‐5  acts  for  a  feature  film  screenplay.    

A  screenplay  is  termed  ‘structured’  because  as  MacDonald  says  in  

Disentangling  the  Screen  Idea:  

The  screenplay  [...]  is  the  record  of  an  idea  for  a  screenwork,  written  in  
a  highly  stylized  form.  It  is  constrained  by  the  rules  of  its  form  on  the  
page,  and  is  the  subject  of  industrial  norms  and  conventions.  
(2004b:  89).  

Argentini  (1998)  and  others  (e.g.  Cole  &  Haag  (1989)  and  Riley  (2005)),  
have  produced  style  manuals  and  guides  for  defining  the  various  screenplay  
elements  and  for  formatting  the  script  for  presentation  to  a  reader.  These  
conventions  have  evolved  over  time  and  at  different  periods  have  reflected  
different  styles  of  screenwriting.  For  example  during  the  1980’s  it  was  
considered  acceptable  for  screenwriters  to  include  directions  in  the  script  
(e.g.  references  to  shots  and  camera  angles  or  how  the  actor  should  deliver  

Stewart  McKie   -­‐  19  -­‐   As  of  15/08/2014  

their  lines).  Today  this  is  considered  bad  form  according  to  commentators  
such  as  Flinn  (1999:109)  and  Epstein  (2002:109-­‐110).  Screenplays  not  
produced  in  accordance  with  the  prevailing  formatting  conventions  are  
likely  to  be  viewed  unfavorably  by  those  involved  in  their  appraisal,  
purchase  or  production  since  they  are  said  to  involve  more  effort  to  read  or  
evaluate  when  they  deviate  from  convention.    

Screenplay  content  is  “audio-­‐visual”  because  it  contains  references  to  sound  
(e.g.  spoken  dialog  and  sound  effects)  and  movement  (e.g.  human  and  non-­‐
human  action)  that  are  expressly  intended  for  realization  as  an  audio-­‐visual  
experience  for  an  audience.  Novels  contain  references  to  sound  and  
movement  but  the  intention  of  the  writer  is  that  these  references  help  create  
a  picture  in  the  mind  of  readers  and  stimulate  their  imagination.  In  a  
screenplay  sound  and  movement  must  be  translated  directly  into  an  entirely  
separate,  tangible  end  product  i.e.  a  sequence  of  individual  film  frames  with  
an  accompanying  soundtrack.  This  change  of  form  is  a  key  difference  
between  the  end  product  of  novel  writing  and  the  end  product  of  
screenwriting.  The  end  product  of  a  novel  exists  in  the  reader’s  imagination;  
the  end  product  of  a  screenplay  is  a  fixed  interpretation  of  the  text  
(interpreted  by  a  ‘director’)  –  fixed  on  physical  film  or  as  a  digital  video  file  
format  -­‐  that  delivers  a  specific  experience  intended  to  audio-­‐visually  
stimulate  the  audience’s  imagination.  

A  screenplay  is  a  relatively  substantial  document  in  terms  of  content.  Most  
feature  length  movies  are  90-­‐120  minutes  in  length  -­‐  seldom  shorter  and  
sometimes  longer.  An  industry  rule-­‐of-­‐thumb  is  that  a  page  of  script  (in  the  
traditional  typewriting  fixed-­‐spacing  font  of  Courier  12)  roughly  equates  to  
1  minute  of  produced  screen  time.  Therefore  most  feature  film  screenplays  
are  90-­‐120  A4  or  US-­‐letter  size  pages  in  length.    In  the  past,  feature  film  
scripts,  like  novel  manuscripts,  were  handwritten  or  typewritten.  Today’s  
preferred  fixed  font  for  screenplays,  courier 12,  reflects  this  typewriter  
and  early-­‐computing  heritage  (see  Millard,  2010:15-­‐16)  and  the  fact  that  a  
fixed  font,  rather  than  a  variable-­‐width  font,  is  used  ensures  some  degree  of  

Stewart  McKie   -­‐  20  -­‐   As  of  15/08/2014  

standardization  of  the  content  so  that  the  potential  movie  running-­‐time  can  
indicated  from  the  number  of  screenplay  pages.    

Today,  many  screenwriters  produce  their  work  using  a  screenwriting  

software  application  on  their  personal  computer.  Dozens  of  screenwriting  
software  applications  are  available  for  all  the  popular  operating  systems  
platforms  including  Microsoft  Windows,  Apple  Mac  OSX  and  open-­‐source  
Linux.  Consequently  most  screenplays  are  now  stored  and  published  as  an  
electronic  file  in  any  of  a  variety  of  popular  formats,  for  example:  text  (.txt),  
html  (.htm),  Adobe  PDF  (.pdf),  Microsoft  Word  (.doc),  XML  (.xml)  or  in  the  
proprietary  format  of  a  particular  screenwriting  application  such  as  Final  
Draft’s  original  file  format  .fdr  (switched  to  the  xml-­‐based  .fdx  in  version  8).  
Unfortunately  there  is  no  one  standard  screenplay  file  format  used  by  all  
screenwriting  applications.  Although  Final  Draft’s  .fdx  format  may  be  
considered  the  de-­‐facto  ‘industry  standard’  as  most  other  screenwriting  
applications  can  either  import  and/or  export  to  this  file  format,  due  to  Final  
Draft’s  perceived  market  penetration  and  industry  practitioner  popularity.    

Datafication  

Quantifying  screenplay  content  in  the  form  of  data  is  one  implementation  of  
what  Mayer-­‐Schönberger  and  Cukier  refer  to  in  Big  Data  as  ‘datafication’  
(2013:  73-­‐97).  They  make  the  point  that,  ‘The  act  of  digitization  -­‐  turning  
analog  information  into  computer-­‐readable  format  –  by  itself  does  not  
datafy.’  So  the  datafication  of  a  screenplay  does  not  only  mean  the  
presentation  of  the  script  in  a  digital  format,  such  as  in  the  form  of  an  Adobe  
PDF  file.  Datafication  means  the  decomposition  of  the  textual  screenplay  
content  into  a  series  of  data  points  designed  and  stored  to  make  the  content  
machine-­‐readable  and  to  facilitate  machine  processing.    

In  an  example  relevant  to  screenwriting  (2013:  84),  the  authors  explain  the  
utility  of  making  words  into  data  by  referencing  Google’s  effort  to  make  as  
many  of  the  world’s  books  searchable,  by  ‘datafying’  the  individual  words  in  
the  texts,  which  enables  their  Ngram  viewer  

Stewart  McKie   -­‐  21  -­‐   As  of  15/08/2014  

(http://books.google.com/ngrams/,  accessed  26  April,  2013)  to  generate  
interesting  analysis  of  the  use  of  words  or  phrases  over  time.  

Figure  1.1  -­‐  Google  Ngram  Viewer  plots  specific  name  mentions  over  time  

As  more  and  more  physical  world  entities  and  events  become  ‘datafied’  –  
locations  (e.g.  as  latitude  and  longitude  co-­‐ordinates),  relationships  (e.g.  as  
social  network  interactions),  the  current  state  of  an  Internet-­‐of-­‐Things  
object  etc.  –  the  datafication  of  screenplay  content  is  an  inevitable  additional  
‘brick  in  the  wall’  and  something  with  the  potential  to  significantly  change  
the  practice  of  screenwriting  and  the  notion  of  the  screenplay  as  a  digital  
artefact.  

A  key  outcome  of  making  screenplay  content  machine-­‐readable  is  that  the  
content  is  easier  to  analyze  and  visualize  (terms  collectively  referred  to  as  
‘analytics’  below)  programmatically.    By  analyze  I  mean  applying  an  analysis  
technique  to  the  data  (e.g.  clustering)  and  by  visualize  I  mean  presenting  the  
result(s)  of  that  analysis  in  a  way  that  facilitates  better  understanding  of  the  
meaning  of  the  data  (e.g.  as  a  chart  or  a  timeline).  Screenwriting  2.0  assumes  
that  every  screenwriting  application  will  include  ‘embedded’  analytics  –  
unlike  today.  

In  fact,  in  the  time  it  would  take  a  typical  human  reader  to  read  a  single  
script  and  deliver  some  qualitative  analysis  based  on  experience  or  ‘gut-­‐feel’,  
a  computer  could  read  an  entire  corpus  of  datafied  scripts  to  process  and  

Stewart  McKie   -­‐  22  -­‐   As  of  15/08/2014  

correlate  what  could  be  a  significant  volume  and  variety  of  quantifiable  data  
to  deliver  insights  based  on  the  evidence  of  the  data  rather  than  the  intuition  
of  a  human  reader.  Quantitative  analysis  becomes  more  important  once  
more  screenplays  are  datafied.  Franzosi  (2010:4)  answers  his  own  question  
‘Words  are  beautiful:  Why  do  you  want  to  turn  them  into  numbers?’  with  
this  answer  (2010:5),  ‘I  quantify  simply  because  I  have  far  too  much  
information  to  deal  with  qualitatively’.  

This  quantitative  ‘data-­‐based’  approach  does  not  replace  the  qualitative  

approach;  it  merely  supplements  and  complements  it,  by  adding  an  
evidential  machine-­‐analyzed  perspective  to  an  intuitive  human-­‐analyzed  
perspective.  This  may  result  in  a  more-­‐rounded  analysis  of  the  screenplay  
than  either  approach  on  its  own.    The  potential  of  evidential  analytics  of  
screenplay  content,  enabled  by  screenplay  as  data,  is  a  key  outcome  of  
datafication  that  is  explored  in  this  thesis.  

Research  Question  

Here,  the  screenplay  is  primarily  considered  as  a  digital  data  artefact  rather  
than  as  a  digital  document  artefact.  And  also  as  a  dynamic  artefact  where  the  
data  is  subject  to  ongoing  change  during  a  lifecyle  that  will  involve  a  
changing  ‘community  of  interest’  comprising  various  stakeholders  with  
different  content  management  roles.  

Datafication  of  the  screenplay  has  the  potential  to  positively  impact  the  
nature  of  the  screenplay  as  a  digital  artefact,  and  the  way  screenplay  content  
can  be  analyzed  and  visualized  and  screenwriting  as  a  collaborative,  social  
process.  Therefore  the  research  question  that  informs  this  thesis  is:  

What  are  the  possibilities  of  screenplay  ‘datafication’?  

How  the  screenplay  as  data  can  impact  creating  and  managing,  presenting  
and  sharing,  analyzing  and  visualizing  screenplay  content.  

Creating  and  Managing  

Stewart  McKie   -­‐  23  -­‐   As  of  15/08/2014  

Creating  and  managing  screenplays  refers  to  how  the  textual  content  is  
stored  and  organized  -­‐  for  example,  in  a  database  or  a  file,  in  a  script-­‐centric  
paradigm  or  a  scene-­‐centric  paradigm  –  and  the  importance  of  structural  
groupings  and  layers,  such  as  sequences  and  storylines,  for  organizing  and  
navigating  the  screenplay  content.      

Presenting  and  Sharing  

The  presentation  of  a  screenplay  should  not  be  entirely  focused  on  the  script  
as  a  document  with  an  emphasis  on  viewing  the  script  simply  as  a  digital  
version  of  the  traditional  physical  printed  artefact.  The  screenplay  as  online,  
digital  artefact  demands  more  attention  is  paid  to  the  visual  look  and  feel  of  
the  content  presentation  and  to  the  variety  of  ways  the  content  can  be  
organized,  navigated  and  searched.  The  social  aspect  of  screenwriting  as  a  
collaborative,  co-­‐creation  process  demands  that  screenplays  are  easy  to  
share  and  interact  with  to  support  the  kinds  of  activities  associated  with  
content  sharing  by  today’s  social-­‐networking  generation  including  
commenting,  rating/ranking  and  content  linking.    

Analyzing  and  Visualizing  

The  screenplay  content  must  be  stored  and  organized  in  a  way  so  as  to  
facilitate  content  analysis  and  content  visualization  in  order  to  support  
deriving  ‘insight’  from  any  quantitative  analysis  of  the  screenplay  data.  The  
availability  of  content  analysis  and  visualization  should  be  a  key  tool  in  
helping  the  screenwriter  to  ‘ask  more  questions’  of  their  script  in  order  to  
improve  its  quality  and  /or  commercial  value  through  analysis  of  content  
from  different  perspectives.    

To  investigate  this  question,  my  research  has  focused  on  three  areas  each  
representing  a  chapter  in  this  thesis:    

1.  Approaching  the  screenplay  as  a  data,  not  document  artefact  or  

‘screenplay  as  data’.  

Stewart  McKie   -­‐  24  -­‐   As  of  15/08/2014  

2.  Recognizing  that  screenwriting  is  a  social  and  business  process  rather  
than  an  individual  and  solely  creative  process  or  ‘screenwriting  as  social  
network’.  

3.  Leveraging  the  availability  of  the  screenplay  as  data  for  analytical  and  
visualization  purposes  or  to  deliver  ‘screenplay  analytics’.  

My  final  chapter,  before  those  discussing  the  practical  deliverables  from  my  
research,  is  ‘screenplay  as  design’  where  I  discuss  how  a  datafied  screenplay  
provides  the  platform  that  facilitates  possible  ways  to  design  and  
automatically  generate  the  ‘framework’  for  a  screenplay.      

Screenwriting  2.0  

When  I  use  the  term  ‘Screenwriting  2.0’  I  am  proposing  a  term  that  refers  to  
an  evolutionary  development  in  screenwriting  technology  and  practice  
analogous  to  that  characterized  by  the  term  ‘Web  2.0’.    The  Web  2.0  term  has  
of  course  been  hijacked  by  the  technology  press  as  a  marketing  term  after  it  
was  coined  by  Darcy  DiNucci  in  Fragmented  Future    (see  
http://www.darcyd.com/fragmented_future.pdf,  accessed  17  Dec.  2013)  
and  popularized  by  Tim  O’Reilly’s  Web  2.0  conferences  from  2004  onwards  
(see  What  Is  Web  2.0?  http://oreilly.com/web2/archive/what-­‐is-­‐web-­‐
20.html,  accessed  17  Dec.  2013).  Web  2.0  has  come  to  mean  the  convergence  
of  a  number  of  technologies  –  including,  among  others:  the  cloud,  mobile  
devices,  global  positioning  satellites  (GPS),  pervasive  social  networks,  
improved  data  visualization,  a  focus  on  website  design  and  usability  -­‐  and  
ways  of  using  the  web  that  denote  a  meaningful  shift  away  from  the  earlier  
versions  of  the  Web.  

Back  in  1999,  DiNucci  suggested  that  ‘Today’s  Web  is  essentially  a  
prototype,  a  proof  of  concept.’  (1999:  32).  I  suggest  that  the  same  could  have  
been  said  about  screenwriting  applications  in  2006  -­‐  applications  that  could  
generally  be  characterized  as  allowing  a  single  user  to  access  the  application  
to  write  a  screenplay  document,  formatted  to  industry  conventions,  and  to  

Stewart  McKie   -­‐  25  -­‐   As  of  15/08/2014  

print  it  out  on  paper  or  share  electronically  in  the  form  of  a  PDF  file.  As  
Andrew  Kenneth  Gay  says,  ‘Given  access  to  a  time  machine,  a  Hollywood  
screenwriter  working  half  a  century  ago  would  encounter  little  difficulty  
adjusting  to  the  conventions  of  professional  screenwriting  in  2012’  (see  
Screenwriting  2.0  in  the  Classroom?  Teaching  the  Digital  Screenplay,  
http://framescinemajournal.com/article/screenwriting-­‐2-­‐0-­‐in-­‐the-­‐
classroom/,  accessed  17  Dec.  2013)  to  which  we  might  add  adjusting  to  the  
technology  of  screenwriting.  

When  I  began  my  research  in  2006,  few  screenwriting  applications  stored  
their  screenplay  content  as  data  (in  a  database  as  opposed  to  in  a  file),  only  
one  (to  my  knowledge)  offered  any  kind  of  screenplay  analytics,  and  none  
were  designed  to  operate  as  a  multi-­‐user  social  network.  To  me,  these  are  at  
least  three  hallmarks  of  the  evolution  towards  Screenwriting  2.0  that  
differentiate  this  iteration  from  the  first  generation  of  screenwriting  
applications  or  Screenwriting  1.0  if  you  will.  The  fundamental  purpose  of  
this  thesis  is  to  discuss  these  hallmark  characteristics  and  apply  them  to  the  
creation  of  a  Screenwriting  2.0  prototype  application.    

Sources  

I  have  selected  four  information  sources  for  my  literature  review,  which  of  
necessity  encompasses  non-­‐literary  artefacts  such  as  software  tools  and  
applications:  

1. Academic  research  focused  on,  or  applicable  to,  screenplay  content  

management  and  the  technology  of  screenwriting.  

2. Screenwriting,  novel-­‐  and  play-­‐writing  “how-­‐to”  manuals  and  other  

works  relevant  to  screenwriting,  which  are  not  considered  ‘academic’  
but  are  primarily  intended  to  help  writers  produce  successful,  that  is  
commercially  valuable,  screenplays.  

3. The  current  generation  of  story-­‐planning  and  scriptwriting  software  

applications  and  tools  available  in  the  marketplace.  

Stewart  McKie   -­‐  26  -­‐   As  of  15/08/2014  

 4. A  range  of  generic  textual  analysis  and  data  visualization  tools  that  
are  not  specifically  designed  for  analyzing  or  visualizing  screenplays  
but  have  the  potential  to  be  applied  to  screenplays.    

Most  of  the  non-­‐academic  literature  relating  to  screenwriting  is  primarily  
concerned  with  how  to  write  a  "better"  screenplay  and  generally  “better"  is  
taken  mean  a  screenplay  that  has  an  improved  chance  of  succeeding  
commercially.  This  is  understandable,  because  unlike  a  novel,  screenplays  
are  written  to  make  into  movies  rather  than  read  for  pleasure.  So  unless  a  
screenplay  is  bought  it  is  unlikely  to  be  produced,  which  usually  defeats  the  
purpose  of  writing  it.    However,  this  commercial  bent  does  not  mean  that  
how-­‐to  manuals  have  no  merit.  

But  what  should  be  noted  about  many  screenwriting-­‐related  how-­‐to  

manuals  is  that  any  screenplay  analysis  they  contain  is  often  not  of  a  script  
text  per  se  –  although  this  is  not  the  case  for  manuals  focused  on  specific  
script-­‐improvement  tasks,  such  as  writing  better  dialog  for  example  -­‐  but  of  
the  movie  that  was  made  from  a  script:  Movie  content  analysis  rather  than  
screenplay  content  analysis.  This  is  an  important  distinction  because  the  
focus  of  this  analysis  is  based  on  the  visual  end  product  rather  than  the  
textual  screenplay  (i.e.  the  output  rather  than  the  input).  It  could  also  be  
argued  that  analyzing  a  visual  movie  is  easier  than  analyzing  the  script  itself  
because  whereas  almost  anyone  who  watches  a  movie  can  describe  the  
plot(s)  and  characters  of  a  movie  just  from  seeing  it,  it  is  harder  to  do  this  by  
only  reading  the  script  without  having  seen  a  movie  produced  from  it.    

The  realized  visualization  of  the  movie  itself  assists  the  viewer  to  raise  the  
level  and  quality  of  their  analysis  of  its  content.  Just  as  handling  or  using  a  
finished  product  aids  the  quality  of  analysis  as  opposed  to  simply  reading  
the  product’s  technical  specification  or  viewing  a  mockup  sketch  or  
photograph  of  the  product.  To  analyze  screenplay  textual  content  is  simply  
not  the  same  thing  as  analyzing  audio-­‐visual  movie  content  derived  from  the  
screenplay.    They  are  two  different  analytical  subjects.  

Stewart  McKie   -­‐  27  -­‐   As  of  15/08/2014  

Methodology  

This  is  a  PhD  by  practice  that  is  framed  as  a  software  development  project.  
The  deliverable  of  the  ‘practice’  element  is  a  series  of  prototype  applications  
that  relate  to  the  research  question.  These  prototypes  are  informed  by  the  
research  into  screenwriting  theory  and  practice  as  represented  by  the  
‘written’  part  of  the  thesis.  The  prototypes  are  working  applications  that  
were/are  delivered  online  and  publicly  accessible  to  anyone  who  registers  
to  use  them.  

Start  Point  

Inevitably  my  thesis  is  informed  by  my  own  ‘start  point’,  which  when  I  
started  this  PhD  in  2006  included  a  MA  in  Screenwriting,  a  MSc  in  
Organization  Consulting  and  my  ‘day  job’  as  an  enterprise  application  
business  analyst.  

My  MA  course  provided  me  with  a  basic  understanding  of  screenwriting  

practice.  But  what  I  noticed  was  that  the  content  of  the  course  lacked  any  
focus  on  the  technology  of  screenwriting.    And  apart  from  an  assignment  to  
write  a  ‘script  report’,  assuming  the  role  of  a  script  reader,  there  was  no  
other  discussion  of  screenplay  analysis  let  alone  the  use  of  technology  to  
deliver  it.  This  all  struck  me  as  odd,  given  that  we  live  in  a  digital  age,  many  
people  on  the  course  wrote  their  screenplays  using  screenwriting  
applications  and  the  analysis  of  many  other  kinds  of  business  data  was,  and  
continues  to  be,  a  focus  in  the  continuous  improvement  of  business  software  
applications  generally.  

My  MSc  was  focused  on  the  relational  and  reflective  aspects  of  consulting.  
We  were  encouraged  to  view  consulting  as  collaboration  –  between  client  
and  consultant  –  in  which  any  outcomes,  whether  positive  or  negative,  are  
co-­‐created.  There  was  an  emphasis  on  the  dynamics  of  consulting  and  
paying  close  attention,  both  ‘in  the  moment’  and  ‘on  reflection’  to  what  you  
and  the  client  were  actually  saying  and  doing.    One  of  the  methodologies  we  

Stewart  McKie   -­‐  28  -­‐   As  of  15/08/2014  

experimented  with  was  Action  Research,  which  I  will  return  to  later.  

My  daily  work  as  a  business  analysis  often  involves  me  in  projects  to  select  
and/or  implement  relatively  complex  enterprise  resource  planning  (ERP)  
software.    A  core  focus  of  these  kinds  of  projects  is  understanding  the  scope  
of  operational  domains  in  a  business,  the  data  that  is  managed  and  the  
processes  used  to  create  and  manage  that  data,  including  user  roles  and  
activities.    This  perspective  also  informs  the  technology  deliverables  from  
my  thesis.    

The  lack  of  technology  emphasis  on  my  MA  course  led  me  to  investigate  
what  technology  was  available  for  screenwriters.  Some  formal  output  from  
this  investigation  was  a  series  of  articles  I  wrote  for  Scriptwriter  magazine  in  
2006  that  reviewed  specific  screenwriting  applications  and  the  
screenwriting  technology  market  generally  (see  McKie  2006a  and  2006b  for  
two  examples).      

From  my  investigation  and  trial  use  of  a  number  of  commercial  
screenwriting  applications  it  became  clear  that  the  technology  was  limited  
in  scope  and  that  among  the  functional  deficiencies  were  the  lack  of  content  
analytics  and  the  management  of  screenplays  as  a  single  user  ‘closed  
document’  artefact  rather  than  a  social  ‘open  data’  artefact.  I  was  
particularly  interested  in  the  possibilities  for  screenwriters  to  gain  more  
insight  about  their  scripts  through  the  use  of  content  analytics,  something  
that  only  one  screenwriting  application  –  Sophocles  -­‐  really  addressed  in  any  
significant  way  at  the  time.  

Analyzing  the  content  of  screenplays,  by  leveraging  the  wide  range  analytic  
and  visualization  technology  available  for  data  analysis,  was  not  easy  simply  
because  although  many  screenplays  were  available  in  a  ‘digitized’  form  (i.e.  
available  in  an  electronic  file  format  such  as  Adobe  PDF)  they  were  not  
‘datafied’  (i.e.  available  as  data  in  a  database).    This  meant  that  in  order  to  
convert  a  digitized  script  into  a  datafied  script  generally  required  some  
manual  ‘wrangling’  or  programmatic  intervention  to  get  the  data  into  the  

Stewart  McKie   -­‐  29  -­‐   As  of  15/08/2014  

format  needed  so  that  analytic/visualization  tools  could  be  used  to  analyze  
the  data  content  of  the  scripts.  Clearly  a  significant  issue  was  that  the  
prevailing  focus  in  screenwriting  applications  on  ‘screenplay  as  document’,  
rather  than  ‘screenplay  as  data’,  was  holding  back  the  development  of  
screenplay  analytics  (the  analysis  and  visualization  of  screenplay  content).  

It  was  clear  from  my  initial  research  into  the  ‘manual’  (rather  than  
automatically  generated  by  software)  visualization  of  screenplay  content  in  
various  academic  and  how-­‐to  manuals  that  there  were  examples  of  
visualizations,  particularly  of  the  analysis  of  screenplay  structures,  that  
could  be  possible  outputs  generated  from  screenplay  analytic  functionality  
in  screenwriting  applications.  Samples  of  these  visualizations  are  discussed  
in  chapter  2.  

From  discussions  in  2006-­‐7  with  my  then  supervisors  Prof.  Fionn  Murtagh  
and  Adam  Ganz,  screenplay  analytics  seemed  like  a  useful  line  for  further  
research,  and  in  fact  this  early  focus  did  lead  to  a  number  of  tangible  outputs  
from  both  my  supervisors  and  myself.    

Murtagh  went  on  to  produce  a  number  of  papers  that  focused  on  the  content  
analysis  of  both  movie  and  TV  series  screenplays  (most  of  these  papers  are  
listed  here  -­‐  http://www.narrativization.com  -­‐  accessed  June  17,  2014).  
Ganz  organized  Film,  Visualization,  Narrative,  a  workshop  run  at  Royal  
Holloway,  University  of  London  on  17  Nov.  2006,  supported  by  the  AHRC  
ICT  Methods  Network  and  LCACE  (London  Centre  for  Arts  and  Cultural  
Enterprise)  and  then  edited  the  Reconstruction  Vol.  8,  No.  3,  2008  issue  
focused  on  Visualization  and  Narrative,  which  included  my  own  contribution  
Screenplay  Visualization:  Concepts  and  Practice  (see  
http://reconstruction.eserver.org/Issues/083/mckie.shtml  -­‐  accessed  
June17,  2014).      

I  presented  Scriptcloud  -­‐  my  first  screenplay  analytics  prototype  application  

-­‐  at  the  Second  International  Workshop  on  Semantic  Media  Adaptation  and  
Personalization  (SMAP)  held  at  Brunel  University  in  2007.  My  paper,  

Stewart  McKie   -­‐  30  -­‐   As  of  15/08/2014  

Scriptclud  (sic).com:  Content  Clouds  for  Screenplays  was  also  included  in  the  
conference  proceedings  (see  http://dl.acm.org/citation.cfm?id=1339060  -­‐  
accessed  June  17,  2014).    I  also  secured  LCACE  funding  for  Scriptgeist,  my  
next  screenplay  analytics  prototype  application  released  in  2008,  with  the  
help  of  Lydia  Daniels  of  the  Royal  Holloway  Enterprise  center  and  supported  
by  Adam  Ganz.  

This  early  work  on  screenplay  analytics  was  also  referenced  in  a  Nature  
online  news  item  Here's  looking  at  you,  kid.  Software  promises  to  identify  
blockbuster  scripts,  (Nature,  vol.  453,  p.  708,  4  June  2008:  see  
http://www.nature.com/news/2008/080604/full/453708a.html  -­‐  accessed  
June  17,  2014).  

Next  Steps  

At  this  point,  my  research  had  what  appeared  to  be  a  useful  focus  –  
screenplay  analytics  –  so  there  were  a  number  of  ways  I  could  have  
proceeded  in  terms  of  my  research  methodology  in  the  context  of  my  
framing  of  my  PhD  by  practice  as  a  software  development  project.  

One  might  have  been  a  case  study  approach,  whereby  I  studied  a  specific  
screenwriters  practice  or  a  small  group  of  screenwriters  practice  in  order  to  
extrapolate  ‘learnings’  from  these  case  studies  to  inform  my  software  
development.  Although  I  solicited  potential  participants  to  work  this  way  via  
the  Internet  no-­‐one  was  really  interested  enough  in  screenwriting  
technology  or  screenplay  analytics  to  devote  the  time  and  energy  needed  to  
participate  in  this  way  and  as  I  did  not  know  any  suitable  participants  
personally,  I  had  to  reject  this  as  a  way  forward.  

I  could  have  informed  my  development  by  use  of  one  or  more  surveys  about  
peoples’  views  on  the  technical  development  of  screenwriting  technology.  In  
fact  I  did  a  basic  online  survey  on  this  topic  just  before  the  start  of  my  PhD,  
the  results  of  which  were  summarized  by  screenwriting  blogger  Alex  
Epstein  here:  

Stewart  McKie   -­‐  31  -­‐   As  of  15/08/2014  

http://complicationsensue.blogspot.co.uk/2006/11/polling-­‐numbers.html,  
accessed  14  March  2014.  

Unfortunately,  in  general  in  the  Internet  age,  many  people  are  ‘surveyed  out’  
due  to  the  many  survey  solicitations  they  now  receive  by  email  (I  myself  get  
at  least  one  a  week),  and  screenwriters  from  active  online  communities  like  
Shootingpeople.org  did  not  seem  that  interested  in  responding  to  surveys  
and  particularly  not  when  the  subject-­‐matter  was  technology-­‐centric.  The  
survey  referenced  above  attracted  less  than  50  respondents.  Surveys  are  
also  difficult  to  design  effectively  and  interpret  usefully  if  you  are  not  a  
professional  statistician.    So  again  I  did  not  feel  this  was  a  direction  worth  
pursuing.  

I  could  also  have  recruited  some  kind  of  ‘beta  test’  group  from  the  start  and  
tried  to  involve  them  throughout  my  research.  But  this  was  sure  to  be  a  big  
ask  given  that  my  part-­‐time  studies  could  take  6-­‐7  years  and  few  people  will  
remain  engaged  with  someone  else’s  project  for  that  period  of  time  unless  it  
has  a  significant  value  proposition  for  them.  In  fact  towards  the  end  of  my  
studies  I  managed  to  recruit  a  small  beta  test  group  of  interested  
academics/students  to  test  a  Scenepad  prototype  deliverable,  but  only  one  
of  them  truly  engaged  with  the  project  over  the  beta  period,  which  in  this  
case  was  only  3  months  and  participants  were  hand-­‐held  along  the  way.    

So  I  decided  to  start  like  so  many  software  developers  of  online  applications  
start  –  by  building  a  prototype,  making  it  available  online  and  seeing  if  
anyone  would  use  it.  The  hope  was  that  if  I  built  it  ‘they  would  come’.  If  I  was  
lucky,  maybe  it  would  ‘go  viral’  in  some  way,  as  certain  applications  that  
somehow  appeal  to  the  zeitgeist  manage  to  do  every  now  and  then.    This  is  a  
risky  approach  as,  without  any  marketing  funding  to  promote  the  
applications,  usage  depends  largely  on  positive  ‘word  of  mouth’  marketing  
from  users  themselves.  In  essence  the  idea  was  to  let  the  application  
prototypes  ‘speak  for  themselves’  and  see  what  emerged  as  a  result.    So  this  
became  my  start  point.  

Stewart  McKie   -­‐  32  -­‐   As  of  15/08/2014  

What  must  be  made  clear  here  is  that  I  did  not  expect  anyone  to  engage  with  
my  prototypes.  In  a  sense  the  development  and  delivery  of  the  online  
prototypes  is  best  characterized  as  a  kind  of    ‘thought  experiment’.  If  I  was  
lucky,  some  people  may  engage  with  the  prototypes  and  provide  useful  
feedback  for  further  development  but  if  not  this  would  not  invalidate  the  
purpose  of  the  exercise.  Without  anyone  using  the  prototypes  I  may  not  be  
able  to  observe  actual  use  of  the  software  or  conduct  genuine  experiments  
with  the  functionality  delivered  (e.g.  A/B  testing  commonly  used  for  
validating  website  page  design  –  see  for  example  
https://www.optimizely.com,  accessed  July  09,  2014)  but  the  prototypes  do  
function  as  a  series  of    ‘imaginary’  scenarios  in  order  to  imagine  what  a  
datafied  screenplay  could  deliver  in  terms  of  analytic  and  social  potential.        

Qualitative  Research  Focus  

The  focus  of  my  research  is  more  qualitative  than  quantitative.  For  example,  
it  was  not  about  gaining  as  many  users  as  possible  for  my  prototypes,  or  
positing  user  surveys  to  garner  ‘votes’  for  features  and  functions,  or  
gathering  statistics  about  the  usage  of  specific  features  and  functions  in  my  
prototypes,  simply  because  without  statistically  significant  user  engagement  
with  my  prototypes  this  kind  of  quantitative  analysis  is  impossible.  

It  was  however,  about  the  kind  of  qualitative  research  as  defined  by  Savin-­‐
Baden  and  Howell  Major  (2013:11):  ‘We  define  qualitative  research  simply  
as  social  research  that  is  aimed  at  investigating  the  way  in  which  people  
make  sense  of  their  ideas  and  experiences.’    

In  my  research  context  I  correlate  ideas  with  ‘screenplays’  and  experiences  as  
the  experience  of  ‘developing  and  writing  a  screenplay’;  the  tools  to  help  
people  make  sense  comprise  the  software  prototype  deliverables  themselves  
and  the  users’  use  of  them.  

My  research  approach  was  informed  by  the  principles  of  ‘action  research’,  
originally  attributed  to  Lewin  (1946:35),  which  I  had  been  introduced  to  and  

Stewart  McKie   -­‐  33  -­‐   As  of  15/08/2014  

used  during  my  MSc  in  Organization  Consulting  at  Ashridge.  In  terms  of  a  
specific  definition  of  action  research,  I  refer  to  that  of  Kemmis  (2012:417)  as  
quoted  in  Savin-­‐Baden  and  Howell  Major  (2013:245):  ‘Action  research  
concerns  action,  and  transforming  people’s  practices  (as  well  as  their  
understandings  of  their  practices  and  the  conditions  under  which  they  
practise(sic))’.  

In  my  research  context  I  correlate  action  with  ‘delivering  software  

prototypes  for  use’  so  that  the  functionality  that  is  delivered  has  the  
potential  to  be  used  for  transforming  people’s  practices.  

Action  research  implies  the  use  of  a  spiral  cycle  of  activities  that  broadly  
speaking  comprises  planning  things  (plan),  doing  things  (do  or  act),  
observing  the  results  (observe)  and  reflecting  on  the  results  (reflect)  in  
order  to  (in  some  cases)  reiterate  the  cycle  one  or  more  times  to  achieve  a  
qualitative  improvement.    Again,  in  my  research  context,  I  am  more  aligned  
with  a  variant  of  the  approach  outlined  by  Jean  McNiff  (see  
http://www.jeanmcniff.com/ar-­‐booklet.asp,  accessed  11  March  2014):  

• identify  an  area  of  practice  to  be  investigated  i.e.  screenwriting  
• imagine  a  solution  i.e.  new  screenwriting  technology  
• implement  the  solution  i.e.  a  screenwriting  application  prototype  
• evaluate  the  solution  i.e.  in  the  light  of  user’s  feedback/usage  
• change  practice  in  light  of  the  evaluation  i.e.  refine  existing  or  deliver  a  
new  prototype  

An  issue  with  the  use  of  action  research  is  whether  the  intention  is  to  
improve  the  practice  of  the  action  researcher  or  that  of  the  group  of  
practitioners  that  the  action  researcher  involves  in  the  action  research  
project.  Since  in  this  case  it  was  in  a  sense  both,  then  the  latter  intention  may  
surface  a  problem  identified  by  Savin-­‐Baden  and  Howell  Major  (2013:254):    

In  some  forms  of  action  research,  where  the  researchers  design  the  
action  plan  for  the  practitioners  (rather  than  with  them),  there  is  a  
danger  that  there  will  be  little  ownership  by  the  practitioners  and  this  
in  turn  will  affect  the  degree  to  which  change  is  ‘owned’,  implemented  
and  really  possible.  

Stewart  McKie   -­‐  34  -­‐   As  of  15/08/2014  

This  is  a  potential  problem  in  any  ‘speculative’  software  application  
development  where  the  software  deliverable,  i.e.  the  action  plan  as  
prototype,  is  designed  by  the  developer  and  presented  ‘as-­‐is’  to  potential  
users.  It  was  a  problem  I  faced  but  unfortunately  the  general  lack  of  any  user  
feedback  meant  that  it  was  not  an  easy  problem  to  overcome  since  so  few  
users  engaged  with  the  prototypes  themselves  or  with  me  as  the  developer  -­‐  
despite  being  provided  with  online  feedback  forms  and  my  offline  contact  
numbers  and  email  address.  

Data  Collection  and  Documents  

In  my  research  context,  data  collection  was  largely  about  understanding:    

1. how  many  people  were  using  and  engaging  with  my  prototypes  
2. what  kind  of  people  were  signing  up  to  use  the  prototypes    
3. what  they  liked  and  disliked  about  the  features  and  functionality  
4. what  ideas  and  suggestions  they  had  for  changes  or  new  functionality  
5. how  many  scripts  they  uploaded  into  and/or  wrote  in  the  applications    

I  could  determine  (1)  from  how  many  users  had  signed  up  and  how  often  
they  logged  into  a  prototype  since  this  data  is  collected  by  the  system  itself.    
But  in  practice  many  registrants  did  not  provide  much  in  the  way  of  
metadata  about  themselves  in  their  user  profiles  so  (2)  was  unreliable.  And  
although  it  is  easy  to  log  how  many  page  views  a  specific  page  gets  in  an  
online  application,  this  does  not  really  help  with  understanding  (3)  in  a  
qualitative  way.    Both  (3)  and  (4)  depended  on  the  receipt  of  verbal  or  
virtual  feedback  from  users  via  contact  forms  or  feedback  forms  or  
comments  within  the  application.    

(5)  is  an  important  metric  but  one  that  essentially  turned  out  to  be  pretty  
much  a  one-­‐to-­‐one  relationship  (i.e.  1  user,  1  script).  Professional  
screenwriters  who  might  be  fortunate  enough  to  be  managing  multiple  
scripts  at  a  time  were  unlikely  to  risk  using  these  unproven  tools  and  most  
people  who  actually  tried  the  applications  probably  only  had  one  screenplay  
that  they  were  currently  focused  on  and  prepared  to  ‘dabble  with’  in  the  
applications.    So  this  data  point  also  proved  to  deliver  minimal  insight.    

Stewart  McKie   -­‐  35  -­‐   As  of  15/08/2014  

Fieldwork,  Interviews  and  Observations  

 Despite  the  deliverables  being  online  applications  used  by  remote  users  
(albeit  from  all  over  the  world)  and  therefore  functioning  as  ‘surrogate’  
fieldwork,  I  did  some  activities  that  could  be  classed  as  traditional    
‘fieldwork,  interviews  and  observation’  as  part  of  my  research.  This  
fieldwork  took  place  in  the  context  of  working  with  three  small  groups  of  
students  from  the  ‘teaching  and  learning’  community  of  interest,  as  outlined  
below.    In  all  three  cases  I  worked  face-­‐to-­‐face  with  the  students  asking  
questions,  observing  behavior  and  discussing  any  comments  posted  directly  
to  the  system  by  the  students  or  given  verbally.    

My  interviews  and  observation,  a  mixture  of  ‘focus-­‐group’  (or  constituted  

group  –  see  Bruhn  Jensen  2002:241)  and  ‘one-­‐to-­‐one’  oriented,  did  not  
follow  a  script  or  present  a  structured  checklist  of  questions  for  students  to  
answer  or  formally  record  peoples’  behavior  in  any  ethnographic  fashion  
but  were  more  informal  (although  all  three  required  significant  preparation  
work  to  ‘design’  the  formal  part  of  the  sessions):  

• At  my  University  College  Falmouth  session  (c.  2  hours),  I  presented  

some  principles  of  screenplay  analytics,  gave  a  demonstration  of  my  
Scriptgeist  prototype  and  then  discussed  the  utility  of  screenplay  
analytics  with  the  students  in  a  Q&A  session.  
 
• At  my  Bournemouth  University  session  (c.  2  hours),  I  gave  a  demo  of  
my  Scenewrite  prototype  and  then  facilitated  the  students  use  of  the  
system  in  a  ‘floorwalker’  training  role  so  they  could  engage  with  the  
system  immediately.  One  student  did  continue  to  engage  with  the  
Scenewrite  prototype  post-­‐session  but  subsequently  gave  up  and  did  
not  answer  emails.  
 
• At  my  Royal  Holloway  session  (c.  3  hours),  I  introduced  and  
demonstrated  my  Sceneclass  add-­‐in  for  Wordpress,  facilitated  the  
students’  use  of  the  system  in  a  ‘floorwalker’  training  role  and  then  
reviewed  their  posted  comments  and  experience  of  using  the  system  
in  the  closing  feedback  session.  

In  terms  of  Savin-­‐Baden  and  Howell  Major’s  continuum  of  observation  roles  
(2013:394  figure  25.1)  I  would  say  my  personal  role  in  these  sessions  was  
located  between  ‘balanced  participation’  and  ‘active  participation’.  It  could  

Stewart  McKie   -­‐  36  -­‐   As  of  15/08/2014  

not  involve  ‘complete  participation’  since  the  point  was  for  the  students  to  
respond  to/use  the  system  prototypes,  and  not  me.  However,  I  think  it  is  fair  
to  say  that  fieldwork,  interviews  and  observation  did  comprise  a  minor  but  
demonstrable  activity  within  my  research  methodology.  

Data  Analysis  and  Interpretation  

The  data  collected  from  my  prototypes  during  my  research  was  hardly  
significant  enough  to  draw  any  meaningful  conclusions  from  other  than  
‘anecdotal’  observations  such  as:  

• User  engagement  went  down  as  functionality  went  up  

 
• Users  showed  more  engagement  with  a  screenplay  oriented  online  
game  (Geistmeister  in  ScriptGeist)  than  with  actually  using  prototypes  
to  write  and  analyze  screenplays  
 
• Screenwriters  do  not  seem  to  value  the  utility  of  analytics  to  help  
them  ‘ask  more  questions  of  their  scripts’.  

Whereas  my  first  prototype,  the  word  frequency  cloud  generator  

Scriptcloud,  attracted  a  few  hundred  users,  my  final  prototype  ‘submission’  
deliverable  Scenepad  –  containing  significantly  more  functionality  including  
screenplay  writing  capability  -­‐  attracted  only  a  handful  and  an  earlier  
iteration  of  Scenepad  did  not  manage  to  sustain  the  engagement  of  a  small  
group  of  motivated  beta-­‐test  users  who  had  signed  up  to  participate  in  a  
highly  structured  beta  test  program.  

My  Geistmeister  screenplay  guessing  game  that  was  part  of  the  Scriptgeist  
prototype  generated  much  more  engagement  that  the  rest  of  the  application  
that  actually  delivered  significant  (in  terms  of  comparable  commercial  
applications)  screenplay  analytics  functionality.  

Anecdotal  responses  to  my  positing  of  the  potential  usefulness  of  screenplay  
analytics  on  blogs  and  forums  were  not  positive.  Of  the  very  few  blogger  
comments  I  received,  at  least  two  specifically  questioned  the  need  for  
‘innovative’  screenwriting  technology  or  the  utility  of  screenplay  analytics  as  

Stewart  McKie   -­‐  37  -­‐   As  of  15/08/2014  

shown  below  in  the  screenshots  below:  

Figure  1.2  –  Comment  posted  by  Alex  Epstein  on  his  Complications  Ensue  blog  

 (see  http://complicationsensue.blogspot.co.uk/2006/11/polling-­‐
numbers.html,  accessed  14  March  2014.)  

Stewart  McKie   -­‐  38  -­‐   As  of  15/08/2014  

Figure  1.3  –  Comment  posted  by  a  Stage32.com  user  on  their  blog  and  sent  to  me  by  email  on  
31  December  2013  

Since  there  was  so  little  data  collected,  either  from  ‘external’  blog  comments  
or  as  ‘internal’  feedback  on  the  applications  themselves,  there  was  no  point  
in  to  trying  to  draw  any  conclusions  from  it,  so  data  analysis  and  
interpretation  turned  out  to  be  minimal.  

Prototype  Development  Methodology  

The  development  of  the  software  prototypes  in  response  to  my  research  
question  was  a  key  deliverable  of  the  practice  element  of  this  PhD.  In  most  
commercial/business  settings,  a  software  application  development  project  is  
generally  one  of  two  kinds:  

1. An  application  development  carried  out  as  a  speculative  venture  by  a  

software  developer/development  firm  where  the  potential  user  
community  is  usually  not  specifically  known.  
 
2. An  application  development  that  is  internal  to  a  specific  organization  
carried  out  by  internal/external  resources  and  where  the  actual  user  
community  is  specifically  known.  

In  the  case  of  (1)  the  development  project  typically  involves  one  or  more  
‘developer’  roles  –  coders,  user  interface  designers  etc.  –  but  may  not  involve  
any  business  analysts  or  actual  users  other  than  possibly  a  few  ‘beta’  users  
who  have  agreed  to  participate  in  some  kind  of  testing  program  prior  to  the  
application’s  formal  release.  

In  the  case  of  (2)  the  development  project  typically  involves  three  parties:  

1. The  developer(s)  
2. The  business  analyst(s)  
3. The  business  user(s)/subject  matter  or  domain  expert(s)    

Case  (2)  may  be  a  more  complex  and  ‘overseen’  project  because  often  more  
people/roles  are  involved  and  specific  departmental/organizational  budgets  
are  being  consumed  to  fund  the  work.  Although  case  (1)  projects  can  be  
equally  large  and  complex,  today  many  so-­‐called  ‘app’  developments  can  

Stewart  McKie   -­‐  39  -­‐   As  of  15/08/2014  

essentially  only  involve  one  person,  which  is  the  case  here.    

So  here,  unlike  many  case  (1)  or  (2)  projects,  a  team  of  developers  did  not  
create  the  deliverables.  In  most  cases,  my  prototype  deliverables  involved  a  
partnership  –  involving  myself  as  ‘designer’  and  ‘data  modeler’,  working  
with  a  developer  partner  as  ‘coder’.    The  users  were  represented  only  by  
those  people  who  managed  to  find  and  sign  up  for  the  applications  online.  In  
this  specific  case,  I  represented  the  developer,  the  business  analyst  and  the  
(screenwriting)  domain  expert.  In  reality  this  is  not  necessarily  a  good  idea  
as  it  tends  to  encourage  solipsistic  practice.  

As  with  virtually  all  software  developments,  large  chunks  of  code  were  not  
developed  either  by  myself  or  by  my  coding  partners  but  by  others  who  
have  made  their  efforts  available  to  all  through  various  kinds  of  open  source  
and  commercial  licenses.  Use  of  or  adaptation  of  these  open  source/licensed  
components  is  standard  practice  in  the  commercial  software  industry.    

There  are  essentially  three  skillsets  used  to  deliver  the  kind  of  software  
application  prototypes  necessary  for  this  PhD,  reflecting  the  so-­‐called  ‘3-­‐tier’  
development  architecture  comprising:  

1. Presentation  tier  skillset:  user  interface  design  and  coding  

2. Business  logic  tier  skillset:  business  rules  /event  design  and  coding  
3. Data  management  tier  skillset:  database  design  and  coding    

In  terms  of  the  early  prototype  deliverables  I  acted  as  the  ‘designer’  for  all  
three  tiers  and  the  ‘coder’  for  tier  3.  As  I  became  a  reasonably  proficient  PHP  
programmer  myself  during  my  PhD  studies,  I  acted  as  the  designer  and  
coder  for  all  three  tiers  in  the  later  deliverables.    

Because  a  development  team  was  not  involved  in  the  creation  of  the  
prototype  applications  there  was  no  need  to  use  one  of  the  formal  software  
application  development  methodologies  that  are  usually  adopted  in  team-­‐
centric  developments,  for  example  the  Scrum  methodology  (see  Sims  2012).  
However  in  order  to  relate  some  of  what  was  done  to  a  popular  software  
development  methodologies,  I  will  indicate  how  there  was  some  overlap  

Stewart  McKie   -­‐  40  -­‐   As  of  15/08/2014  

between  my  practice  and  that  of  certain  software  development  
methodologies.        

In  Scrum  role  terms  I  acted  as  a  combination  of  ‘product  owner’  and  ‘team  
member’.  However  since  no  formal  team-­‐centric  Scrum  process  was  
followed  I  did  not  perform  the  role  of  ‘scrum  master’  since  no  actual  scrums  
took  place.  No  Scrum  artefacts,  such  as  user  stories,  backlogs  and  tasks  were  
created  nor  was  a  scrum  ‘sprint’  cycle  followed.      

In  terms  of  the  Agile  manifesto  (see  

http://agilemanifesto.org/principles.html  accessed  5  February  2014)  only  
two  of  the  12  principles  are  applicable  here:  

Deliver  working  software  frequently,  from  a  couple  of  weeks  to  a  

couple  of  months,  with  a  preference  to  the  shorter  timescale.  

Working  software  is  the  primary  measure  of  progress.  

And  in  terms  of  Agile  values,  both  ‘Working  software  over  comprehensive  
documentation’  and  ‘Responding  to  change  over  following  a  plan’  informed  
the  deliverable  developments.  Similarly,  in  terms  of  the  Crystal  Clear  
properties  (Cockburn  2005:17),  my  practice  espoused  both  ‘Frequent  
Delivery’  and  ‘Reflective  Improvement’.  

But  fundamentally  the  core  methodology  was  simply  that  of  an  iterative  
development  throughout  the  duration  of  my  research,  whereby  each  
prototype  iteration  became  functionally  richer  as  my  research  progressed  
with  the  final  ‘submission’  deliverable  being  the  most  functionally  rich  of  all.  
None  of  the  functional  development  was  in  direct  response  to  user  
request/demand  since  this  did  not  happen.  

Development  Choices  

As  with  any  software  development  it  was  necessary  to  make  choices  as  to  
the  technology  used  to  develop  with  and  the  platforms  targeted  to  develop  
for.  Given  that  this  was  largely  a  development  managed  and  delivered  by  

Stewart  McKie   -­‐  41  -­‐   As  of  15/08/2014  

myself,  it  is  natural  that  my  own  skillset  and  experience  played  a  significant  
role  in  my  development  choices.  

From  a  practice  perspective,  I  faced  some  challenges.  Any  screenplay  

content  needed  to  be  ‘datafied’  and  this  datafied  content  had  to  be  accessible  
to  analytic  and  visualization  tools.    I  had  to  be  confident  that  I  could  manage  
this  datafication  and  that  the  datafied  content  itself  would  be  reasonably  
easy  to  analyze  using  off-­‐the-­‐shelf  tools.    I  am  not  a  computer  scientist  so  I  
am  not  trying  to  invent  new  algorithms  for  these  purposes,  merely  to  use  
readily  available,  popular  and  low-­‐cost  or  free  open  source  technology  
wherever  possible.  

In  terms  of  the  3-­‐tier  application  delivery  architecture  outlined  above  there  
are  at  least  three  main  development  technology  choices  to  be  made:  

1. The  delivery  user  interface  (UI)  and  target  user  device  

2. The  coding  language  to  be  used  for  the  client  and  server  business  logic  
code  
3. The  database  management  system  (DBMS)  to  be  used  for  data  storage  
and  management  

At  the  time  of  my  PhD,  the  most  popular  and  pervasive,  delivery  user  
interfaces  and  devices  include:  

1. Web  UI  –  including  any  device  capable  of  running  a  web  browser  
2. Mac  –  the  Apple  Mac  operating  system  (OSX)  and  Mac  computer  
3. Windows  –  the  Microsoft  Windows  operating  system  and  a  PC  
4. Linux  –  the  Linux  operating  system  and  any  device  capable  of  running  
Linux  
5. IOS  –  the  operating  system  for  the  Apple  iPhone  and  iPad  devices  
6. Android  –  devices  that  support  Google  Android  operating  system    

Developing  for  (5)  and  (6)  requires  a  relatively  specialist  and  expensive  
skillset  that  I  did  not  have  and  also  is  intended  to  create  applications  
primarily  designed  to  operate  on  mobile  phones  and  tablets.    There  are  
screenwriting  applications  that  run  on  mobile  phones  and  tablets  but  in  my  
view  this  is  not  the  primary  target  device  for  this  kind  of  application  use  
case.    

Stewart  McKie   -­‐  42  -­‐   As  of  15/08/2014  

Developing  for  2-­‐4  also  requires  a  relatively  specialist  and  expensive  skillset  
that  I  did  not  have  but  clearly  these  are  key  target  devices  for  screenwriting  
applications  as  many  commercially  available  screenwriting  applications  
were  designed  to  run  on  Macs  or  Windows  PCs  or  both.  

I  decided  to  focus  on  (1)  because  I  had  prior  experience  of  developing  web-­‐
based  applications,  web  applications  run  on  most  devices  (both  ‘tethered’  
and  mobile)  and  the  costs  and  barriers  to  entry  are  low.  

There  are  a  number  of  programming  languages  I  could  have  used  for  client  
and  server  side  development  but  I  decided  to  use  JavaScript/HTML/CSS  on  
the  client  and  PHP  on  the  server  as  again  I  had  some  familiarity  with  these  
languages,  they  are  free  to  use  and  there  are  many  free/low  cost  
components  available  in  the  open  source  community  to  facilitate  
development  using  these  languages.    I  certainly  did  not  want  to  have  to  learn  
a  wholly  new  set  of  programming  skills  as  part  of  my  PhD  by  practice.  

There  are  a  number  of  databases  (e.g.  SQL  and  NoSQL)  or  file  systems  (e.g.  
XML)  that  I  could  have  used  for  data  storage  and  management  but  I  decided  
to  use  a  structured  query  language  (SQL)  based  relational  database  
management  system  (RDBMS)  as  the  database  engine  because  I  understand  
SQL  and  how  to  design  and  work  with  relational  data  models.  I  chose  to  use  
the  open  source  MySQL  database  for  data  storage  because  it  is  reliable,  
scalable  and  free  to  use  and  also  available  on  the  Internet  Service  Provider  
(ISP)  hosting  account  that  I  use  to  host  and  deliver  my  online  applications.    

Community  of  Interest    

Any  software  development  essentially  involves  two  parties:  the  

developers/analysts  and  the  users.  In  the  development  of  software  within  a  
specific  business  these  parties  are  easy  to  define  –  usually  the  
developers/analysts  sit  within  the  IT  department  and  the  users  sit  within  
the  operational  line  of  business  units.  In  this  case,  my  partner  coder  and  I  
functioned  as  the  developer  party  but  a  decision  needed  to  be  made  as  to  

Stewart  McKie   -­‐  43  -­‐   As  of  15/08/2014  

who  to  target  as  the  user  party.    

Here  my  methodology  is  informed  by  Etienne  Wenger’s  concept  of  a  
community  of  practice  (see  http://wenger-­‐trayner.com,  A  brief  introduction  
to  Communities  of  practice,  accessed  5  February  2014).    My  ‘domain  of  
interest  ‘  is  that  of  screenwriting  in  which  the  ‘shared  competence’,  in  
relation  to  my  research  question,  is  specifically  that  of  writing  and/or  
analyzing  movie  screenplays  using  a  software  application.    Clearly  there  are  
at  least  four  potential  user  communities  that  exist  in  this  domain  with  
specific  kinds  of  competency  focus:  

1. Writers  –  i.e.  practicing  amateur  or  professional  screenwriters  who  

write  or  want  to  write  screenplays  for  a  hobby  or  a  living.  
 
2. Analysts  –  i.e.  individuals  in  the  business  of  analyzing  screenplays  
including  professional  agents,  readers  and  other  ‘story  analysts’.  
 
3. Teachers  &  Learners  –  i.e.  those  involved  in  the  teaching  and  learning  
of  screenwriting  practice  at  school/college/university.  
 
4. Producers  –  i.e.  the  cast  and  crew  engaged  in  realizing  an  audio/visual  
movie  product  from  a  script  that  acts  as  the  production  ‘plan’.  

I  initially  rejected  (4)  as  a  potential  user  community.  The  reason  for  this  was  
that  the  ‘producing’  community  is  less  interested  in  writing  and  analyzing  
the  content  of  a  screenplay  and  more  in  the  practical  effort  of  converting  the  
screenplay  into  a  finished  movie.  In  software  functionality  terms,  this  user  
community  is  more  concerned  with  tasks  such  as  generating  prop  lists  and  
daily  call  sheets  and  with  managing  script  revisions  and  monitoring  the  
production  budget.  This  was  not  functionality  that  I  initially  intended  to  
deliver  and  so  this  community  was  not  a  good  fit  in  terms  of  my  research  
question.  

I  initially  thought  that  (1)  and  especially  (2)  would  be  my  primary  focus.  
However,  the  anecdotal  response  to  contacts  I  made  with  a  handful  of  
practitioners  I  knew  or  reached  out  to  in  the  ‘analysts’  domain  was  not  
positive.    The  general  view  seemed  to  be  a  lack  of  interest  in  tools  that  
helped  to  analyze  a  screenplay,  whether  because  this  was  seen  as  

Stewart  McKie   -­‐  44  -­‐   As  of  15/08/2014  

antithetical  to  the  author/analyst’s  ‘creative’  process  or  some  kind  of  threat  
to  the  validity  or  commercial  viability  of  fee-­‐based  qualitative  screenplay  
analysis  services.    

This  feedback  led  me  to  focus  initially  on  (1)  and  as  I  did  not  know  many  
screenwriting  practitioners  personally  I  focused  my  efforts  on  recruiting  a  
user  community  by  reaching  out  to  various  screenwriting-­‐focused  bloggers  
and  those  online  communities  that  appeared  to  include  a  significant  number  
of  screenwriters.  Usually  this  was  by  means  of  a  direct  contact  email,  posting  
comments  on  other  people’s  blogs  or  a  posting  a  solicitation  to  try  out  a  
prototype  on  the  online  forum  of  a  screenwriting  user  community.          

One  screenwriting  blogger  who  did  engage  with  me  initially  was  Alex  
Epstein,  author  of  the  Crafty  Screenwriting  books.  He  made  a  few  posts  that  
referred  to  my  work  that  attracted  a  handful  of  comments  from  members  of  
his  blog  community–  see:  

• http://complicationsensue.blogspot.co.uk/2006/11/polling-­‐
numbers.html  
• http://complicationsensue.blogspot.co.uk/2007/03/ive-­‐looked-­‐at-­‐
clouds-­‐that-­‐way.html  
• http://complicationsensue.blogspot.co.uk/2008/08/scriptgeist.html  

(All  above  accessed  5  February  2014)  

I  also  joined  and  participated  in  the  online  community  at  

http://shootingpeople.org  to  solicit  screenwriters  to  try  the  initial  
scriptcloud.com  and  scriptgeist.com  prototypes.    As  this  community  requires  
users  to  pay  a  subscription  fee  to  participate,  I  assumed  this  would  be  
indicative  of  higher  quality  and  more  engaged  participants.  And  in  fact  many  
of  the  hundreds  of  users  who  signed  up  to  use  my  early  prototypes  did  
originate  from  the  Shooting  People  community.  However,  it  appeared  that  
over  time  people  became  disinterested  in  the  potential  of  screenwriting  2.0  
technology.  In  addition,  no  blog  posts  I  made  on  my  own  blog  at  
phd.tripos.biz  or  any  of  my  prototype  applications  ever  gained  much  
feedback  from  users  and  certainly  never  went  ‘viral’  in  any  way.  

Stewart  McKie   -­‐  45  -­‐   As  of  15/08/2014  

What  quickly  became  clear  to  me  was  that  any  user  community  I  could  
engage  with  was  less  a  community  of  practice  and  more  a  community  of  
interest.  As  Wenger  states  (p.2),  ‘A  website  in  itself  is  not  a  community  of  
practice’  and  ‘A  community  of  practice  is  not  merely  a  community  of  
interest-­‐-­‐people  who  like  certain  kinds  of  movies,  for  instance.’  

So  despite  providing  feedback  forms  on  my  prototypes  for  users  and  open  
commenting  on  my  blog  (originally)  I  quickly  found  that  the  few  users  who  
did  engage  with  the  prototypes  were  not  that  interested  in  developing  
shared  practice  in  the  context  of  my  research  question  but  merely  in  
‘dabbling’  with  it.    It  was  clear  that  I  was  working  with  a  kind  of  community  
of  interest  (CoI),  along  the  lines  of  that  discussed  by  Fischer,  (see  Gerhard  
Fischer,  EXTERNAL  AND  SHAREABLE  ARTIFACTS  AS  OPPORTUNITIES  FOR  
SOCIAL  CREATIVITY  IN  COMMUNITIES  OF  INTEREST  
http://l3d.cs.colorado.edu/~gerhard/papers/ccmcd2001.pdf  ,  accessed  5  
February  2014)  rather  than  a  community  of  practice  (CoP).  

Fischer  (p.4)  states  that  communities  of  interest  (CoIs):  

are  “defined”  by  their  shared  interest  in  the  framing  and  resolution  of  a  
design  problem.  CoIs  often  are  more  temporary  than  CoPs:  they  come  
together  in  the  context  of  a  specific  project  and  dissolve  after  the  
project  has  ended.  CoIs  have  great  potential  to  be  more  innovative  and  
more  transforming  than  a  single  CoP  if  they  can  exploit  the  “symmetry  
of  ignorance”  as  a  opportunity  for  social  creativity.

I  had  hoped  that  the  apparent    ‘symmetry  of  ignorance’  about,  in  particular,  
social  screenwriting  and  screenplay  analytics  (via  software)  might  lead  to  
some  social  creativity.    But  I  was  proved  wrong.  

Certainly  my  potential  CoI  did  not  satisfy  Fischer’s  assertion  that  ‘CoIs  
[Fischer,  2001]  bring  together  stakeholders  from  different  CoPs  to  solve  a  
particular  (design)  problem  of  common  concern’,  since  it  is  doubtful  
whether  either  the  ‘developers’  or  ‘users’  in  this  case  can  be  considered  to  be  
examples  of  CoPs.  And  perhaps  this  was  because  I  did  not  recognize  or  
confront  a  specific  difficulty  faced  by  a  CoI,  according  to  Fischer:  

Stewart  McKie   -­‐  46  -­‐   As  of  15/08/2014  

 Fundamental  challenges  facing  CoIs  are  found  in  building  a  shared  
understanding  [Resnick  et  al.,  1991]  of  the  task-­‐at-­‐hand,  which  often  
does  not  exist  at  the  beginning,  but  is  evolved  incrementally  and  
collaboratively  and  emerges  in  people’s  minds  and  in  external  artifacts.

The  concept  of  Screenwriting  2.0  and  specifically  the  practice  of  screenplay  
analytics  on  datafied  scripts  were  not  mainstream  ideas  when  I  started  this  
PhD  and  I  largely  failed  to  facilitate  the  incremental  and  collaborative  
evolution  of  either  with  those  users  that  did  engage  with  my  prototypes.  

Remote  vs.  Local  Usage  

Another  issue  that  every  developer  of  speculative  online  applications  faces  is  
that  their  user  community  is  by  definition  remote  and  often,  as  in  this  case,  
not  linked  by  any  shared  bonds  e.g.  part  of  a  family  or  business  or  other  
recognizable  offline  community.    This  led  me  to  reconsider  reaching  out  to  
the  ‘teachers  &  learners’  user  group  since  in  this  case  I  expected  to  be  able  to  
work  locally  and  face-­‐to-­‐face  with  small  groups  of  users,  rather  than  only  
remotely.  As  a  result,  I  engaged  with  3  specific  ‘educational’  user  groups:  

1. An  after-­‐school  screenwriting  class  at  my  local  secondary  school  

(Shaftesbury  School  in  Dorset)  comprising  year  9-­‐11  students.  

2. Undergraduate-­‐level  student  groups  at  two  universities  

(Bournemouth  and  University  College,  Falmouth).  

3. A  graduate-­‐level  student  group  at  Royal  Holloway  College,  London.  

In  the  case  of  (1)  and  (2)  the  groups  were  shown  the  Scenewrite  prototypes,  
and  one  or  two  students  within  each  group  subsequently  used  the  
applications  to  write  and  analyze  their  own  screenplay  scenes.  However  no  
student  continued  to  use  Scenewrite  after  an  initial  try.    

In  the  case  of  (3),  all  students  did  actively  engage  with  the  Sceneclass  
prototype  (a  Wordpress  add-­‐in)  during  the  session.  Yet  despite  clearly  
favourable  feedback  on  the  day,  again  no  student  continued  to  use  
Sceneclass  after  their  initial  try.  On  the  basis  of  this  experience  I  decided  not  
to  focus  more  attention  on  the  potential  teaching  and  learning  user  
community  as  it  was  not  clear  that  the  results  with  these  three  user  

Stewart  McKie   -­‐  47  -­‐   As  of  15/08/2014  

communities  face-­‐to-­‐face  were  likely  to  be  significantly  more  positive  than  
working  with  a  remote  user  community.  

It  is  fair  to  say  that  despite  numerous  remote  and  local  attempts  to  create  
and  evolve  a  community  of  practice  or  a  community  of  interest,  I  was  
unsuccessful  at  doing  this  with  any  of  the  three  possible  ‘target’  user  
communities  I  identified  at  the  start  of  this  PhD.  In  fact  it  seemed  to  me  that  
the  more  functionality  I  delivered  -­‐  that  in  my  view,  validated  the  utility  of  
screenwriting  2.0  and  screenplay  analytics  in  particular  -­‐  the  less  users  
engaged  with  the  prototypes  and  the  fewer  users  that  were  attracted  to  even  
try  the  deliverables.    

Prototyping  as  a  Development  Methodology  

There  are  a  number  of  software  development  life  cycle  (SDLC)  models  that  
can  be  used  to  inform  software  application  development  including  Agile,  
Big-­‐Bang,  Iterative,  Prototyping,  RAD,  Spiral  and  V-­‐Model.  My  intention  is  
not  to  discuss  these  models  here,  as  they  all  have  pros  and  cons  in  relation  to  
a  specific  development  project,  but  simply  to  state  that  I  chose  to  use  
Prototyping  as  my  primary  informing  model.    

In  a  presentation  about  chapter  8,  Prototyping  and  Rapid  Application  

Development,  from  the  fifth  edition  of  their  book  Systems  Analysis  and  Design,  
Kendall  and  Kendall  (2002)  define  prototyping  as  ‘an  information  gathering  
technique  and  as  useful  in  seeking  user  reactions,  suggestions,  innovations  
and  revision  plans’  (see  slide  8-­‐3  in  
http://www.mgt.ncu.edu.tw/~ylchen/sasd-­‐slide/chap08.ppt  ,  accessed  11  
March  2014).    

Like  Kendall  and  Kendall,  Beaudouin-­‐Lafon  and  Mackay  in  chapter  52  of  
Protoyping  Tools  and  Techniques  (see  
https://www.lri.fr/~mackay/pdffiles/Prototype.chapter.pdf  ,  accessed  11  
March  2014)  also  emphasize  (p.1-­‐2)  the  utility  of  a  prototype  to  support  
‘creativity,  communication’  and  ‘early  evaluation’  (of  design  options).  They  
differentiate  between  ‘offline’  and  ‘online’  prototype  representations  that  

Stewart  McKie   -­‐  48  -­‐   As  of  15/08/2014  

either  require  or  do  not  require  a  computer.  They  also  mention  
‘evolutionary  prototypes’  (p.1-­‐6)  i.e.  prototypes  that  evolve  into  a  final  
application  deliverable  and  sometimes  involve  ‘participatory  (also  called  
Cooperative)  design’  (p1-­‐7)  whereby  users  collaborate  with  designers  to  
more  or  less  co-­‐create  the  application  together.  

In  The  qualitative  research  process,  Bruhn  Jensen  (2002:236)  makes  the  

point  that  ‘Qualitative  researchers  tend  to  conceive  of  their  studies,  most  
generally,  as  an  iterative  or  repeated  process’,  which  fits  well  with  the  
practice  of  evolutionary  prototyping  in  software  development.  And  in  Bruhn  
Jensen’s  terms,  prototyping  can  be  seen  as  performing  both  a  ‘sampling’  
function  in  the  sense  that  each  prototype  presents  a  sample  of  functionality  
that  users  may  respond  to  or  not  and  each  prototype  also  creates  a  
‘naturalistic  context’  since  by  definition  the  application  itself  is  the  context  in  
which  the  users  work.  

My  prototypes  are  functional,  interactive  online  applications,  not  offline,  

static  and  non-­‐functional  ‘wireframes’  or  paper-­‐based  diagrams  merely  
indicative  or  illustrative  of  a  user  interface  or  functionality.  And  my  
prototypes  are  evolutionary  in  nature  because,  iteratively,  they  became  
more  functional  over  time  -­‐  from  Scriptcloud  to  Scenepad  (as  discussed  in  
more  detail  in  chapter  5  below  -­‐  The  Road  to  Scenepad).  

The  table  below  shows  how  the  application  functionality  was  evolved  from  
basic  word  clouds  in  Scriptcloud,  to  analytic  charts  in  Scriptgeist,  to  basic  
social  scenewriting  in  Sceneclass  and  finally  to  more  or  less  complete  
screenwriting  applications  in  Scenewrite  and  Scenepad.  The  evolution  was  
not  done  by  enhancing  a  single  deliverable  from  ‘first  cut’  to  ‘final  cut’  
prototype.  This  may  have  been  an  error  of  judgment  on  my  part.  As  one  of  
my  supervisors  (Adam  Ganz)  suggested,  perhaps  it  was  a  mistake  to  deliver  
a  number  of  different  prototypes  each  with  different  names  and  web  site  
addresses.  Because  by  not  focusing  on  a  single  prototype  all  the  way  through  
my  research  I  was  confusing  users  and  forcing  them  to  keep  switching  
applications  thereby  limiting  their  longevity  as  users.  This  may  be  a  factor  in  

Stewart  McKie   -­‐  49  -­‐   As  of  15/08/2014  

relation  to  why  the  user  count  went  down  rather  than  up  as  the  prototype  
evolved,  as  users  did  not  automatically  ‘carryover’  from  one  prototype  to  the  
other  and  became  harder  to  convince  as  the  functional  complexity  became  
harder  to  understand.  Unfortunately  for  various  reasons,  including  lack  of  
developer  commitment/availability  forcing  me  to  change  development  
partner  with  each  prototype  version  and  availability  of  funding  to  
underwrite  the  earlier  coding  effort,  it  was  easier  or  necessary  to  start  again  
and  build  from  scratch  with  each  new  prototype.    

FUNCTION   Scriptcloud   Scriptgeist   Sceneclass   Scenewrite   Scenepad  

           
Upload  script  (Text)   X   X   -­‐   -­‐   -­‐  

Generate  word  cloud   X   X   -­‐   X   X  

Generate  script  charts   -­‐   X   -­‐   X   X  

Generate  script  entities   -­‐   X   -­‐   X   X  

Write  script  scenes   -­‐   -­‐   X   X   X  

Scene  commenting   -­‐   -­‐   X   X   X  

Scene  ‘liking’   -­‐   -­‐   X   X   X  

Script  storyboards   -­‐   -­‐   -­‐   X   X  

Script  multi-­‐level  PDF   -­‐   -­‐   -­‐   X   X  

output  
Script  business   -­‐   -­‐   -­‐   X   X  
manager  
Script  content  hubs   -­‐   -­‐   -­‐   X   X  

Script  scene  versioning   -­‐   -­‐   -­‐   X   X  

Upload  script  (FDX)   -­‐   -­‐   -­‐   -­‐   X  

Script  mobile-­‐access   -­‐   -­‐   -­‐   -­‐   X  

site  
Script  sentiment   -­‐   -­‐   -­‐   -­‐   X  
analysis  
Script  production   -­‐   -­‐   -­‐   -­‐   X  
manager  

Stewart  McKie   -­‐  50  -­‐   As  of  15/08/2014  

Table  1.1  –  Functional  evolution  of  prototype  deliverables  

In  conclusion,  to  use  Bruhn  Jensen’s  terms  (2002:237),  my  research  design  
methodology  can  be  summarized  as:  

• Strategy  –  to  deliver  a  series  of  evolutionary  online  software  

prototypes  to  test  the  utility  of  functionality  that  relates  to  my  
research  question.  
 
• Tactics  –  to  expose  these  prototypes  to  communities  of  interest  
comprising  users  that  are  most  likely  to  engage  with  the  prototypes.  
 
• Techniques    -­‐  to  collect  verbal  and  virtual  feedback  from  these  users  
and  reflect  on/respond  to  this  feedback  to  iteratively  evolve  the  
design  of  the  prototypes.  

My  first  development  challenge  was  to  figure  out  how  to  ‘datafy’  screenplay  
content  effectively,  in  order  to  facilitate  screenplay  analytics  and  other  
functionality  that  depends  on  the  availability  of  ‘screenplay  as  data’  rather  
than  ‘screenplay  as  document’.    

Stewart  McKie   -­‐  51  -­‐   As  of  15/08/2014  

1  Screenplay  as  Data  

We  are  not  used  to  thinking  of  screenplays  as  data  but  rather  as  documents.    

As  Millard  says  in  The  screenplay  as  prototype,  ‘The  screenplay  is  a  document  
that  exists  as  a  carry-­‐over  from  a  pre-­‐digital  era’  (in  Analysing  the  Screenplay  
2011:  146).    Here,  ‘screenplay  as  data’  means  decomposing  and  digitally  
storing  the  screenplay’s  textual  content  as  a  collection  of  individual  data  
points  at  a  level  of  granularity  that  facilitates  reassembling  of  these  data  
points  in  various  ways  to  present  various  kinds  of  ‘wholes’  that  represent  
the  sum  of  various  ‘parts’:  In  short,  the  datafication  of  screenplay  content.  

Typically  we  think  of  and  envision  screenplays  as  documents.  We  often  use  
them  printed  out  on  paper.    There  is  a  great  deal  of  attention  focused  on  the  
look  and  feel  of  this  venerable  document  -­‐  use  of  the  Courier  12  font,  binding  
the  pages  using  brass  brads,  and  industry  standard  formatting  conventions  
for  the  presentational  layout  of  the  text  on  the  page.  Screenwriting  
technology  has  largely  adopted  the  physical  paper-­‐document  format  as  the  
template  digital  format  for  a  screenplay  when  viewed  and  managed  on-­‐
screen  therefore  screenplays  continue  to  be  essentially  presented,  defined  
and  discussed  as  document  artefacts,  whether  paper  or  digital.  But  a  
document-­‐centric  artefact  is  not  the  same  as  a  data-­‐centric  artefact.    

We  consider  documents  as  artefacts  designed  for  reading  by  a  human  

‘consumer’  of  the  document  content.  It  is  the  human  process  of  reading  that  
is  used  to  access  the  information  the  document  communicates.  So  a  
document-­‐centric  screenplay  is  primarily  intended  for  visual  consumption  
and  manual  use  by  humans  -­‐  for  example  to  be  read  and  annotated  with  a  
pen  or  highlighter  and  to  allow  additional  (or  revision)  physical  pages  to  be  
easily  inserted  into  and  ‘cut’  pages  to  be  removed  from,  the  printed  script.    

There  are  a  number  of  ways  of  presenting  all  or  parts  of  script  content  for  
human  consumption  on  physical  media,  for  example:  

Stewart  McKie   -­‐  52  -­‐   As  of  15/08/2014  

 • As  a  paper  document  printed  out  for  reading  

• As  a  series  of  3x5  index  cards  pr  post-­‐it  notes  outlining  the  content  of  
a  series  of  scenes  

• As  hand-­‐drawn  boards  representing  a  storyboard  visualization  of  

scenes  

Some  or  all  of  these  views  of  screenplay  content  are  represented  digitally  in  
many  of  the  current  generation  of  screenwriting  tools.  The  paper  
representation  has  been  transitioned  almost  directly  to  a  digital  
representation.  The  script  has  become  a  document  format  (e.g.  PDF)  that  can  
be  stored  online  and  scrolled  through  and  searched  on-­‐screen.  Index  cards  
are  visually  represented  on-­‐screen  and  display  selected  data  about  a  scene.  
Storyboards  are  presented  as  a  collection  or  portfolio  of  digital  images  
rather  than  physical,  hand-­‐drawn  sketches.  Virtual  cards  and  boards  can  be  
‘dragged  and  dropped’  with  a  mouse  to  easily  re-­‐order  them  on  screen  to  
simulate  the  physical  process  of  unpinning  and  repositioning  them  on  a  
corkboard.    

But  in  software  development  terms,  a  document-­‐centric  artefact  concerns  

itself  primarily  with  only  the  ‘surface’  presentation  layer  of  the  ‘3-­‐tier’  
client/server  architecture  and  design  pattern  –  originated  by  John  J.  
Donovan  at  the  Open  Environment  Corporation  (1994)  -­‐  comprising  
presentation,  application  and  data  tiers.  

In  a  3-­‐tier  architecture,  the  presentation  tier  is  what  an  application  user  sees  
and  interacts  with  e.g.  the  user  interface  (UI)  of  a  software  application.  The  
application  tier  contains  the  rules  and  business  logic  that  define  how  the  
data  is  extracted  from  the  data  tier,  how  you  can  see  what  has  been  
retrieved  and  what  you  can  do  with  what  you  can  see.  The  data  tier  manages  
the  data  (stored  in  files  or  a  database  for  example)  that  is  actually  presented  
to  the  user  through  the  lens  of  the  user  interface  and  as  a  result  of  the  
application  of  the  business  logic  (for  example  you  may  only  see  certain  data  

Stewart  McKie   -­‐  53  -­‐   As  of  15/08/2014  

that  is  based  on  business  logic  that  defines  your  access  rights  according  to  
your  application  user  role).    

A  document-­‐centric  artifact  is  focused  on  the  presentation  or  ‘look  and  feel’  
layer  of  the  content  rather  than  on  the  content  business  rules  and  data  and  
on  the  document  as  a  whole,  as  a  container  of  parts,  rather  than  as  a  network  
of  individual  elements  and  their  relationships.  In  this  sense  the  screenplay  
as  document  places  more  importance  on  the  script  as  whole  rather  than  the  
parts  of  the  script  –  the  watch  face  rather  than  the  components  that  make  it  
function.    

But  a  document  centric  artefact  is  not  the  same  as  a  data-­‐centric  artefact.  

1.1  The  Data-­‐Centric  Screenplay  

The  document-­‐centric  screenplay  is  not  designed  or  intended  for  machine  
processing.    There  are  a  number  of  ways  of  decomposing  screenplay  content  
to  make  the  content  better  suited  for  machine  (i.e.  computer)  processing,  for  
example:  

• As  a  combination  of  data  and  metadata  

• As  a  hierarchy  of  relationships    

• As  text  ‘marked  up’  with  identification  ‘tags’  

• As  text  ‘marked  down’  with  specific  characters  to  trigger  specific  

presentation  formatting  rules  

• As  text  tagged  as  parts  of  speech  (i.e.  dialog  text  identified  as  verbs  
and  nouns)  

A  data-­‐centric  artefact  is  not  specifically  concerned  with  the  presentation  

and  navigation  of  the  document  as  a  whole.  That  is  something  that  will  itself  
be  machine-­‐processed  when  the  data  is  rendered.  A  data  centric  artefact  is  
designed  both  for  analysis  as  a  whole  (the  ‘forest’)  and  in  parts  (the  ‘trees’)  

Stewart  McKie   -­‐  54  -­‐   As  of  15/08/2014  

and  recognizes  that  a  range  of  data  relationships  may  need  to  be  defined  and  
maintained  are  drivers  of,  but  otherwise  nothing  to  do  with  the  surface  
presentation  content  of  the  artefact  but  which  function  to  facilitate  its  
analysis  and  visualization  in  ways  other  than  merely  as  an  on-­‐screen  
document.  

You  are  even  less  likely  to  find  screenplays  discussed  as  the  specification  of  a  
physical  piece  of  information,  which  is  the  definition  of  an  ‘artifact’  in  the  
Unified  Modeling  Language  (UML),  the  de-­‐facto  standard  for  building  Object-­‐
Oriented  software.    The  Object  Management  Group  (OMG)  (see  
www.omg.org,  accessed  12  February  2014),  who  manage  the  specification  of  
UML,  state:  

The  Unified  Modeling  Language  (UML)  is  a  graphical  language  for  

visualizing,  specifying,  constructing,  and  documenting  the  artifacts  of  a  
software-­‐intensive  system.  The  UML  offers  a  standard  way  to  write  a  
system's  blueprints,  including  conceptual  things  such  as  business  
processes  and  system  functions  as  well  as  concrete  things  such  as  
programming  language  statements,  database  schemas,  and  reusable  
software  components.    

You  could  quite  easily  apply  this  definition  to  screenplays  by  rewriting  this  
OMG  statement  as  follows:  

A  screenplay  is  a  textual  language  for  visualizing,  specifying,  

constructing,  and  documenting  the  audio-­‐visual  artifacts  of  a  movie.  
The  screenplay  offers  a  standard  way  to  write  a  movie's  blueprints,  
including  conceptual  things  such  as  scenes,  shots  and  transitions  as  
well  as  concrete  things  such  as  people,  props  and  visual  effects.  

A  screenplay  uses  a  relatively  sparse  ‘language’  to  define  a  means  of  

realizing  an  end  product  –  in  this  case  a  movie  rather  than  a  software  
application.  But  a  screenplay  is  more  than  a  document  defined  by  a  set  of  
formatting  rules  that  determine  the  way  it  is  presented,  it  already  comprises  
two  core  kinds  of  potential  data  content:  data  and  metadata.  

1.2  Data  and  Metadata  

The  data  of  a  screenplay  comprises  only  the  blocks  of  scene  description  or  

Stewart  McKie   -­‐  55  -­‐   As  of  15/08/2014  

action  and  the  blocks  of  character  dialog  that  make  up  the  main  content  of  
scene.  Everything  else  –  scene  headings,  captions,  character  cues,  
parentheticals  and  transitions  etc.  –  is  metadata,  because  it  provides  context  
for  the  data  in  terms  of  its  on-­‐(cinema)  screen  presentation.    This  
combination  of  data  vs.  metadata  is  an  essential  characteristic  of  a  
screenplay,  although  it  is  generally  unrecognized  in  this  way.    And  a  
screenplay  has  considerably  more  metadata  than  say,  a  novel,  where  the  
only  metadata  is  likely  to  be  chapter  numbers/headings.  Novel  metadata  is  
implicit  within  the  text  rather  than  explicit  as  in  a  screenplay  where,  for  
example,  the  text  is  broken  up  by  explicitly  named  scenes,  character  cues  
and  transition  instructions  for  example,  that  are  not  found  in  a  novel  form.  

Metadata  is  what  helps  a  program  to  understand  and  utilize  data  by  
providing  essential  context.  A  simple  example  of  the  power  of  metadata  is  
how  it  is  used  by  Disney’s  Story  (see  http://story.us,  accessed  12  February  
2014),  an  iPhone  app  used  to  create  simple  storylines  from  the  masses  of  
photos  captured  on  an  iPhone  camera  roll.  The  app  uses  the  photo  metadata  
(captured  and  stored  with  the  photo  data)  to  link  photos  together  into  
proto-­‐narratives  or  storylines.  Metadata  such  as  the  date/time  and  location  
of  the  photo  is  leveraged  both  to  group  and  order  photos  into  ‘starter’  
storylines  that  the  user  can  improve  and  enhance  using  the  app.      

If  you  view  a  typical  content  page  in  a  screenplay  document  you  are  actually  
viewing  two  separate  kinds  of  data  content:  The  screenplay  data  and  the  
screenplay  metadata  that  describes  or  contextualizes  the  data  to  give  it  
coherence  and  meaning.    For  example,  consider  this  snippet  of  dialog  from  
Casablanca  (1942,  Michael  Curtiz):  

ILSA
Play it, Sam.

‘Play  it,  Sam.’  is  data.  ‘ILSA’  is  metadata  because  it  tells  us  something  about  
the  data  (i.e.  in  this  case,  which  character  speaks  it).    A  scene  heading  is  also  
an  example  of  explicit  metadata  in  a  screenplay  as  it  provides  some  essential  

Stewart  McKie   -­‐  56  -­‐   As  of  15/08/2014  

context  for  all  the  scene  content  that  follows.    

Imagine  what  a  given  screenplay  would  look  like  with  all  the  action  and  
dialog  blocks  removed.  First  the  script  content  would  be  more  compact  as  
the  bulk  of  a  screenplay  is  typically  contained  in  the  data,  not  the  metadata.  
What  would  be  left  is  a  series  of  scene  headings,  character  names  and  
parentheticals  and  transition  instructions.  What  might  be  called  the  
“framework”  of  the  script.  In  essence  you  would  have  the  skeleton  but  not  
the  organs  or  ‘guts’  of  the  narrative  of  a  movie.  Alternatively  you  could  have  
just  the  screenplay  data  –  comprising  a  continuous  series  of  action  and  
dialog  blocks  with  no  scene  headings,  character  names  and  parentheticals  
and  transition  instructions.    

One  can  try  to  film  a  screenplay  that  comprises  only  metadata  or  only  data.  
In  the  first  case  you  would  be  aware  of  specific  locations  and  time  settings,  
populated  by  specific  characters  but  they  would  do  and  say  nothing  since  
there  is  no  data  to  provide  their  actions  and  dialog.  In  the  second  case  you  
would  be  able  to  discern  some  kind  of  narrative,  it’s  just  that  the  movie  
could  take  place  anywhere,  at  anytime  with  any  characters  since  none  of  this  
is  specified  as  the  screenplay  metadata  is  missing.  Clearly  it’s  the  
combination  of  data  and  metadata  that  delivers  a  complete  movie  realization  
and  potentially  satisfying  experience  for  an  audience  rather  than  simply  one  
or  the  other.  

But  a  writer  writes  a  character  name,  not  metadata  and  writes  dialog,  not  
data.  As  humans  who  understand  the  purpose  of  this  content  we  do  not  need  
to  think  of  it  this  way  but  machines  need  more  help.    A  machine  needs  some  
explicit  structure  to  have  some  indication  of  what  the  metadata  or  data  is  
defined  as,  within  the  context  of  a  screenplay,  and  how  it  should  be  
formatted  when  rendered  on  screen/the  printed  page.      

1.3  Unstructured  vs.  Structured  Data  

If  you  create  a  screenplay  as  a  regular  document  in  a  standard  word  

Stewart  McKie   -­‐  57  -­‐   As  of  15/08/2014  

processor  like  Microsoft  Word,  you  are  creating  what  is  generally  regarded  
as  an  ‘unstructured’  data  artefact.  Apart  from  a  series  of  formatting  
instructions  e.g.  capitalize  and  center  this  text  (i.e.  to  indicate  a  character  
cue),  your  word  processing  document  has  no  clue  what  the  data  within  the  
document  actually  means  or  the  context  it  is  being  used  in.  For  example  if  
you  are  writing  a  James  Bond  movie  screenplay  in  Word,  your  word  
processing  software  has  no  way  of  knowing  that  the  word  BOND  in  capital  
letters  is  in  fact  the  name  of  a  character  in  your  screenplay.  

In  your  ‘unstructured’  screenplay  document  in  Word,  you  can  search  for  
BOND  and  you  can  use  bookmarking  to  notate  each  occurrence  of  the  word  
BOND  in  your  document  so  that  you  can  navigate  your  script  content  by  
character  cue.  But  if,  in  effect,  you  said  to  your  word  processor  ‘find  me  
every  bit  of  dialog  that  BOND  speaks’  it  would  have  no  clue  how  to  go  about  
this  since  it  neither  knows  what  constitutes  dialog  in  the  text  nor  that  BOND  
is  a  speaker  of  dialog.  However,  if  your  screenplay  is  stored  in  a  structured  
data  format  you  would  expect  a  reasonably  accurate  response  to  this  kind  of  
query  (if  submitted  in  the  correct  query  syntax  of  course).    

To  help  understand  the  difference  between  unstructured  and  structured  

data  in  screenwriting/screenplay  terms,  let’s  consider  again  the  dialog  
snippet  from  Casablanca  (1942).  

ILSA    
Play  it,  Sam.  

As  a  human  reader  of  the  snippet  with  my  own  mental  contextual  
framework  of  a  screenplay  to  reference,  I  understand  ILSA  to  be  a  character  
and  expect  a  dialog  block,  perhaps  preceded  by  a  parenthetical  instruction,  
to  follow.  A  software  program  used  to  format  and  analyze  a  screenplay  must  
be  explicitly  provided  with  its  own  kind  of  conceptual  framework  in  order  to  
understand  what  this  snippet  is  about.  In  other  words,  the  metadata  and  
data  contained  in  this  snippet  must  be  further  ‘structured’  in  some  way  to  
facilitate  this  understanding.  

Stewart  McKie   -­‐  58  -­‐   As  of  15/08/2014  

Let’s  save  the  dialog  snippet  above  in  three  common  ‘unstructured’  file  
formats  -­‐  .txt  and  .html  and  .rtf  -­‐  to  see  how  it  is  represented.  

Format   Content  

Text  (.txt)   ILSA    

Play  it,  Sam.  

HTML  (.html)   <center><strong>ILSA</strong></br>

Play it, Sam.</center>

Rich  Text   {\rtf1\ansi\ansicpg1252\cocoartf1138\cocoasubrtf510

Format  (.rtf)   {\fonttbl\f0\fswiss\fcharset0 Helvetica;}
{\colortbl;\red255\green255\blue255;}
\paperw11900\paperh16840\margl1440\margr1440\vieww10800\vie
wh8400\viewkind0
\pard\tx566\tx1133\tx1700\tx2267\tx2834\tx3401\tx3968\tx4535\tx5
102\tx5669\tx6236\tx6803\pardirnatural\qc
\f0\b\fs24 \cf0 ILSA
\b0 \
Play it, Sam.}

Table  1.2  –  Text  file  representations  

Other  than  the  line  return  to  separate  the  two  lines,  the  ‘plain’  text  file  has  
no  instructions  to  identify  what  this  data  is  or  how  it  should  be  formatted,  
which  is  why  it  is  not  centered  in  any  way.  The  HTML  file  is  ‘marked’  up  with  
standard  HTML  formatting  information  for  rendering  this  text  on  screen  in  a  
web  browser  (e.g.  <strong>  means  show  the  text  in  boldface)  using  start  and  
end  ‘tags’  (e.g.  <center>  and  </center>).  The  RTF  file  also  includes  additional  
‘rich’  formatting  information  for  rendering  this  text  on  screen.    But  none  of  
these  formats  have  any  kind  of  indication  of  what  this  metadata  and  data  

Stewart  McKie   -­‐  59  -­‐   As  of  15/08/2014  

actually  is.  This  formatting  ‘style’  tagging  is  only  used  to  apply  formatting  
rules  for  rendering  such  as  the  font,  spacing  or  alignment  of  the  content  
encompassed  by  the  style  tags.  However,  here  we  are  not  concerned  with  
presentational  style  tagging  but  with  the  “semantic  tagging”  of  the  
screenplay  content  to  help  understand  what  the  data  actually  means  in  its  
current  context.  

Consider  if  our  dialog  snippet  was  tagged  in  this  way:    

<character>ILSA</character>
<dialog>Play it, Sam</dialog>

In  this  example  the  notation  <character>  or  <dialog>  is  a  “start”  tag  to  
indicate  that  what  follows  is  a  piece  of  dialog  or  character  name  and  the  
notation  </character>  or  </dialog>  is  an  “end”  tag  that  indicates  the  end  of  
the  extent  of  the  dialog  item.  It’s  easy  to  see  how  a  machine  that  understands  
how  to  parse  this  kind  of  ‘tagged’  data  can  find  ‘character’  data  or  ‘dialog’  
data  simply  by  looking  for  text  surrounded  by  the  ‘character’  and  ‘dialog’  
tags.  

Traditionally,  screenwriting  applications  recognize  the  need  to  store  

screenplay  metadata  and  data  in  ways  that  facilitate  correct  formatting  of  
the  data  when  rendered.  But  often,  to  further  ‘structure’  the  text  to  better  
identify  metadata  or  data  for  analysis  purposes,  this  relied  on  manual  
tagging  by  the  writer.    

In  order  to  prepare  a  script  for  production,  some  screenwriting  software  

packages  allow  words  within  the  text  of  a  script  to  be  ‘marked-­‐up’  or  
‘tagged’  by  the  writer  or  some  other  editor  of  the  script  with  production-­‐
related  metadata  tags.  For  example,  Final  Draft  provides  a  means  for  the  
script  author  to  add  metadata  to  a  script  by  manually  ‘tagging’  text  in  the  
script  to  identify  it  –  say  to  identify  a  specific  noun  in  the  script  as  a  prop.  
For  example,  references  to  sound,  props  or  special  effects  included  in  the  
scene  action  description  can  be  tagged  as  such:  

Stewart  McKie   -­‐  60  -­‐   As  of  15/08/2014  

Suddenly, the bomb <sound>EXPLODED</sound>, triggering
mayhem in the crowd.

Bond whipped his <prop>WALTHER PPK</prop> from its

holster.

The cop <sfx>MORPHS</sfx> into a molten metal

humanoid.

The  traditional  formatting  convention  of  capitalizing  these  production-­‐

related  items  in  a  printed-­‐paper  script  is  a  simple  way  to  help  production  
managers  to  visually  identify  what  was  relevant  to  them  in  the  screenplay  to  
help  them  to  plan  and  setup  scenes/shots.  Whereas  manually  tagging  the  
script  text  in  this  way  to  ‘bake-­‐in’  metadata  identifiers  enables  the  
screenplay  software  to  automatically  extract  specific  production  reports  
from  the  electronic  copy  of  the  script  and  to  report  these  references  in  a  
meaningful  way.  The  software  simply  searches  for  text  within  a  specified  tag  
set  to  find  say,  all  the  sounds,  props  or  special  effects.  

This  kind  of  tagging  can  also  have  a  commercial  benefit.    When  offered  
‘action’  text  tagged  like  this:  

<action><text>Bond  checked  his  watch:  O500  hours.</text></action>  

A  software  program  could  specifically  read  only  the  action  text  in  the  script  
to  find  product  placement  opportunities  for  brand  sponsorship  to  pitch  to  a  
company  like  Rolex  by  identifying  the  noun  ‘watch’.  And  by  identifying  
‘checked’  as  a  verb,  the  program  could  also  start  to  build  a  profile  of  what  
the  character  Bond  does  in  order  automatically  build  up  a  character  profile  
from  the  evidence  of  the  data  rather  than  by  the  writer  describing  his  
character  manually  in  the  form  of  a  hand-­‐crafted  ‘bio’.  Does  Bond  do  lots  of  
checking?  Maybe  Bond  shows  signs  of  being  a  meticulous  character?    This  is  
an  example  of  how  quantitative  analysis  of  the  script  content  may  assist  with  
qualitative  assessment  of  a  role.  Now,  evidence  from  the  data  helps  both  to  

Stewart  McKie   -­‐  61  -­‐   As  of  15/08/2014  

define  and  validate  his  actual  functional  role  in  the  screenplay  as  a  spy.  

Tagging  screenplay  data  in  this  fashion  is  only  one  way  of  ‘datafying’  the  
textual  content,  by  means  of  embedding  identification  tags  in  the  text.  
Tabulating  the  data  is  another  way  to  achieve  a  similar  purpose.  
Conceptually,  a  table  (in  practice  represented  by  say  a  database  table  or  a  
spreadsheet  file  for  example)  identifies  the  purpose  of  the  metadata  or  data  
by  defining  that  data  in  terms  of  a  row  column  intersection.  So  (in  simplistic  
terms),  our  data  snippet  might  be  represented  in  table  format  as:  

Table   Row  ID   Type   Character   Text  

Snippet   1   Dialog   ILSA   Play  it,  Sam.  

Table  1.3    -­‐  Row/Column  Intersection  

Using  a  database  management  system,  the  row/column  intersections  in  the  

‘snippet’  table  can  be  used  to  query,  for  example:  

• All  ILSA’s  text  that  is  of  type  dialog  

• Only  text  that  is  type  dialog  
• How  many  times  ILSA  speaks  any  dialog  
• All  dialog  that  mentions  ‘Sam’    
• All  dialog  spoken  by  ILSA  that  contains  an  action  verb  e.g.  ‘Play’  
• All  dialog  spoken  by  ILSA  that  refers  to  another  character  e.g.  ‘Sam’  

Now  the  metadata  and  data  has  been  formally  structured  for  identification  
purposes  so  that  any  program  that  understands  that  formal  structure  can  
query  both  in  various  combinations  to  deliver  information  to  the  user.    

Once  the  screenplay  is  represented  as  a  data-­‐centric  rather  than  document-­‐
centric  artefact  and  the  data  structured  so  that  programs  can  both  identify  
its  content  purpose  as  well  as  how  to  format  it  when  rendered,  both  data  

Stewart  McKie   -­‐  62  -­‐   As  of  15/08/2014  

and  metadata  assume  an  additional  role  above  and  beyond  merely  
describing  who  does  and  says  what,  where  and  when:  they  help  to  prepare  
the  script  content  for  programmatic  analysis:  Metadata  can  be  analyzed  on  
its  own,  data  can  be  analyzed  on  its  own  and  the  combination  of  data  plus  
metadata  can  also  be  analyzed.  

A  key  aim  of  screenwriting  2.0  applications  that  assume  screenplays  as  data  
is  to  facilitate  automatically  recognizing  semantic  concepts  already  
embedded  in  the  screenplay  content  so  the  author  does  not  have  the  burden  
of  this  additional  manual  process.    So  if  the  application  can  identify  nouns  
used  in  dialog  or  action  text  because  it  can  identify  such  text  via  metadata,  
the  application  can  then  extract  a  list  of  all  nouns  avoiding  the  need  for  the  
writer  to  tag  them.  This  is  useful  because  this  part  of  speech  is  most  likely  to  
represent  the  props  used  in  the  movie’s  production.  Counting  the  frequency  
of  each  noun  then  indicates  which  props  will  get  the  most  use  and  may  need  
to  be  more  carefully  selected.  

According  to  some  sources  (see  Demystifying  Big  Data  p.10)  only  15%  of  
data  is  structured  i.e.  stored  in  databases  and  spreadsheets  and  85%  of  data  
is  unstructured  in  the  form  of  emails,  videos,  blogs  etc.  (even  though  much  of  
this  data  is  actually  also  stored  in  databases).  But  here  what  we  mean  by  
‘structured’  data  is  data  that  has  both  metadata  to  provide  context  and  data  
that  is  subject  a  mutually  agreed  standard  for  how  that  metadata  is  defined  
and  applied  to  the  data  What  is  not  meant  here  is  the  typical  meaning  of  the  
screenwriting  term  ‘structure’  in  connection  with  screenplays  -­‐  where  what  
comes  to  mind  is  the  Aristotelian/Field  3-­‐act  structure  or  Vogler’s  (1992)  
Hero’s  Journey  or  Murdock’s  (1990)  Heroine’s  Journey.    

There  are  at  least  three  ways  of  storing  screenplay  data  in  a  structured  
format  to  make  them  machine-­‐comprehensible  and  more  suitable  for  
analytics:  As  data  stored  in  a  relational  database  management  system  
(RDBMS)  or  in  a  ‘NoSQL’  database  or  in  a  file  (or  database)  that  uses  
Extensible  Markup  Language  (XML)  to  ‘tag’  the  data  with  metadata.  I  refer  to  
these  as  ‘tabulated’,  ‘documented’  and  ‘tagged’  screenplays  below.    

Stewart  McKie   -­‐  63  -­‐   As  of  15/08/2014  

1.4  Tabulated  Screenplays  

Instead  of  storing  screenplay  data  and  metadata  in  an  ‘unstructured’  
document  file  (e.g.  .rtf  or  .pdf)  or  plain  text  file  (e.g.  .txt),  it  could  be  stored  in  
a  relational  database  management  system  (RDBMS)  that  uses  a  separate,  
formal  schema  to  define  how  the  data  is  stored  and  related  in  the  database  
and  a  formal  language  for  (among  other  things)  querying  that  data  for  
analysis  purposes  -­‐  Structured  Query  Language  (SQL).  This  approach  has  
been  used  before  for  a  similar  purpose,  for  example  by  Franzosi  (2010:  75-­‐
82),  to  store  and  analyze  what  he  calls  the  ‘story  grammar’  of  any  kind  of  
narrative  text  in  a  Microsoft  Access  application  called  PC-­‐ACE.      

Here,  the  schema  (or  entity-­‐relationship  model)  defines  (among  other  

things)  the  database  tables  used  to  represent  the  main  data  entities  of  the  
script  and  the  table  columns  used  to  store  both  data  and  metadata  within  
those  tables.  

Simplistically,  a  screenplay  could  be  represented  by  just  3  related  database  

tables  (see  figure  1.1):  

1. A  script  table  
2. A  scene  table  
3. A  ‘snippet’  table  (storing  action  and  dialog  data)  

The  script  table  has  one  row  per  script.  The  scene  table  has  one  row  per  
scene  within  a  script.  The  snippet  table  has  one  row  per  data  element  (e.g.  
action  or  dialog  block)  within  a  scene.    

These  tables  are  related  by  means  of  ‘foreign  keys’.  In  relational  terms,  if  the  
primary  key  of  the  script  table  is  script_id  then  this  is  used  as  a  foreign  key  
in  the  scene  table  to  link  one  or  more  scene  rows  to  a  script  row.  If  the  
primary  key  of  the  scene  table  is  scene_id  then  this  is  used  as  a  foreign  key  in  
the  snippet  table  to  link  one  or  more  snippet  rows  to  scene  rows.  

Stewart  McKie   -­‐  64  -­‐   As  of  15/08/2014  

  

Figure  1.2  -­‐  Script  as  tables  (simplified)  

So  a  single  screenplay  with  60  scenes  and  many  action  and  dialog  blocks  
within  those  scenes  is  likely  to  be  stored  as:  

• 1  row  in  the  script  table  

• 60  rows  in  the  scene  table  
• Hundreds  of  rows  in  the  snippet  table  

Most  of  the  ‘data’  stored  in  these  tables  is  in  fact  metadata.  The  snippet  table  
stores  both,  with  a  row  comprising  a  series  of  columns  in  which  data  is  
represented  by,  for  example,  the  dialog  text  of  the  snippet  e.g.  ‘Play  it,  Sam’  
whereas  metadata  is  represented  by,  for  example,  the  type  of  snippet  
represented  e.g.  ‘Dialog’  or  an  identifier  of  the  character  who  spoke  it.  

By  tabulating  the  screenplay  data  and  metadata  in  this  way,  the  screenplay  
is  stored  in  a  structured  format  that  includes  formally  defined  relationships  
between  data  tables.  To  find  out  information  about  a  script  e.g.  the  title  or  
author,  you  query  the  script  table.  To  find  out  information  about  a  scene,  e.g.  
the  location  or  time  of  day  you  query  the  scene  table  for  a  specific  script.  To  
find  out  information  about  the  dialog  or  action  in  a  script  scene,  you  query  
the  snippet  table.  To  get  a  copy  of  the  complete  script  document,  that  
requires  data  from  all  three  tables,  you  would  need  to  assemble  (join)  data  
sourced  from  each  table  based  the  relationships  between  them.  

All  my  software  prototypes  (described  below)  are  based  on  tabulating  the  
data  in  the  Open  Source  RDBMS  MySQL.  The  main  advantage  of  tabulating  

Stewart  McKie   -­‐  65  -­‐   As  of  15/08/2014  

screenplays  in  this  way  is  that  it  makes  deriving  analytics  from  the  data  
much  easier  as  there  are  many  tools  available  to  analyze  and  visualize  data  
stored  in  SQL  databases.    A  potential  disadvantage  is  that  all  scenes  and  the  
script  as  a  whole  have  to  be  assembled  ‘on-­‐demand’  from  the  scenes  and  the    
snippets  linked  to  each  scene.  Another  is  that  as  the  script  is  no  longer  
encapsulated  within  a  single  document,  but  decomposed  into  smaller  
content  units,  it  requires  more  formal  management  and  may  not  present  the  
traditional  free-­‐format  document  ‘feel’  to  users.    

Franzosi  (2010:  chapter  3)  also  makes  a  number  of  useful  points  about  using  
an  RDBMS  to  store  data  and  using  SQL  not  just  to  query  the  data  but  also  to  
update  the  data  to  better  reflect  the  analytical  needs  (2010:93-­‐95).  So  if,  for  
example,  you  wanted  to  identify  all  violent  scenes  in  a  screenplay  (without  
manually  doing  this  yourself)  you  could  add  a  column  to  your  scene  table  –  
say  scene_type  –  and  then  use  a  SQL  update  statement  to  find  examples  of  
terms  in  the  textual  data  of  the  scene  (i.e.  dialog  and  action  snippets)  that  
can  be  ‘mapped’  to  the  scene_type  attribute  ‘violent’.  For  example  the  terms    
‘killed’,  ’stabbed’,  ’shot’  can  be  identified  in  the  text  and  the  scene  flagged  as  
scene_type  ‘violent’.  A  similar  data  update  process  could  also  be  used  to  
identify  sexual  scenes  from  the  screenplay  text.  The  combination  of  these  2  
techniques  could  identify  scripts  likely  to  have  ‘adult’  content  without  the  
need  to  read  the  script.  

1.5  Documented  Screenplays  

Recently,  a  number  of  ‘NoSQL’  databases  have  emerged  that  store  and  
manage  data  differently  from  a  SQL  RDBMS  (notably  they  do  not  use  SQL  as  
the  language  for  managing  and  querying  the  database)  and  are  better  suited  
to  managing  specific  kinds  of  textual  content.  Vaish  (2013:11)  identifies  five  
types  of  NoSQL  database,  segregated  by  data  model:  

1.  Document  
2.  Key-­‐Value  
3.  XML  

Stewart  McKie   -­‐  66  -­‐   As  of  15/08/2014  

 4.  Column  
5.  Graph  

My  aim  here  is  not  to  compare  and  contrast  these  5  types  but  simply  to  say  
that  while  any  of  these  5  types  could  be  used  to  store  screenplay  data,  some  
are  more  likely  candidates  than  others  due  to  the  relatively  stable  structure  
of  a  screenplay  and  the  nature  of  its  content.  For  example,  NoSQL  graph  
databases  (like  Neo4J  or  FlockDB  for  example)  are  best  suited  for  storing  
only  that  data  from  a  screenplay  that  is  specifically  intended  to  produce  
social  network  graphs  (e.g.  to  visualize  the  character  relationships  in  a  
screenplay).  

Here,  as  a  contrast  to  the  SQL  RDBMS  or  ‘tabulated’  datafication  approach  
outlined  above,  I  will  outline  the  use  of  a  NoSQL  Document  database  (like  
Couchbase  –  see  http://www.couchbase.com/docs/couchbase-­‐devguide-­‐
2.0/modeling-­‐documents.html  -­‐  accessed  10  June  2013)  to  store  screenplay  
data.  
 
Document-­‐databases  have  become  popular  as  a  data  store  for  web  
applications  such  as  social  networks  where  the  data  being  managed  is  pre-­‐
dominantly  high  volumes  of  textual  (vs.  say  numeric)  content  such  as  
comments  or  chat  or  tweet-­‐like  data.    Instead  of  requiring  a  pre-­‐defined  
schema  to  structure  the  data  into  tables  comprising  a  set  of  columns  with  
data  represented  and  accessed  as  content  rows,  document  databases  store  
data  in  document  entities  with  the  data  stored  in  value  pairs.  So  separate  
documents  could  represent  a  script,  a  scene  and  a  snippet  (see  figure  1.3).  

Stewart  McKie   -­‐  67  -­‐   As  of  15/08/2014  

Figure  1.3  -­‐  Script  as  documents  (simplified)    

Here  the  document  is  like  a  table,  the  value  pairs  like  columns  and  the  
linking  relationship  between  documents  is  based  on  a  document  reference  
identifier  rather  than  primary  and  foreign  key  references  in  the  relational  
model.    

Document  databases  are  claimed  to  have  some  advantages  over  the  
relational  model  in  that  they  allow  easier  distribution  of  these  virtual  
documents  across  physical  servers,  require  less  schema  administration  
(because  the  document  itself  is  its  own  schema)  and  offer  more  efficient  data  
updating  if  data  changes  regularly  (as  it  is  likely  to  do  while  drafting  
versions  of  a  screenplay).  Documents  can  also  include  linked  data  ‘nested’  
within  a  single  document    -­‐  such  as  a  scene  document  that  also  contains  all  
the  comments  posted  about  that  scene  –  so  that  from  a  content  perspective  
the  document  is  fully  integrated  and  can  ‘stand  alone’  for  analysis  purposes.  
In  the  relational  model  this  would  require  two  related  tables:  one  for  scenes  
and  one  for  comments  linked  to  those  scenes.  A  disadvantage  of  these  
NoSQL  databases  over  SQL  databases  is  that  they  are  newer    -­‐  my  lack  of  
knowledge  of  these  databases  is  the  main  reason  why  I  did  not  use  this  kind  
of  data  store  for  my  practical  deliverable  Scenepad  –  and  they  are  not  as  
accessible  from  as  wide  a  range  of  analytic  and  visualization  tools  as  SQL  
databases.  However  it  seems  highly  likely  that  NoSQL  databases  will  be  used  
by  future  screenwriting  2.0  applications.  

1.6  Tagged  Screenplays  

Instead  of  storing  screenplay  data  and  metadata  in  a  database,  the  data  
could  be  stored  in  an  Extensible  Markup  Language  (XML)  file.  As  mentioned  
above,  this  is  the  basis  for  the  file  format  now  used  by  Final  Draft,  so  this  
type  of  data  format  is  already  in  widespread  use.  Note  that  Final  Draft  stores  
the  XML-­‐tagged  screenplay  data  in  a  discrete,  named  file  and  not  in  a  
database  designed  to  store  XML-­‐tagged  data  where  both  the  data  and  the  
tags  are  stored  in  the  database  itself.  

Stewart  McKie   -­‐  68  -­‐   As  of  15/08/2014  

However,  XML  databases  can  also  be  used  to  store  XML-­‐tagged  screenplay  
data  in  a  database  management  system  rather  than  in  a  file.  In  this  case,  each  
individual  ‘instance’  file  is  fed  into  the  database  and  the  tagged  content  
‘shredded’  into  the  more  granular  data  storage  structure  of  the  database.  An  
XML  database  would  be  the  ideal  repository  for  managing  a  corpus  of  
multiple  Final  Draft  .fdx  files  because  the  power  of  the  database  query  tools  
can  be  brought  to  bear  on  the  corpus  as  a  whole  rather  than  having  to  work  
with  each  screenplay  file-­‐by-­‐file  as  is  the  case  when  each  screenplay  is  
‘stored’  individually  in  its  own  .fdx  file.    Creating  such  a  corpus  of  .fdx  
screenplays  in  an  XML  database  is  a  significant  research  and  commercial  
opportunity.  

When  screenplay  data  is  stored  in  XML  it  functions  as  an  XML  ‘instance’  
document  that  is  part  of  a  related  set  of  documents  that  typically  includes  
both  schema  (.xsd)  and  stylesheet  (.xsl)  files  (and  may  include  others)  used  
for  defining  the  content  and  presentation  of  the  instance  document  as  a  
specific  kind  of  artefact  within  a  specific  information  domain  (in  this  case  a  
screenplay  artefact  within  the  screenwriting  domain).  The  tags  used  to  
markup  the  screenplay  metadata  and  data  content,  are  defined  in  the  
schema  and  their  rendering/presentation  format  described  in  the  stylesheet.  
So  if  a  Casablanca  screenplay  were  stored  in  this  way  it  would  probably  
consist  of  at  least  3  files:  

• casablanca.xml  –  the  data/metadata  instance  document  

• casablanca.xsd  –  the  XML  ‘reference’  schema  document  
• casablanca.xsl  –  the  stylesheet  document  

The  screenwriting  application  would  include  a  ‘parser’  function  that  is  able  
to  read  and  format  the  content  of  the  Casablanca.xml  file  with  reference  to  
the  xsd  and  xsl  files.  Note  that  a  Final  Draft  .fdx  file  combines  all  the  
screenplay  data  and  metadata  and  XML  tags,  including  the  relevant  
presentational  attributes,  into  a  single  file  and  it  is  the  Final  Draft  
application  itself  that  interprets  and  presents  this  XML  file  to  the  user.  

Stewart  McKie   -­‐  69  -­‐   As  of  15/08/2014  

XML  is  not  only  useful  for  providing  a  means  to  structure  the  data,  but  also  
has  a  potential  standardization  purpose  that  could  benefit  the  screenwriting  
fraternity  in  other  ways.  For  example,  if  a  ‘standard’  XML  format  can  be  
agreed  for  describing  screenplays,  in  the  form  of  an  agreed  schema  and  
presentational  stylesheet  (as  has  happened  in  other  business  domains  –  see  
section  1.6  below),  then  it  will  becomes  much  easier  both  to  export/import  a  
homogenous  screenplay  format  between  heterogeneous  screenwriting  
applications  and  to  apply  similar  analyses  and  visualizations  from  a  single  
screenplay  analytics  application  across  content  sourced  from  multiple  
screenwriting  applications.          

A  disadvantage  of  screenplay  tagging  is  that  the  addition  of  markup  data  to  
the  script  will  significantly  increase  the  size  of  the  screenplay  file  reflecting  
the  addition  of  start  and  end  markup  tags  to  the  text.  But  today’s  cheap  
storage  and  processing  power  makes  this  irrelevant  and  this  does  not  matter  
because  the  marked  up  screenplay  file  is  not  a  text  for  reading  by  humans  
(although  some  may  find  this  an  interesting  pursuit),  but  for  processing  by  a  
software  program.    

XML  is  a  specific  way  to  tag  data  and  metadata  but  ‘tagging’  can  be  done  in  
other  ways.  For  example  Franzosi  (2010:62-­‐66)  describes  a  user  manually  
‘assigning’  ‘codes’  to  text  using  the  Computer-­‐Assisted  Qualitative  Data  
Analysis  Software  (CAQDAS)  ATLAS  to  add  the  metadata  attributes.  He  
describes  assigning  the  code  ‘Participant:Actor:Fascists’    to  the  term  
‘fascists’  in  a  text  to  in  order  to  identify  the  subject  of  a  subject-­‐verb-­‐object  
triplet  to  facilitate  further  qualitative  textual  analysis.  Once  all  text  is  
assigned  codes  manually  in  this  way  (see  ibid:  figure  3.20)  it  becomes  
relatively  simple  to  query  the  database  to  retrieve  examples  or  a  count  of  all  
‘actors’  in  the  text  (ibid:66).    

XML  tagging  can  also  be  used  to  model  a  hierarchy,  by  encapsulating  one  set  
of  tags  within  another.  An  example  of  a  screenplay  schema  model  that  
reflects  a  construct  of  XML  element  tags  and  attributes  is  shown  in  figure  1.4  
below.  The  <script>  tags  encapsulate  the  <sequence>  and  <scene>  tags  and  

Stewart  McKie   -­‐  70  -­‐   As  of  15/08/2014  

the  <scene>  tags  encapsulate  <action>  and  <dialog>  tags.  The  term  “Text”  
refers  to  the  actual  text  of  a  specific  dialog  or  action  block  instance  in  the  
script.  

Figure  1.4  -­‐  An  Example  of  a  Screenplay  Hierarchy  Model    

An  XML  tag  (or  element)  may  also  be  associated  with  and  qualified  by  
metadata  of  its  own,  so-­‐called  “attributes”  (not  included  in  figure  1.4)  used  
to  uniquely  identify  a  specific  instance  of  tagged  data  in  some  way,  so  for  
example:  

• The  attributes  of  the  <script>  tag  may  be:  Title,  author  and  date.  

• The  attributes  of  the  <sequence>  tag  may  be:  Title  and  order  (i.e.  
each  sequence  has  its  own  title  and  running  order  in  the  script).  

• The  attributes  of  the  <scene>  tag  may  be:  INT/EXT,  location,  
DAY/NIGHT,  transition  instruction  and  order  (i.e.  each  scene  has  
its  own  heading/slugline  and  running  order  in  the  script  (or  
sequence)  and  potentially  a  scene-­‐end  transition  instruction).  

• The  attributes  of  the  <dialog>  tag  may  be:  Character,  

parenthetical,  transition  instruction  and  order  (i.e.  each  dialog  
block  is  spoken  by  a  character  in  a  particular  way  and  in  a  

Stewart  McKie   -­‐  71  -­‐   As  of  15/08/2014  

 specific  order  within  the  scene  with  potentially  its  own  transition  
instruction).  

• The  attributes  of  the  <action>  tag  may  be:  Transition  instruction  
and  order  (i.e.  each  action  block  has  a  specific  running  order  in  
the  scene  and  potentially  its  own  transition  instruction).  

Tagging  may  be  an  appropriate  structured  data  format  for  screenplays  since  
it  has  already  been  used  to  bring  the  benefits  of  structured  data  and  
standardization  to  other  business  domains  focused  on  textual  content,  for  
example  corporate  regulatory  reporting.  

1.7  XBRL  and  Corporate  Reporting  

To  understand  the  potential  of  screenplay  as  data,  it  helps  to  compare  the  
screenplay  to  another  kind  of  document  content.  I  have  chosen  the  
mandatory  corporate  reporting  submitted  by  businesses  to  regulators  in  
business  jurisdictions  around  the  world.  This  kind  of  reporting  often  takes  
the  form  of  quarterly  or  annual  ‘returns’  such  as  the10-­‐Q  and  10-­‐K  reports  
submitted  by  registered  businesses  to  the  Securities  and  Exchange  
Commission  (S.E.C.)  in  the  USA  who  mandated  the  use  of  Extensible  
Business  Reporting  Language  (XBRL)  submissions  of  the  10-­‐Q  and  10-­‐K  
reports  by  corporate  filers  starting  in  2009.  XBRL  is  a  specific  
implementation  of  XML  used  to  ‘tag’  these  reports  so  that  most  of  the  data  
that  was  previously  reported  in  an  ‘unstructured’  way  (as  an  HTML  
document)  in  the  past  now  has  to  be  provided  as  both  an  unstructured  
HTML  document  file  and  a  structured  XBRL  data  file.  

Until  very  recently  most  businesses  filed  their  corporate  returns  to  
regulators  around  the  world  either  as  paper/PDF/HTML  documents  or  by  
filling  in  an  online  form  –  a  digital  representation  of  the  paper  document.  
Unfortunately  this  kind  of  delivery  of  corporate  data  is  not  conducive  to  the  
kind  of  analysis  that  government  and  other  regulators  need  to  make  of  
company  data  (e.g.  to  quickly  identify  trends  or  fraudulent  activity  or  simply  

Stewart  McKie   -­‐  72  -­‐   As  of  15/08/2014  

arithmetic  errors)  and  the  like-­‐for-­‐like  analysis  that  the  investor  community  
needs  to  make  decisions  about  whether  or  not  to  invest  in  a  company.  The  
reason  for  this  is  that  the  data  may  not  be  consistent,  is  subject  to  
transposition  errors  as  data  is  ‘taken  off’  documents  and  re-­‐keyed  into  the  
databases  or  spreadsheet  models  or  that  one  filer’s  revenue  or  expense  
calculations  are  not  made  on  the  same  basis  as  those  of  others  because  the  
data  submission  was  not  structured  and  standardized  in  a  mutually  agreed  
way  (i.e.  mutually  agreed  by  both  the  producers  and  consumers  of  the  data).  

Regulators  and  financial  services  investment  analysts  that  consume  the  data  
produced  by  corporates  in  their  returns  are  using  the  data  for  serious  
purposes  –  for  example  to  detect  financial  fraud  or  to  make  influential  equity  
‘buy’,  ‘hold’  or  ‘sell’  recommendations  to  the  brokers  and  individual  
investors  that  subscribe  to  their  services.  Some  of  the  benefits  of  XBRL  
submission  are  claimed  to  be  faster  processing  of  the  submissions  
themselves  and  of  analysis  of  the  submission  content  and  greater  reliability  
of  the  data  submitted  as  to  provide  the  kind  of  comparability  that  company  
regulators  and  investors  need  to  analyze  company  performance  efficiently  
and  effectively  requires  a  more  rigorous  data  delivery.  

This  is  why  many  regulators  around  the  world,  including  HMRC  in  the  UK  
and  the  S.E.C.  in  the  USA,  have  mandated  that  certain  quarterly  and  annual  
corporate  reporting  is  provided  in  a  specific  format  based  on  Extensible  
Business  Reporting  Language  (XBRL).  XBRL  is  used  to  ‘markup’  reporting  
data  according  to  a  specific  data  standard  in  the  form  of  a  mandated  XBRL  
schema  or  taxonomy  that  all  filers  must  use.  The  data  can  still  be  presented  
in  the  traditional  way  but  the  surface  presentation  now  has  the  potential  for  
analytical  depth  because  the  data  structure  is  more  rigorous,  based  on  a  
single  standard  being  used  to  define  data  concepts,  rules  and  relationships  
(e.g.  formulas)  as  reflected  in  the  data  markup.    

The  data  markup,  in  this  case  XBRL  tags,  is  subject  to  a  regulator-­‐mandated  
schema.  This  ensures  that  the  analysis  programs  used  by  both  regulators  
and  other  information  consumers,  can  expect  to  compare  like-­‐for-­‐like  tagged  

Stewart  McKie   -­‐  73  -­‐   As  of  15/08/2014  

data  to  look  for  exceptions,  to  identify  patterns  and  trends  or  to  create  
summary  aggregations  based  on  mutually  agreed  formulas  also  defined  
within  the  shared  and  mutually  agreed  XBRL  files  referenced  by  these  
programs.  

Appendix  A  outlines  a  very  simple  example  of  using  XBRL  as  provided  by  
Charles  Hoffman  (aka  the  ‘father’  of  XBRL)  for  learning  purposes  –  see  
http://xbrl.squarespace.com/journal/2008/12/18/hello-­‐world-­‐xbrl-­‐
example.html  -­‐  accessed  6  June  2013).  Note  that  all  data  content  is  enclosed  
within  <start>  and  </end  >  tags.    

In  this  case,  the  instance  document  HelloWorld.xml  relies  on  the  

HelloWorld.xsd  schema  document  for  its  data  processing  logic  so  the  schema  
acts  as  the  key  reference  for  any  program  that  is  reading  the  instance  
document.  In  practice  this  means  that  if  many  businesses  create  instance  
files  to  report  data  that  refer  to  the  same  schema  then  there  is  a  data  
consistency  and  standardization  in  play  that  makes  the  data  derived  from  
processing  the  instance  document  more  likely  to  be  reliable  and  
comparisons  more  accurate.  In  the  case  of  XBRL  reporting  to  the  S.E.C.,  the  
S.E.C.,  in  its  role  as  information  regulator,  ‘owns’,  maintains  and  enforces  the  
schema  that  all  corporate  information  providers  utilize.    

Corporate  reports  submitted  to  HMRC  or  the  S.E.C.  for  example  are  much  
more  complex  instance  files  and  subject  to  much  more  complex  schemas  
than  this  HelloWorld  example,  but  conceptually  they  function  in  the  same  
way.    

By  submitting  the  corporate  reporting  data  in  this  XBRL  format,  the  
traditional  return  document  has  been  datafied  so  that  each  filing  can  be  
delivered  as  an  electronic  upload  that  is  received,  processed  and  ‘first  pass’  
analyzed  entirely  by  machine.  In  the  case  of  S.E.C.  XBRL  submissions  one  
immediate  benefit  is  that  the  corporate  submission  data  is  also  available  
virtually  immediately  for  public  consumption  via  an  RSS  feed  that  is  also  
generated  automatically  from  the  submission  file,  programmatically.    

Stewart  McKie   -­‐  74  -­‐   As  of  15/08/2014  

However,  a  screenplay  is  not  a  corporate  report  and  there  is  no  mandate  to  
create  screenplays  using  a  specific  data  format,  nor  is  there  an  agreed  XBRL  
schema  for  screenplays  (although  there  might  be  one  day).  But  screenplays  
could  be  treated  in  the  same  way  given  that  agents,  movie  studios  and  
production  companies  that  receive  ‘spec’  scripts  for  evaluation  essentially  
play  a  similar  content-­‐receiver  role  to  a  corporate  regulator  and  it  could  be  
advantageous  for  all  such  receivers  to  receive  their  screenplay  data  in  the  
same  format.  But  this  is  likely  to  be  some  years  off  yet.  

As  yet,  unlike  the  corporate  reporting  domain,  there  is  no  standard  file  
format  or  XML  schema  for  sharing  screenplay  content  despite  the  efforts  of  
individuals  such  as  Sean  Moubry  (see  http://screenplayXML.org    –  accessed  
10  June  2013).  However,  with  the  transition  of  Final  Draft  from  the  
proprietary  .fdr  format  to  the  XML-­‐based  .fdx  format  (in  version  8  released  
April  2009)  and  Adobe  Story’s  story  XML  format,  there  is  now  real  industry  
momentum  towards  XML  as  the  file  format  of  choice.  For  example,  other  
screenwriting  applications  like  ScriptsPro  (for  the  iPad/iPhone)  have  
already  adopted  the  Final  Draft  .fdx  format  as  their  default  storage  format.  

Final  Draft  .FDX  files  use  ‘paragraph’  tags  to  identify  screenplay  
data/metadata  and  also  to  help  drive  formatting  of  the  data  when  presented  
on-­‐screen  or  on-­‐paper.  The  example  below  from  the  fdx  file  of  a  version  of  
the  script  of  Alien  (1979)  begins  with  a  <Content>  tag,  to  define  the  start  of  
the  script,  and  uses  the  <Paragraph  Type  =””>  to  identify  the  different  
screenplay  elements  (e.g.  Transition,  Scene  Heading  and  Action  below)  and  
to  drive  the  content  formatting.    

(This  example  also  shows  an  error  in  the  Final  Draft  file  import  logic  as  the  
caption  ‘SOMETIME  IN  THE  FUTURE’  has  been  incorrectly  defined  as  a  
transition  element  because  it  ended  in  a  colon,  which  the  import  business  
rules  assume  is  a  signifier  of  a  transition.)  

<Content>  
<Paragraph  Type="Transition">  

Stewart  McKie   -­‐  75  -­‐   As  of  15/08/2014  

  <Text>FADE  IN:</Text>  
</Paragraph>  
<Paragraph  Type="Transition">  
<Text>SOMETIME  IN  THE  FUTURE:</Text>  
</Paragraph>  
<Paragraph  Type="Scene  Heading">  
 <SceneProperties  Length="1/8"  Page="1"  Title=""/>  
<Text>INT.  ENGINE  ROOM</Text>  
</Paragraph>  
<Paragraph  Type="Action">  
<Text>Empty,  cavernous.</Text>  
</Paragraph>  

Other  options  gaining  some  attention  are  Kent  Tessman’s  Open  Screenplay  
Format  (http://www.kenttessman.com/2012/02/open-­‐screenplay-­‐format,  
accessed  28  August,  2012)  used  in  some  recently  released  screenwriting  
applications  (like  Tessman’s  own  Fade  In)  and  John  August’s  Fountain  
format  (http://www.fountain.io,  accessed  29  March,  2013).  These  file  
formats  are  designed  for  use  by  any  screenwriting  application  and  therefore  
less  tied  to  the  proprietary  internals  and  needs  of  specific  applications  like  
Final  Draft  or  Story.  

1.8  Analytic  Metadata  for  Screenwriters  

Production-­‐related  markup  of  the  screenplay  text  (e.g.  for  props  or  sounds)  
is  directed  at  a  specific  set  of  script  stakeholders,  namely  the  various  
production  managers.  It  is  not  intended  to  help  the  screenwriter  create  a  
higher  quality,  more  saleable  script.  Markup  to  help  the  writer  improve  their  
script  is  lacking  in  almost  all  screenwriting  software.  

For  example,  character  names  are  an  important  piece  of  metadata  in  a  script.  
They  are  linked  to  sections  of  dialog,  and  may  also  be  mentioned  both  in  the  
dialog  of  others  and  in  scene  action  descriptions.  But  to  get  any  analytical  
value  from  character-­‐related  data/metadata  in  the  script  means  that  it  must  

Stewart  McKie   -­‐  76  -­‐   As  of  15/08/2014  

be  tagged  everywhere  it  occurs  in  a  script.  In  the  screenplay  of  Casablanca,  
the  name  “RICK”  must  be  tagged  as  an  element-­‐type  “character”  everywhere  
it  is  found,  in  order  to  define  instances  of  RICK  in  the  script  content  as  
representing  the  character  Rick  (and  not  say  a  hayrick),  for  example:  

<character>RICK</character>
<dialog>I stick my neck out for nobody</dialog>

<action><character>RICK</character> moves out of the

shadows and into view.</action>

Alternatively,  if  these  dialog  and  action  snippets  are  stored  separately  as  
rows  in  a  database  table  then  the  database  row  itself  can  be  linked  
relationally  to  the  character  ‘Rick’  (itself  a  row  in  another  table)  so  we  know  
it  was  ‘character’  Rick  who  spoke  or  ‘character’  Rick  who  was  in  the  action.  
Either  way,  once  Rick  and  the  other  characters  in  the  script  have  been  
identified,  it  is  realistic  to  expect  a  program  capable  of  reading  both  the  
screenplay  data  and  metadata  to  determine  the  following:  

– what  dialog  Rick  speaks  

– who  Rick  speaks  to  
– when  Rick  is  referred  to  by  other  characters  
– which  scenes  Rick  plays  a  role  in    
– which  action  Rick  is  involved  in  or  referred  to  in  

As  a  minimum,  knowing  what  dialog  Rick  speaks  enables  dialog  analysis  to  
produce  some  kind  of  linguistic  fingerprint  for  Rick;  knowing  which  
characters  Rick  speaks  to  and  is  spoken  of  by,  clarifies  the  relationship  
network  that  Rick  participates  in  and  knowing  which  scenes  Rick  plays  a  
role  in  or  which  action  he  is  involved  in  helps  to  establish  both  his  narrative  
“throughline”  and  character  “arc”  in  the  screenplay.    

Now  imagine  that  Rick  is  identified  not  just  as  a  character  but  that  his  
character  is  also  separately  described  by  further,  enhancing  metadata  

Stewart  McKie   -­‐  77  -­‐   As  of  15/08/2014  

“attributes”  such  as  role  (e.g.  protagonist  or  antagonist),  gender,  race,  sexual  
preference  etc.  This  opens  up  even  more  analytic  potential  by  expanding  
“existing”  tags,  intrinsic  to  the  screenplay,  with  the  addition  of    “enhanced”  
tags  added  later  by  the  writer/analyst.  Clearly  generating  metadata  tags  
from  and  adding  metadata  tags  to  existing  script  content  is  an  important  
value-­‐add  process  in  terms  of  maximizing  the  analytic  and  visualization  
potential  of  the  screenplay  content.    This  is  the  kind  of  metadata  that  my  
Scenepad  prototype  allows.  

1.9  Screenplay  Elements  

The  XML/XBRL  concept  of  ‘elements’  (that  define  the  tags  used  to  identify  
data/metadata  in  content)  can  easily  be  applied  to  screenplays  as  a  means  of  
understanding  how  screenplay  content  could  be  datafied.    Final  Draft  uses  
the  term  ‘elements’  to  define  the  list  of  formatting  options  for  the  
screenplay’s  text  including:  General,  Scene  Heading,  Action,  Character,  
Parenthetical,  Dialogue,  Transition,  Shot  and  Cast  List.  You  type  in  the  text  
and  pre-­‐  or  post-­‐apply  the  element  ‘tag’  to  format  the  text  appropriately.  

For  analysis  purposes,  it’s  also  useful  for  the  content  of  a  screenplay  to  be  
viewed  as  consisting  of  an  assembly  of  individual,  but  related,  textual  
elements.  In  How  to  Be  Your  Own  Script  Doctor  Kenning  (2006:6)  states  that:    

Analyzing  a  screenplay  is  an  intricate  process  that  is  the  same  for  any  
document  –  it  entails  breaking  apart  the  individual  elements  and  
relationally  subjecting  those  elements  to  each  other  and  to  the  whole.  

Although  she  refers  mainly  to  higher-­‐level  (or  lower  granularity)  elements  
than  I  discuss  below,  it  is  important  to  recognize  that  a  screenplay  can  be  
approached  as  a  construction  of  a  series  of  inter-­‐related  elements.  

In  a  screenplay,  each  element  has  both  a  content  purpose  and  a  specific  

formatting  convention  to  enable  the  element  to  be  viewed  correctly  (i.e.  to  
industry  standards)  when  rendered  on-­‐screen  or  printed  on  paper.  In  the  
brief  discussion  of  screenplay  elements  that  follows,  standard  script-­‐
formatting  conventions  are  neither  discussed  nor  complied  with  in  the  

Stewart  McKie   -­‐  78  -­‐   As  of  15/08/2014  

examples,  only  the  purpose  of  the  element  is  outlined.  Some  of  the  examples  
below  are  based  on  a  version  of  the  script  for  Casablanca  (1942).  A  visual  
overview  of  various  typical  screenplay  elements  is  shown  in  figure  1.5  (see  
http://steveboese.squarespace.com/storage/sample-­‐screenplay-­‐page.gif      -­‐  
accessed  7  June  2013).  

Stewart  McKie   -­‐  79  -­‐   As  of  15/08/2014  

Figure  1.5  –  Typical  Screenplay  Elements  

Fade  In/Fade  Out  

These  terms  are  often  used  to  start  and  end  a  screenplay,  and  if  included  
they  act  as  markers  as  to  the  extent  of  the  script  content.  A  screenplay  file  
may  also  include  a  title  page  and  possibly  other  front  matter  content  relating  
to  what  might  be  seen  while  initial  the  movie  credits  are  showing,  but  the  
first  FADE  IN  usually  indicates  where  the  actual  screenplay  content  starts.  
Similarly,  FADE  OUT  or  END  would  indicate  where  the  screenplay  ends.  If  
these  extent  terms  are  missing,  then  it  can  be  assumed  that  the  screenplay  
starts  when  the  first  identifiable  scene  heading  occurs  and  ends  when  the  
last  identifiable  scene  finishes.  

The  Scene  

The  primary  building  block  of  a  screenplay  is  the  scene,  which  usually  
combines  both  dialog  and  action  taking  place  at  a  specific  internal  or  
external  location  during  a  specific  time  period  at  a  specific  time  of  day.  
Scenes  may  be  intentionally  grouped  into  contiguous  dramatic  units  
(sequences)  or  into  not  necessarily-­‐contiguous  dramatic  units  (storylines)  
and  may  be  also  be  deliberately  written  to  facilitate  the  decomposition  of  the  
scene  content  into  individual  “shot-­‐lists”  for  use  during  production.    

A  scene  typically  comprises:  

• The  scene  heading  

• One  or  more  blocks  of  scene  and/or  action  description  

• One  or  more  blocks  of  dialog  and/or  voiceover  narration  

• One  or  more  audio/visual  transition  instructions  occurring  either  

at  the  end  of  the  scene  or  within  the  scene  

Scenes  have  an  identifiable  beginning  in  the  script  content.  Conventionally,  

Stewart  McKie   -­‐  80  -­‐   As  of  15/08/2014  

all  scenes  should  begin  with  a  scene  heading  (starting  with  INT  or  EXT).  
Some  scenes  end  with  a  transition  instruction  (e.g.  CUT  TO).  Dialog  begins  
with  the  cue  or  name  of  the  character  speaking  the  dialog.  These  start  
signifiers  provide  obvious  entry  points  for  some  kind  of  programmatic  script  
content  analysis  method.    

Scene  Heading  (or  “Slugline”)  and  Captions  

A  screenplay  consists  of  a  series  of  scenes  that  may  or  may  not  be  in  
chronological  order  in  story  terms.  Each  scene  has  a  heading,  for  example  
INT.  RICK’S  PLACE  -­‐  NIGHT,  which  consists  of  at  least  three  sub-­‐elements:  

1. INT.  or  EXT.  -­‐  indicating  whether  the  scene  will  be  shot  internally  
(e.g.  in  a  room)  or  externally  (e.g.  on  a  street).  

2. LOCATION  -­‐  a  brief  description  of  where  the  scene  takes  place  (e.g.  
which  room  in  a  house  or  a  physical  place).  

3. DAY  or  NIGHT  –  a  time  of  day  indicating  whether  the  scene  takes  
place  sometime  during  the  day  or  sometime  during  the  night.  

Occasionally  a  scene  heading  may  be  extended  to  contain  extra  information  
such  as  the  nature  of  the  scene  and/or  a  specific  time  of  day,  the  season,  
present/past/future  setting  -­‐  for  example:  EXT.  THE  AFRICAN  SAVANNAH  -­‐  
DAY  –  PREHISTORIC  TIMES  (DAWN).  It  may  also  be  followed  by  a  caption  
that  establishes  some  exact  criteria  for  the  scene  location  and  time  that  adds  
essential  contextual  information  for  the  scene  e.g.  The  White  House  –  The  
Morning  of  9/11.  

Action/Scene  Description  

A  scene  usually  comprises  both  action  (or  scene  directions)  and  dialog  (or  
spoken  parts)  but  could  comprise  only  one  or  the  other.  An  action-­‐only  
scene  is  quite  common.  However,  a  dialog  scene  without  any  action  is  
unlikely,  since  some  kind  of  scene  description  is  usually  essential  to  

Stewart  McKie   -­‐  81  -­‐   As  of  15/08/2014  

establish  a  context  for  the  dialog  to  take  place.  

Action  descriptions  may  or  may  not  involve  naming  any  of  the  characters  
speaking  or  being  spoken  to  in  the  scene  but  will  include  the  mention  of  
props,  special  effects  (SFX),  specific  sounds  and  other  production-­‐related  
items.  These  character  names  and  items  may  be  capitalized  in  the  script  as  
this  used  to  be  a  way  of  highlighting  them  within  the  textual  content  of  a  
printed  document  for  production  planning  purposes.  Today,  if  scripts  are  
viewed  online,  there  is  less  need  to  capitalize  since  these  keywords  can  be  
color-­‐coded  based  on  user-­‐defined  tagging.  

Character  and  Dialog  

If  there  is  dialog  in  the  scene,  it  is  spoken  by  a  character  (even  if  that  
character  is  invisible  to  the  camera  e.g.  a  narrator  providing  a  voiceover),  for  
example:  

ILSA
Play it, Sam.

The  character  would  usually  speak  "on  camera"  (i.e.  within  the  frame  of  the  
shot)  but  may  speak  off-­‐camera  (O.C.),  off-­‐screen  (O.S.)  or  as  a  voice-­‐over  
(V.O.).  Using  a  parenthetical,  the  screenwriter  may  also  provide  a  specific  
suggestion  to  the  actor  in  relation  to  the  dialog  that  follows,  for  example:  

RICK
(indifferently)
If I gave you any thought, I probably would.

The  first  time  a  significant  character  is  mentioned  in  the  script,  their  name  is  
capitalized  –  traditionally  an  indication  to  the  actor  of  when  their  part  
“starts”  in  the  script.  Again,  today,  this  convention  is  hardly  necessary  as  it  
easy  to  find  the  first  mention  of  a  character  in  an  online  script  through  a  
search  function.  

Stewart  McKie   -­‐  82  -­‐   As  of  15/08/2014  

Transitions  

A  transition  (e.g.  CUT  TO:  or  DISSOLVE  TO:)  moves  the  focus  of  the  
audience’s  attention  to  create  a  specific  effect  within  the  visual  flow  of  the  
movie’s  narrative.  A  transition  can  occur  either  within  a  scene,  say  to  cut  to  
and  from  different  locations  to  set  the  context  for  a  telephone  conversation,  
or  at  the  end  of  a  scene  to  move  to  the  next  scene  in  a  specific  visual  way.    

Transitions  affect  both  the  look  and  feel  and  the  pacing  of  a  movie.  However  
it  should  be  noted  that  transitions  written  into  the  script  by  the  screenwriter  
may  be  ignored  altogether  in  production  and  in  any  case  have  much  less  
impact  on  the  final  product  than  the  shot-­‐by-­‐shot  transitions  decided  upon  
by  the  movie’s  camera  director  during  production  or  the  editor,  post-­‐
production.    

Other  Elements  

A  screenplay,  created  in  screenwriting  software  that  supports  this  

capability,  could  also  include  structural  act  and/or  sequence  headings  
inserted  into  the  script  and  sequential  scene  numbers.  These  headings  are  
not  part  of  the  screenplay  content  as  such,  but  function  to  group  scenes  
together  for  script  organization  and  locate  scenes  for  script  navigation.  An  
act  is  a  collection  of  sequences  or  scenes  (if  sequences  are  not  used)  and  a  
sequence  is  a  collection  of  scenes  that  comprise  a  deliberate  unit  designed  to  
achieve  a  specific  dramatic  or  narrative  purpose.    

Where  dialog  is  broken  up  by  action  or  a  page-­‐break,  the  terms  MORE  or  
CONT’D  may  be  inserted  to  indicate  a  break  in  the  flow  of  the  dialog  content  
across  printed  pages.  

The  screenplay  may  also  include  specific  camera  directions  such  as  ANGLE  
ON  (i.e.  shot  angle)  or  CLOSE  UP  (i.e.  shot  framing),  although  this  is  less  
common  today  than  it  used  to  be.  And  although  more  common  in  TV  scripts,  
a  storyline  identifier  is  also  another  way  of  grouping  scenes,  In  this  case,  
scenes  that  belong  to  a  specific  narrative  thread  in  the  screenplay.    

Stewart  McKie   -­‐  83  -­‐   As  of  15/08/2014  

A  screenplay  is  a  dynamic  artefact  so  during  the  production  process,  
revisions  to  the  script  may  also  be  added  and  provided  on  different  coloured  
paper  sheets,  these  revision  pages  signifying  a  new  ‘version’  of  the  previous  
script  content.    

Screenplay  Text    

Arguably,  the  lowest  level  of  meaningful  granularity  in  a  screenplay  is  the  
individual  word  (contained  in  action  description  or  dialog),  but  the  lowest  
level  of  element  granularity  is  the  blocks  of  dialog  or  action  description  in  
the  screenplay.  This  is  the  true  ‘text’  of  the  screenplay.  This  text  is  analogous  
to  what  Florian  Korsakow’s  Korsakow  system  (see  http://korsakow.org)  
calls  the  ‘smallest  narrative  unit’  (SNU)  -­‐  although  he  uses  it  to  refer  to  short  
video  clips  that  can  be  combined  in  his  ‘dynamic  storytelling’  system  to  
create  new  storylines.      

The  text  comprises  individual  words  and  the  syntactic  units  e.g.  sentences  
that  organize  the  words  into  units  of  meaning,  both  of  which  are  potential  
subjects  for  textual  analysis.  Analysis  of  words  and  word  frequency  may  an  
indicator  of  a  number  of  ‘traits’  of  the  script  as  a  whole:  time  period,  
location,  mood,  theme  etc.    Sentence  structure  and  words  grouped  into  
terms  may  also  be  indicative  of  time  period,  milieu,  culture,  pace  etc.  Either  
or  both  can  be  used  to  determine  role  or  scene  characteristics  such  as  
positive  or  negative  sentiment,  profane  or  sexual  content,  or  assist  with  
opportunities  for  product  placement.  

The  text  can  be  searched  for  keywords  and  event  notifiers  such  as  discourse  
markers  to  identify  events  within  a  text  that  may  then  indicate  something  
else,  such  as  some  kind  of  ‘change’  in  the  narrative  that  is  preceded  by  the  
event  marker.  

1.10  Screenplay  Decomposition  

Other  than  as  elements,  a  screenplay  can  be  decomposed  in  a  number  of  
ways,  including  as  a  hierarchy,  a  set  of  entities,  parts  of  speech,  and  a  series  

Stewart  McKie   -­‐  84  -­‐   As  of  15/08/2014  

of  content  strata.    

Hierarchy  Decomposition  

As  a  hierarchy,  a  screenplay  may  be  decomposed  as  follows:  

Acts  

             Sequences  

                                             Scenes  

                                                                 Action(s)  

                                                               Transition  

                                                               Character(s)                                                                                

                                                                     Parenthetical  

                                                                     Dialog  

                                                                 Transition  

                                             Transition  

This  hierarchy  provides  a  built-­‐in  "tree"  structure  for  navigating  the  script  
content  and  to  allow  content  viewers  to  easily  “jump-­‐to”  specific  content  
locations  in  the  script  to  read  or  edit  the  textual  content.  As  noted  above,  
XML  supports  the  “nesting”  of  tags  within  tags  and  so  XML  is  very  suitable  
for  representing  this  kind  of  built-­‐in  tree.  Most  screenwriting  software  
already  provides  rudimentary  hierarchy  navigation,  if  only  in  the  form  of  a  
list  of  scenes  that  lets  a  user  click  the  scene  name  in  the  list  to  view/edit  the  
full  scene  content.  

However,  a  screenplay  does  not  just  have  one  hierarchy  as  it  depends  on  the  
point-­‐of-­‐view,  which  in  turn  sets  a  ‘start  point’  for  the  hierarchy.  In  the  

Stewart  McKie   -­‐  85  -­‐   As  of  15/08/2014  

example  above  the  start  point  is  the  script.    But  if  the  start  point  is  a  specific  
role  or  location  then  the  hierarchy  will  be  different,  for  example  to  navigate  
only  the  dialog  snippets  linked  to  a  role  or  the  action  snippets  linked  to  a  
location.  

Entity  Decomposition  

Scripts  may  also  be  viewed  as  a  collection  of    "entities"  –  an  entity  being  an  
attribute  with  meaning.    A  ‘character’  is  an  entity  since  it  has  a  specific  
meaning  in  terms  of  its  role  in  the  script  narrative.    A  ‘location’  is  an  entity  
since  it  has  a  specific  meaning  in  terms  of  where  the  narrative  takes  place.  
An  entity  may  be  an  example  of  a  specific  implementation  or  “instance”  of  a  
certain  higher-­‐level  entity  “class”:  For  example,  characters  and  scene  
locations  as  examples  of  screenplay  entities  reflecting  screenplay-­‐specific  
implementations  of  the  general  entity  classes  of  "people"  and  "places".  
Blocks  of  action  can  also  be  considered  as  screenplay  specific  entities  
representative  of  a  general  entity  class  of  "events".  A  block  of  action  may  be  
further  decomposed  into  other  sub-­‐entities  such  as  place  names,  prop  
names,  sounds,  special  effects  (SFX)  and  so  on.  Transitions  are  also  entities,  
each  transition  being  a  specific  implementation  of  the  “visual  transition”  
class.  

Chatman  (1980:26)  produced  a  form  of  entity-­‐class  view  of  narrative  that  is  
relevant  here,  and  divided  it  into  two  main  branches:  Story  (Content)  and  
Discourse  (Expression).  For  our  purposes,  the  branch  “Story”  leads  to  the  
important  sub-­‐class  branches  of  Events  (action  and  happenings)  and  
Existents  (characters  and  settings  cf.  characters  and  locations  above).  

Stewart  McKie   -­‐  86  -­‐   As  of  15/08/2014  

  

Figure  1.6  -­‐  Chatman’s  Narrative  Entity  Classes  

Linguistic  Decomposition  

As  Chatman  indicates,  a  block  of  action  or  dialog  in  a  screenplay  also  
represents  a  form  of  expression  of  a  specific  kind  of  Discourse  entity.  Here  
the  form  may  be  further  decomposed  linguistically,  for  example  into  parts  of  
speech  (e.g.  verbs  and  adverbs,  nouns  and  adjectives)  and  significant  
keywords,  such  as  any  character  names  involved  in  the  action.  In  the  case  of  
dialog,  linguistic  decomposition  may  be  used  in  order  to  determine  some  
kind  of  tonal  fingerprint  (e.g.  emotional  level  or  the  use  of  sexual/profane  
language)  for  the  content  analyzed.  The  parenthetical  and  voiceover/off-­‐
camera  indicators  may  indicate  vocal  loudness,  auditory  focus  or  a  
character's  "distance"  from  the  scene  as  we  see  it  (e.g.  a  voiceover  may  imply  
that  the  character  is  "looking  back"  over  events  that  have  already  
happened).  Repetitive  use  of  certain  keywords  in  the  dialog  or  action  
throughout  the  script  –  for  example  the  use  of  the  word  ‘family’  in  the  
Godfather  (1972)  script  –  may  be  indicative  of  a  core  theme  in  the  
screenplay  (see  Scriptcloud  prototype  in  ‘Road  to  Scenepad’  below).    

Stewart  McKie   -­‐  87  -­‐   As  of  15/08/2014  

Strata  Decomposition  

But  this  kind  of  entity  decomposition  is  based  on  the  obvious  "surface"  
content  of  the  screenplay,  namely  the  words  themselves  and  their  structural  
organization  in  a  screenplay  format.  Another  kind  of  decomposition  seeks  to  
understand  what  might  be  termed  the  underlying  "strata"  of  the  screenplay  
that  may  or  may  not  be  explicitly  found  or  visible  in  the  textual  surface.    

In  Narrative  Fiction,  Rimmon-­‐Kenan  (2002:10)  expresses  it  this  way,  

‘Whereas  the  surface  structure  of  the  story  is  syntagmatic,  i.e.  governed  by  
temporal  and  causal  principles,  the  deep  structure  is  paradigmatic,  based  on  
static  logical  relations  among  the  elements…’    

Uncovering  these  logical  relations  to  reveal  the  “deep  structure”  is  the  
challenge  here.  Metaphorically,  the  words  and  sentences  represent  the  
visible  landscape,  whereas  the  strata  represent  the  invisible  geological  
layers  below.  These  strata  could  take  the  form  of:  

• Timelines    

• Journeys    

• Arcs  of  change  (i.e.  changes  in  emotional  or  psychological  states)  

• Tones  (e.g.  colours,  temperature)  

• Sound  and  Silence    

• Seasonal  Variations    

• Past,  Present  and  Future    

• Symbols  and  Metaphors    

• Plots  and  Sub  Plots    

• Narrative  conventions  

Stewart  McKie   -­‐  88  -­‐   As  of  15/08/2014  

 • Themes  

• Pacing  

Unlike  say  screenplay  structure,  which  represents  a  ‘fixed’  overlay  of  the  
content,  these  strata  are  more  fluid  and  more  likely  to  be  impacted  by  even  
small  changes  to  the  script  content.  There  may  also  be  certain  strata  
patterns  that  are  particularly  relevant  to  the  conventions  of  specific  movie  
genres  such  as  those  relating  to  death  and  violence  in  horror  movies  or  sex  
and  love  in  romantic  comedies.  Strata  analysis  and  visualization  is  likely  to  
demand  relatively  sophisticated  algorithms  to  detect,  decode  and  display  the  
strata  visually  to  the  writer  after  taking  the  script’s  proposed  genre  into  
account.    

Resource  Description  Framework  (RDF)  

According  to  Tauberer  in  What  is  RDF  (2006:1),  ‘Resource  Description  
Framework  (RDF)  is  the  W3C  standard  for  encoding  knowledge’  and  a  key  
part  of  the  future  semantic  web,  which  he  terms  ‘a  decentralized  platform  
for  distributed  knowledge’.    RDF  is  a  method  of  decomposing  knowledge  
into  individual  facts,  expressed  as  a  Subject-­‐Predicate-­‐Object  combination  
termed  a  “triple”.  Among  other  things,  these  facts  can  also  be  associated  with  
a  Uniform  Resource  Identifier  (URI)  i.e.  a  web  link  and  with  text  value  
attributes  (literal  values).  For  example,  a  URI  could  be  a  web  URL  that  
locates  the  triple  on  a  specific  web  page  and  a  literal  value  could  be  a  
descriptor  that  tells  us  something  more  about  the  triple,  e.g.  to  add  specific  
context  or  additional  content  to  the  triple.      

It  is  quite  practical  to  decompose  a  screenplay  into  a  large  number  of  triples  
that  would  summarize  a  number  of  facts  about  the  script,  for  example:  

<scriptname>is  written  by<author>  

<scriptname>belongs  to<genre>  

<scriptname>is  WGA  registered  as<registration  identifier>  

Stewart  McKie   -­‐  89  -­‐   As  of  15/08/2014  

<character>speaks<dialog>  
<character>speaks  to<character>  
<character>takes  part  in<action>  

<action>requires<special  effect>  
<action>involves<propname>  
 
<dialog>is<narrated>  
<dialog>is  spoken  in<language>  

<scene>is  located  at<location>  

<scene>takes  place  at<day/night>  
<scene>belongs  to<sequence>  

Decomposing  a  single  screenplay  into  RDF  doesn’t  necessarily  make  that  

much  sense,  but  if  a  corpus  of  screenplays  is  decomposed  in  this  way  and  
made  accessible  via  the  Internet,  this  corpus  of  screenplay  facts  in  RDF  
format  could  be  queried  in  ways  that  may  result  in  interesting  analyses  or  
visualizations.      

1.11  Screenplays  as  ‘Big  Data’  

It’s  difficult  to  discuss  screenplays  as  data  without  some  reference  to  the  
data  buzzword  ‘du  jour’,  namely  ‘big  data’.    On  its  own,  a  screenplay  hardly  
qualifies  as  ‘big  data’  –  the  term  usually  used  to  refer  to  the  high-­‐volume  and  
high-­‐velocity  datasets  that  governments,  retailers  and  banks  routinely  deal  
with  for  example.    And  a  single  screenplay  does  not  represent  anywhere  
near  the  kind  of  data  variety,  velocity  and  volume  that  big  data  algorithms  
are  designed  to  thrive  on  –  for  example  to  discover  data  patterns.    

Big  data  is  generally  created  by  large  numbers  of  people  or  devices  
‘transacting’  with  data  collection  systems  on  a  regular  basis.  In  this  sense,  
the  big  data  is  created  by  ‘pushing’  data  into  a  data  collection  system.  For  
example  when  each  bar  coded  product  that  you  buy  at  a  supermarket  is  
logged  into  a  centralized  point-­‐of-­‐sale  system  so  the  data  can  (among  other  

Stewart  McKie   -­‐  90  -­‐   As  of  15/08/2014  

things)  be  used  to  determine  optimum  inventory  levels  by  store  and  across  a  
chain  based  on  actual  and  predicated  spending  or  to  target  offers  to  
consumers  based  on  past  spending  habits.  

Big  data  can  also  be  collected  by  crowdsourcing  data  posted  by  users  to  
online  data  streams.  In  this  sense  the  big  data  is  created  by  ‘pulling’  data  into  
a  data  analysis  system.  For  example  when  Twitter  Tweets  or  Facebook  Likes  
are  used  to  perform  sentiment  analysis  to  look  for  trends  that  may  indicate  
increasing  or  reducing  interest  in  an  event,  person  or  product.  

Compared  to  these  ‘push’  and  ‘pull’  examples,  an  individual  screenplay  
represents  merely  a  tiny  sample  of  a  bigger  data  picture,  like  a  pixel  in  a  
digital  photo,  that  could  be  represented  by  a  corpus  of  screenplays  
representative  of  a  genre,  a  director’s  body  of  work  over  a  long  career  or  all  
scripts  from  a  movie-­‐making  decade  for  example.  That’s  why  the  initial  
interest  in  applying  ‘big  data’  techniques  to  the  world  of  movie  making  has  
not  focused  on  individual  screenplay  content  analysis  to  improve  script  
quality  but  in  activities  like  trying  to  predict  the  best  films  to  acquire  for  
optimum  online  rentals  or  which  movies  could  become  box-­‐office  
blockbusters  or  potential  Oscar  winners.    

The  New  York  Times  reported  in  Giving  Viewers  What  They  Want  (David  Carr  
–  24  February  2013)  that  online  movie  rental  giant  Netflix  analyzes  big  data  
sourced  from  their  customer  activity  to  refine  their  ability  to  decide  which  
shows  to  license  for  rental.  This  big  data  includes  some  30  million  movie  
‘plays’  per  day,  over  4  million  subscriber  ratings,  3  million  searches  plus  
other  data  relating  to  when  subscribers  watch  shows  and  how  they  pause,  
rewind  and  forward  shows.      

The  New  York  Times  also  reported  on  the  commercial  script  analysis  
activities  of  the  Worldwide  Motion  Pictures  Group  in  Solving  Equation  of  a  Hit  
Film  Script,  With  Data  (Brook  Barnes  –  5  May  2013).  Screenplay  analysis  of  
specific  scripts,  combined  with  other  data  from  focus  groups  and  online  
surveys  is  supposed  to  help  studios  that  are  ‘looking  for  clues  to  box  office  

Stewart  McKie   -­‐  91  -­‐   As  of  15/08/2014  

success.’    However  the  company  acknowledges  that  this  script  analysis  ‘is  
not  done  by  machines.’  So  this  does  not  really  qualify  as  a  true  data  analytics  
activity  as  it  relies  on  human  intuition  rather  than  machine  algorithms  to  
derive  interesting  correlations.    

Dr.  Peter  Gloor  at  the  Center  for  Collective  Intelligence  at  MIT’s  Sloan  School  
of  Management  accurately  predicted  the  2007  Best  Motion  Picture  of  the  
Year  Academy  Award  for  The  Departed  (2006)  by  tracking:  

…a  year’s  worth  of  Oscar  conversation  on  IMDB,  use  regression  analysis  on  its  
word  counts,  and  plug  that  into  a  custom-­‐designed  index  to  predict  Oscar  
nominees  and  winners.  

(see  http://www.vanityfair.com/online/oscars/2013/02/mathematical-­‐
formula-­‐oscar-­‐winners-­‐imdb,  accessed  08  March  2012).  

Prediction  markets  like  Intrade.com  (suspended  when  accessed  on  29  March  
2013)  and  farsiteforecast.com  are  among  a  number  of  websites  that  have  
also  engaged  in  Oscar  predictions  by  tapping  into  ‘the  wisdom  of  crowds’  via  
social  networking  sites.  Effective  crowdsourcing  of  big  data  in  this  way  not  
only  depends  on  a  lot  of  data  being  generated  by  and  publicly  available  from,  
the  crowd,  but  also  on  the  quality  of  the  crowd  generating  the  data,  as  
farsiteforecast.com  point  out,  

Essential  to  analyzing  the  success  and  value  of  crowd  sourcing  is  a  firm  grasp  
of  one  thing  –  the  crowd.    The  question  is  –  who  comprises  the  crowd  

(see  http://www.farsiteforecast.com/vanity-­‐fair-­‐polls-­‐and-­‐parties/,  
accessed  18  March  2013)  

and  emphasize  in  figure  1.7  below:  

Stewart  McKie   -­‐  92  -­‐   As  of  15/08/2014  

  

Figure  1.7  –  Quality  Pyramid  of  Crowdsourced  Data  

But  Oscar  predictions  like  this  using  social  networking  buzz  to  predict  
success  are  not  based  on  analyzing  screenplay  content  (predictive  analysis  
of  screenplay  content  is  discussed  below)  however  big  data  techniques  are  
being  used  as  a  means  of  comparing  individual  screenplay  content  to  
relevant  big  data  content  as  discussed  in  ‘dialog  analysis’  below.  

1.12  Screenplay  APIs  

Once  you  have  datafied  a  screenplay  it  becomes  practical  and  perhaps  useful  
to  permit  interaction  with  that  data  by  applications  other  than  the  
‘authoring’  application  (i.e.  the  ‘authoring’  screenwriting  tool).  This  is  
achieved  by  using  an  application  programming  interface  or  API  designed  to  
allow  other  applications  to  read  data  from  the  screenplay  or  write  data  to  
the  screenplay.      

A  key  reason  why  certain  social  networking  sites  like  Facebook  and  Twitter  
and  many  other  Web  2.0  sites  generally  have  grown  so  rapidly  and  managed  
to  build  an  application  ecosystem  around  what  may  be  a  very  simple  
concept  (e.g.  posting  a  140  character  Tweet)  is  by  providing  an  Application  

Stewart  McKie   -­‐  93  -­‐   As  of  15/08/2014  

Programming  Interface  (API)  to  allow  other  applications  to  interact  with  
them.  The  provision  of  an  API  enables  the  global  developer  community  to  
get  more  value  from  or  add  value  to  the  API  source  application.  

The  basis  of  any  API  can  be  summarized  in  two  words  ‘GET’  and  ‘PUT’.    A  
GET  retrieves  data  from  an  application  and  a  PUT  posts  data  to  an  
application.  So  a  GET  could  retrieve  all  Tweets  posted  to  a  user’s  account  
subject  to  specific  criteria  and  a  PUT  could  post  a  Tweet  to  that  account  from  
a  different  application  than  Twitter.  Obviously  APIs  restrict  what  data  can  be  
retrieved  from  an  application  and  what  data  can  be  posted  to  it  but  
essentially  they  open  up  applications  to  the  outside  world  through  what  is  
designed  to  be  a  well-­‐defined  and  secure  interface.  

A  screenplay  could  function  as  the  ‘object’  of  an  API  for  example  to  get  data  
from  the  screenplay  text  or  to  post  data  to  it.    But  this  is  only  realistic  if  the  
screenplay  is  datafied  e.g.  stored  in  an  XML  file  or  a  database.  If  a  corpus  of  
screenplays  were  stored  in  an  online  archive  in  a  standardized  data  format  
with  an  API  this  might  make  various  kinds  of  analysis  by  different  
communities  of  interest-­‐  genre  analysis,  ‘potential  blockbuster’  predictive  
analysis  and  sentiment  analysis,  not  to  mention  general  academic  research  –  
much  easier  to  do.  

A  screenplay  API  might  assist  when  writing  is  being  done  collaboratively  
with  a  partner  or  group  or  as  a  way  for  production  managers  to  interact  
with  a  script  during  production  of  the  movie.  Access  to  an  API  is  also  a  way  
to  facilitate  interaction  with  a  screenplay  as  a  data  ‘source/target’  from  
devices  like  mobile  phones,  where  perhaps  using  the  main  screenwriting  
application  does  not  make  sense.  So  if  you  want  to  post  content  to  the  
screenplay,  captured  on  the  device  (e.g.  a  photo  taken  on  location  or  an  idea  
or  note  scribbled  on  the  spur  of  the  moment)  then  an  API  would  help  to  
enable  you  to  post  this  data  from  your  phone  app  to  the  screenplay  data  
repository.  

Stewart  McKie   -­‐  94  -­‐   As  of  15/08/2014  

1.13  Screenplay  as  Data  Wall  or  Stream  

Online  social  networks  Facebook  and  Twitter  have  popularized  the  concept  
of  data-­‐walls  used  to  display  data-­‐streams.  Users  ‘poke’  data  onto  their  
personal  Facebook  wall,  which  may  then  be  pushed  on  to  the  walls  of  their  
friends.  Users  ‘tweet’  data  into  their  personal  Twitter  stream,  which  may  
then  be  ‘retweeted’  into  the  streams  of  their  followers.  And  there  is  no  
reason  why  a  screenplay  cannot  be  viewed  in  the  same  way  both  in  terms  of  
its  core  content  and  enrichment  content.  

Here,  for  example,  core  content  such  as  scenes  and  dialog/action  data  could  
be  viewed  as  being  ‘poked’  onto  the  screenplay  as  data  wall.  And  enrichment  
content  such  as  comments  about  a  scene  could  be  viewed  as  being  ‘tweeted’  
to  the  scenes  data  stream.  Online  social  applications  also  tend  to  make  user  
activity  more  transparent,  by  displaying  all  user  interaction  in  near  real-­‐time  
in  the  form  of  an  activity  stream,  and  again  this  is  equally  applicable  to  a  
screenplay  in  order  to  make  more  transparent  and  track  the  contribution  
made  to  the  dynamic  screenplay  artefact  by  all  the  stakeholders  in  the  
community  of  interest  around  the  script.    

Indeed  this  activity  stream  tracking  could  even  have  a  commercial  

dimension  in  terms  of  calculating  whether  or  not  a  specific  stakeholder  can  
claim  a  writing  credit  on  the  script  based  on  their  content  contribution,  if  the  
script  is  in  fact  made  into  a  movie.  

1.14  Screenplay  as  Interface  

Again,  the  datafication  of  the  screenplay  changes  the  way  that  an  individual  
user  may  interface  with  the  screenplay.  The  prevailing  interface  paradigm  is  
to  present  the  screenplay  through  the  interface  of  ‘script’  –  the  vertical  
scrolling  ‘toilet  paper’  paradigm  of  blog  posts  and  comments  –  or  through  
the  interface  of  scene    -­‐  a  sheet  or  two  off  the  roll.  But  with  screenplay  as  
data  there  is  no  longer  a  restriction  on  the  way  a  user  interfaces  with  a  
screenplay.  

Stewart  McKie   -­‐  95  -­‐   As  of  15/08/2014  

For  example:  

• A  writer  may  want  to  interface  with  the  script  content  by  working  
only  with  scenes  that  have  a  status  of  ‘draft’  versus  ‘final’;  or  with  
scenes  that  are  specifically  shared  with  her  by  a  partner;  or  with  
scenes  that  have  recently  attracted  the  most  comments  from  
others  with  access  to  the  script.  

• An  actor  may  want  to  interface  with  the  script  by  working  only  
with  scenes  that  she  is  in  and  then  maybe  toggling  the  scene  
content  so  that  only  her  dialog  is  shown;  or  viewing  all  her  dialog  
only  across  the  whole  script.  

This  is  much  easier  to  do  if  the  screenplay  content  is  stored  as  data.  
Suddenly  the  traditional  whole-­‐script  or  even  scene-­‐based  content  ‘straight-­‐
jacket’  can  be  discarded  and  users  allowed  to  interact  with  the  screenplay  
content  through  an  interface  that  reflects  their  role  or  what  it  is  they  actually  
want  to  use  the  content  for.  The  screenplay  as  data  enables  the  user  to  
choose  or  personalize  their  interface  with  the  content.    

Summary  

It  is  clear  that  if  we  consider  screenplays  as  data-­‐centric  rather  than  
document  centric  artefacts  that  comprise  both  data  and  metadata,  we  get  a  
different  perspective  of  screenplay  content  and  how  it  can  be  organized  and  
analyzed.  Plain  text  or  and  other  unstructured  data  storage  formats  do  not  
provide  the  same  rich  potential  for  analysis  as  screenplays  stored  in  
tabulated  (SQL  database),  documented  (document  database)  or  tagged  
(XML)  structured  data  storage  formats.  Datafied  screenplays  are  an  essential  
foundation  for  ‘what  happens  next’.  

Screenplays  stored  as  data  are  also  essential  to  enable  the  application  of  big  
data  techniques  to  analyze  a  corpus  of  screenplays  and  to  ‘open  up’  
screenplays  so  that  a  range  of  different  applications  can  interact  with  the  
content  via  a  screenplay  API.  Although  there  is  not  yet  a  standard  screenplay  
Stewart  McKie   -­‐  96  -­‐   As  of  15/08/2014  
format  used  by  all  screenplay  applications,  as  say  XBRL  is  used  for  
mandatory  corporate  reporting  in  many  countries,  the  Final  Draft  .fdx  
format  is  a  proto-­‐standard  that  could  be  adopted  industry  wide  by  
screenwriting  applications  in  the  future.  

In  chapter  2,  I  discuss  how  screenplay  as  data  provides  the  foundation  for  
screenplay  analytics  –  the  analysis  and  visualization  of  screenplay  content    -­‐  
helping  readers  to  better  understand  a  script  based  on  the  evidence  of  the  
data  and  helping  writers  to  ask  questions  of  their  script  by  surfacing  both  
‘knowns’  and  ‘unknowns’  from  the  evidence  of  the  screenplay  data.  

Stewart  McKie   -­‐  97  -­‐   As  of  15/08/2014  

2  Screenplay  Analytics  

Many  screenwriting  applications  do  not  focus  much  or  any  attention  on  
content  analytics,  thus  depriving  the  screenwriter  of  potentially  valuable  
tools  for  asking  more  questions  of  their  script  in  order  to  improve  its  quality  
and/or  commercial  potential.    

Here,  screenplay  analytics  means  applying  analytic  techniques  and  

algorithms  to  a  screenplay  text  –  both  metadata  and  data  -­‐  using  a  software  
tool.  Analytic  techniques  and  software  can  be  used  for  various  purposes  and  
come  in  various  forms.    Analysis  can  be  used  for  specific  purposes  such  as  
pattern  recognition  or  to  summarize  and  aggregate  data  so  it  can  be  
visualised.    The  insight  derived  from  this  analysis  can  be  used  for  decision  
support  to  facilitate  risk  analysis  or  fraud  detection  for  example.  Analysis  
may  be  delivered  in  the  form  of  row/column  reports,  graphs  and  charts  or  
visual  infographics,  whatever  makes  it  easier  to  assimilate  and  understand  
the  information  in  order  to  facilitate  decision-­‐making.  For  some  time,  in  the  
world  of  business  software  applications,  ‘analytics’  was  more  often  referred  
to  as  ‘decision  support’  as  this  is  often  what  it  is  used  for.    

In  the  domain  of  theatre,  academic  linguistic  analysis  of  the  works  of  
Shakespeare  –  both  as  individual  works  and  as  a  corpus  –  extends  back  at  
least  to  Caroline  Spurgeon’s  1935  work,  Shakespeare’s  Imagery  and  What  it  
Tells  Us  and  includes  studies  by  Mahood  (1957),  Carroll  (1967),  Elam  (1984)  
et  al.  as  well  as  Shakespearian  dictionaries  from  Schmidt  (1902)  to  Crystal  
(2002)  and  concordances  from  Beckett  (1787)  to  Spevack  (1968-­‐70).  
However  these  linguistic  analyses  have  focused  on  the  use  of  language,  the  
constituent  words  and  the  identification  of  thematic  patterns  within  a  very  
specific  content  domain  –  that  of  the  works  of  Shakespeare  -­‐  rather  than  on  
the  development  of  analytical  models  and  visualizations  that  can  be  applied  
to  these  and  other  non-­‐theatrical  dramatic  works  like  screenplays  with  
equal  utility.  

Stewart  McKie   -­‐  98  -­‐   As  of  15/08/2014  

My  focus  here  is  less  on  linguistic  analysis  of  the  language  of  an  individual  
screenplay  or  that  of  a  corpus  of  screenplays,  say  within  a  specific  genre,  but  
on  the  delivery  of  basic  screenplay  analysis  and  visualization  tools  to  help  
the  various  stakeholders  in  the  screenplay,  to  analyze  and  visualize  the  
content  of  a  script  in  various  ways,  including  its  use  of  language.  In  the  
context  of  single  feature  film  screenplay,  we  are  interested  in  analytics  that  
can  positively  influence  the  restructuring  or  rewriting  of  the  screenplay  
content  not  only  to  effect  a  quantitative  (i.e.  to  cut  or  add  content)  impact  on  
the  content  but  also  to  facilitate  a  qualitative  (i.e.  to  revise  content)  
improvement.    The  current  generation  of  screenwriting  software  does  a  
poor  job  of  helping  screenwriters  to  restructure  or  rewrite  their  scripts  
because  so  little  in  the  way  analytic  functionality  is  supported  

That  the  process  of  script  rewriting,  rather  than  script  writing,  is  so  poorly  
supported  in  the  current  generation  of  screenwriting  applications  is  
surprising  because  it  is  well  established  that  screenwriting  is  a  process  and  
that  screenplays  typically  go  through  any  number  of  rewrites  during  
development  and  while  they  are  being  shot  in  production.  According  to  Leily  
Kleinbard  in  The  Atlantic  (Nov.  21,  2012),  David  Magee,  the  screenwriter  of  
Ang  Lee’s  Life  of  Pi  (2012),  originally  considered  Yann  Martel’s  book  
‘unfilmable’,  which  may  be  why  adapting  the  book  into  a  screenplay  took  
‘170  Script  Revisions’.    Perhaps  some  kind  of  screenplay  content  analytics  
could  have  cut  the  number  of  revisions  needed  and  therefore  reduced  the  
script  development  cost?  

Screenplay  analytics  is  likely  to  be  of  most  benefit  not  just  to  writers  but  also  
to  readers  of  scripts,  particularly  the  story  analysts,  story  editors  and  
‘polishers’  employed  by  studios  or  production  companies  to  evaluate  or  
improve  the  ‘quality’  of  a  script.  This  is  because  analytics  is  a  way  to  learn  
more  about  a  script  prior  to  reading  it  or  to  confirm  or  deny  intuitive  
conclusions  based  on  the  evidence  of  the  data  after  it  is  read.    

According  to  Garfinkel  in  Screenplay  Story  Analysis  (2007:xv),  for  an  
experienced  screenplay  reader/analyst,    ‘it  takes  an  individual  about  one  

Stewart  McKie   -­‐  99  -­‐   As  of  15/08/2014  

and  half  hours  to  read  a  decent,  readable  screenplay’.    

However,  it  would  take  several  more  hours,  even  for  a  professional  script  
reader,  to  gain  a  reasonable  understanding  and  appreciation  of  the  quality  of  
the  script  and  to  evaluate  the  commercial  or  artistic  potential  of  the  
screenplay.  This  is  precious  time  for  production  company  executives  and  
studio  script  readers  who  apparently  have  a  never-­‐ending  “slush-­‐pile”  of  
speculative  scripts  scheduled  for  reading  and  appraisal  or  that  are  
submitted  online  to  various  script  ’harvesting’  web  sites  that  have  sprung  up  
to  encourage  and  manage  this  stream  of  ‘user  generated’  content.    

According  to  Edwards  and  Skerbelis  (2009:22-­‐25),  major  Hollywood  studios  

tend  to  have  8-­‐12  readers/analysts  on  staff  who  are  each  reading  8-­‐10  
scripts  a  week  to  produce  a  range  of  coverage  and  other  content  analysis  
reports  and  it  is  ‘the  story  editor’s  responsibility  to  evaluate  readers’  
coverage  on  a  regular  basis  and  to  maintain  the  quality  of  that  coverage.’  So  
anything  that  can  help  to  cut  the  time  to  do  the  evaluations  or  improve  the  
quality  of  the  results  either  by  analysis  of  an  individual  screenplay  or  as  a  
result  of  analysis  of  data-­‐based  results  over  time,  could  enable  studios  to  
read  more  scripts  (increase  throughput)  or  make  better  decisions  (improve  
quality).  

The  lack  of  attention  paid  to  screenplay  content  analytics  (as  opposed  to  say  
movie  analysis  and  Oscar  predictions)  by  both  theorists  and  practitioners  is  
also  surprising  because  analysis  and  visualization  of  the  dramatic  form  is  
not  a  new  concept.  Aristotle’s  Poetics  is  widely  regarded  as  the  earliest  work  
addressing  aspects  of  the  analysis  of  drama  and  this  work  even  includes  a  
reference  to  visualization  (1995:27),  as  Aristotle  recommends  that,    
 
“When  constructing  plots  and  working  them  out  complete  with  their  linguistic  
expression,  one  should  as  far  as  possible  visualize  what  is  happening.  By  
envisaging  things  very  vividly  in  this  way,  as  if  one  were  actually  present  at  the  
events  themselves,  one  can  find  out  what  is  appropriate  and  
inconsistencies  are  least  likely  to  be  overlooked.”  (my  emphasis).    

Stewart  McKie   -­‐  100  -­‐   As  of  15/08/2014  

Clearly  Aristotle  was  not  referring  to  the  analysis  and  visualization  of  
screenplay  content,  but  his  recommendation  is  relevant  here.  Probably  the  
first  published  example  combining  both  the  analysis  and  visualization  of  a  
dramatic  form  is  Freytag’s  pyramid  (figure  2.1),  from  his  Technique  of  the  
Drama  (1899:115  -­‐  originally  written  in  1863)  that  visualized  his  definition  
of  the  five  parts  of  the  drama  as  a  simple  pyramid  -­‐  introduction  (a),  rise  (b),  
climax  (c),  return  or  fall  (d)  and  catastrophe  (e).    

Figure  2.1  -­‐  Freytag’s  Pyramid  

Things  have  changed  a  lot  since  Freytag’s  day,  especially  in  the  last  few  
decades,  as  software  has  emerged  that  can  automate  many  content  analysis  
and  visualization  tasks  (although  we  should  be  clear  that  you  cannot  find  a  
screenwriting  tool  than  can  generate  a  Freytag’s  pyramid  for  you).  For  
example,  software  can  generate  visualizations  from  the  content  of  specific  
Shakespeare  plays,  such  as  the  social  network  of  Hamlet  (see  figure  2.2  
below  from  
http://services.alphaworks.ibm.com/manyeyes/view/SJjqGFsOtha6ymEZW
ajPF2,  accessed  23  September,  2008),  and  to  provide  a  completely  new  
interface  for  viewing  and  interacting  with  the  text  of  a  dramatic  work,  as  
exemplified  by  Watching  the  Script  (see  http://digitalplaybook.humviz.org,  
accessed  23  September  2008)  –  a  content  navigation  interface  shown  in  
figure  2.3  below.  The  software  used  for  both  of  these  examples  could  be  

Stewart  McKie   -­‐  101  -­‐   As  of  15/08/2014  

applied  equally  well  to  screenplay  analysis  and  visualization.  

Figure  2.2:  Character  Associations  in  Shakespeare’s  Hamlet  

 

Figure  2.3  -­‐  Watching  the  Script  

Stewart  McKie   -­‐  102  -­‐   As  of  15/08/2014  

At  the  heart  of  what  screenplay  analytics  is  about  are  Corrigan’s  (2007:22)  
comment  ‘Be  prepared  with  a  questioning  mind  from  the  beginning’  and  
Kenning’s  (2006:15)  ‘To  diagnose  your  script’s  problems,  you  will  ask  
questions  of  your  script  –  a  lot  of  questions’.  Also,  according  to  Card,  
Mackinlay  and  Shneiderman  (1996),  the  purpose  of  visualization  is  ‘to  aid  
analysis  and  facilitate  discovery,  decision-­‐making  and  explanation.’  So  my  
proposition  is  that  by  providing  the  screenplay  as  data  it  will  be  possible  to  
deliver  a  range  of  analytical  reports  and  visual  images  as  output,  to  enable  
any  screenplay  stakeholder  to  ask  more  questions  of  the  script  and  hopefully  
to  make  more  informed  decisions  about  it.    

The  intention  of  this  questioning  process  is  primarily  to  help  the  writing  
stakeholders  (i.e.  the  writer  and/or  co-­‐writers)  in  particular  to  better  
understand  where  to  focus  their  attention  in  order  to  raise  the  quality  of  
their  screenplay  and  perhaps  increase  the  salability  or  competitiveness  of  
their  script.  There  is  certainly  a  commercial  dimension  to  screenplay  
evaluation  that  is  relevant  to  screenplay  analytics  and  visualization.  In  
Writing  Drama  Lavandier  states  this  most  forcefully,  

I  am  convinced  that  the  health  of  the  film  (and  television  drama)  
industry  hinges  on  the  ability  of  all  those  involved  in  decision-­‐making  
to  read,  that  is  to  evaluate,  a  screenplay  correctly.  (my  emphasis).  
(2004:  26)  

Here  it  is  important  to  clarify  that  by  visualization,  we  are  primarily  
concerned  with  visualizing  the  screenplay  content  (input)  rather  than  
visualizing  the  possible  filmic  (output)  from  the  screenplay.  Arguably,  a  
screenplay  is  itself  a  form  of  narrative  visualization  since  it  enforces  a  
specific  presentation  ‘form’  on  the  content.  Yet  the  screenplay  is  not  one  of  
the  7  genres  of  narrative  visualization  proposed  by  Segel  and  Heer  (2010),  
presumably  because  it  is  not  ‘visual’  enough  compared  to  their  other  genre  
selections,  despite  the  fact  that  even  the  traditional  paper-­‐based  screenplay  
format  is  itself  a  visualization  of  the  data  and  metadata  of  a  script,  albeit  a  
simple  one.  

Stewart  McKie   -­‐  103  -­‐   As  of  15/08/2014  

Visualizing  the  output  of  a  screenplay  is  the  realm  of  artistic  storyboarding  
and  pre-­‐visualization  software  and  the  movie  production  process  itself.  In  
this  context,  both  storyboarding  and  pre-­‐visualization  –  when  initially  
performed  in  an  automated  fashion  by  software  –  are  an  existing  form  of  
screenplay  analytics  that  is  well-­‐supported  in  current  tools.  Here  the  
screenplay  content  becomes  the  input  that  is  analyzed  to  generate  the  basis  
for  a  storyboard  or  pre-­‐viz  that  can  be  further  enhanced  by  visual  artists  
using  the  software.  The  analysis  consists  of  identifying  scenes  in  the  
screenplay  and  visual  entities  such  as  characters  and  objects  for  example  so  
that  these  can  be  matched  to  visual  assets  in  the  software’s  database  and  
selectively  rendered  as  ‘proto’  storyboards  or  pre  viz  animations.  

Storyboarding  involves  manually  or  electronically  (i.e.  using  a  storyboarding  

software  package)  sketching  out  a  visual  impression  of  the  individual  
camera  shots  required  to  shoot  a  specific  scene  or  set  of  scenes  in  the  movie  
described  by  the  screenplay.  The  storyboard  is  a  collection  of  "boards"  or  
individual  drawings,  in  shot  sequence,  that  give  an  idea  of  the  setting  and  
props,  the  character  action,  the  prop  or  camera  movement,  the  point  of  view  
and  the  general  look  and  feel  of  the  scene.  The  primary  purpose  of  a  
storyboard  is  to  help  the  director  and/or  his  production  managers  imagine,  
block  and  plan  the  shot.  

Using  the  screenplay  as  a  data  source  to  visualize  an  animated  set  of  shots  is  
known  as  “pre-­‐visualization”  or  “pre-­‐viz”.  Pre-­‐visualization  describes  the  
automatic  creation  of  a  visual  product  from  the  script  by  a  software  package,  
prior  to  shooting  the  movie  with  a  camera.  A  pre-­‐visualization  is  either  static  
–  in  the  form  of  computer-­‐generated  storyboard  of  static  images  -­‐  or  
animated  in  the  form  of  some  kind  of  computer-­‐generated  animation  that  
can  be  ‘played’  on-­‐screen.  Like  storyboarding,  the  primary  purpose  of  a  pre-­‐
viz  is  also  to  help  the  director  and/or  the  production  managers  of  a  movie  to  
imagine,  block  and  plan  the  shots  in  a  scene.  

An  animated  pre-­‐visualization  refers  to  the  use  of  animation  software  to  
provide  some  kind  of  basic  ‘moving  picture’  using  the  images  suggested  by  

Stewart  McKie   -­‐  104  -­‐   As  of  15/08/2014  

the  screenplay  text.  By  creating  an  animation  involving  images  of  the  scene  
setting,  the  scene  characters  and  their  movements  and  perhaps  connecting  
this  to  a  synchronized  audio  track  of  the  dialog  being  spoken  and  a  backing  
soundtrack,  a  far  more  sophisticated  level  of  pre-­‐production  visualization  of  
the  movie  is  possible.    

Cavazza,  Friedman,  Liu,  Ramirez  and  others  have  already  done  much  work  
in  the  area  of  screenplay  pre-­‐visualization  by  applying  Natural  Language  
Processing  (NLP)  techniques  and  tools  to  screenplay  texts.  The  results  of  
this  work  include  prototype  tools  such  as  WordsEye  (Coyne  and  Sproat),  
ScriptViz  (Zhi-­‐Qiang  Liu)  and  Scenemaker  (Hanser,  McKevitt,  Lunney  and  
Condell),  designed  to  help  with  the  automated  conversion  of  screenplay  text  
into  computer-­‐generated  animations.  These  animations  typically  include  
generating  3D  characters  and  modeling  their  expressions  and  actions  and  
visualizing  background  scene  environments  for  the  characters  to  be  situated  
in.  This  in  turn  depends  on  the  ability  to  identify  emotions  in  the  text,  as  
expressed  by  the  both  the  dialog  and  action  of  the  characters,  to  significantly  
enhance  the  basic  positive/negative  evaluation  delivered  by  sentiment  
analysis  for  example.  This  has  been  the  focus  of  work  by  Strapparava  and  
Mihalcea,  and  Aman  and  Szpakowicz  among  others.  

Commercially  released  storyboarding  and  pre-­‐visualization  software  such  as  

Frameforge  3D  and  Avid  for  example,  can  already  import  a  screenplay  file  
generated  by  a  compatible  screenwriting  package  and  automatically  
generate  “skeleton”  shot  storyboards  and/or  basic  scene  animations  directly  
from  the  script  text.    

Here  we  are  concerned  with  ways  of  analyzing  and  visualizing  the  textual  
content  of  the  screenplay  primarily  for  the  benefit  of  the  
screenwriter/reader,  rather  than  to  help  the  director  or  production  
managers  imagine  and  plan  shots  or  scenes  in  pseudo-­‐production  mode  to  
create  a  “proto-­‐movie”.  The  aim  is  to  deliver  a  level  of  qualitative  
improvement  prior  to  investing  money  and  resources  into  any  kind  of  pre-­‐
visualization,  such  as  storyboarding  or  animation,  both  of  which  may  be  

Stewart  McKie   -­‐  105  -­‐   As  of  15/08/2014  

relatively  costly  and  time-­‐consuming  exercises  in  of  themselves  and  in  any  
case  may  be  beyond  the  skills  or  interest  of  the  screenplay  author(s).  

2.1  Why  bother  analyzing  screenplay  content?  

Franzosi  (2010:4-­‐5)  asks  the  question:  ‘Words  are  beautiful:  Why  do  you  
want  to  turn  them  into  numbers?’  He  answers  his  question  thus:  ‘I  quantify  
simply  because  I  have  far  too  much  information  to  deal  with  qualitatively’.  

Analyzing  screenplays  is  not  a  well-­‐established  niche  of  analytics  but  

analyzing  released  movies  is  fast  becoming  a  popular  pastime  of  bloggers,  
screenwriting  manual  writers  and  websites  focused  on  the  more  sensational  
aspects  of  the  movie  world  such  as  the  Oscar  awards.    The  web  sites  
thecredits.org  and  brandwatch.com  collaborated  to  analyze  and  visualize  the  
social  media  ‘buzz’  around  the  85th  Academy  Awards  (2013)  to  try  to  predict  
the  Oscar  awards  by  using  a  social  analytic  algorithm  (authored  by  Edward  
Crook)  to  analyze  postings  on  public  forums,  Facebook,  Twitter  etc.  
(http://www.brandwatch.com/2013/02/introducing-­‐the-­‐social-­‐oscars-­‐
tracking-­‐award-­‐buzz/  -­‐  accessed  25  February  2013).  

Stewart  McKie   -­‐  106  -­‐   As  of  15/08/2014  

Figure  2.4  -­‐  Brandwatch  Social  Oscars  DataViz  

Brandwatch  then  produced  a  comprehensive  ‘wisdom  of  crowds’  pre-­‐Oscar  

infographic  based  on  this  social  buzz,  which  proved  remarkably  accurate  in  
terms  of  its  Oscar  award  predications  (see  
http://www.brandwatch.com/2013/02/the-­‐power-­‐of-­‐crowds-­‐predicting-­‐
the-­‐oscar-­‐winners/-­‐  accessed  25  February  2013).    But  these  visualizations  
tap  into  a  relatively  big  social  dataset  that  reflects  international  interest  in  
movies  in  general  and  the  Oscars  in  particular.  Screenplays  do  not  generate  
the  same  general  interest  as  movies  and  either  they  do  not  involve  a  social  
network  at  all  or  do  not  involve  a  significant  enough  ‘crowd’  from  whom  
‘wisdom’  can  be  sourced  for  analysis  and  visualization  purposes.      

Nevertheless,  the  content  of  a  screenplay  must  be  –  and  is  -­‐  of  interest  to  
someone.  A  screenplay  is  part  of  a  commercial  value  chain  and,  until  it  is  
realized  as  a  movie,  represents  latent  value  for  every  stakeholder  with  a  
financial  interest  in  the  production  and  commercial  success  of  the  output  
movie.  These  stakeholders  typically  include  the  producer,  director,  key  
talent  and  the  writer(s).  

The  writer  is  usually  the  initial  beneficiary  of  this  latent  value.  So  what  is  a  
script  worth  to  a  writer?  The  Writer’s  Guild  of  America  (WGA)  is  certainly  
the  largest  and  most  influential  body  in  the  world  that  represents  
professional  screenwriters.  The  WGA  Schedule  of  Minimums  (2004),  
effective  for  the  period  of  11/1/06-­‐10/31/07,  requires  that  WGA  
represented  writers  be  paid  between  $56,500  -­‐  $106,070  for  an  original  
screenplay  including  treatment.    

In  reality,  many  “hot”  screenplays  are  bought  for  considerably  more  than  
this  WGA  minimum.  Using  sources  such  as  Daily  Variety,  Hollywood  Reporter,  
DoneDealPro.com  and  Trackingboard.com,  Scott  Myers  reported  scripts  
selling  for  up  to  $2  million  on  his  blog  
(http://www.gointothestory.com/2009/01/spec-­‐script-­‐sales-­‐analysis-­‐
2008-­‐big.html  -­‐  accessed  27  July  2009).  However,  very  few  spec  scripts  are  

Stewart  McKie   -­‐  107  -­‐   As  of  15/08/2014  

actually  sold  each  year.  According  to  Hollywood  Reporter  (Dec  23-­‐25,  2006  
issue),  only  77  spec  scripts  were  purchased  in  2005  
(http://fencingwiththefog.blogspot.ca/2006/01/screenplay-­‐sales-­‐
statistics.html  -­‐  accessed  18  March  2014).  

Not  only  are  screenplays  worth  considerable  sums  of  money  to  a  writer  in  
upfront  fees,  there  may  be  significant  “back-­‐end”  compensation  as  well,  if  
the  writer’s  contract  includes  rewrite  fees  or  percentage  points  of  the  
movie’s  eventual  gross  profit.  In  this  case  the  royalty  revenue  stream  from  
the  “points”,  resulting  from  the  initial  release  revenues  and  other  residuals  
from  DVD/cable/pay-­‐TV  release  etc.,  may  well-­‐outstrip  the  upfront  
payment.  In  rough  terms,  the  initial  payment  may  equate  to  at  least  a  
reasonable  year’s  salary  and  the  points  royalties  anything  up  to  moderate  
personal  wealth.    

A  screenplay  is  a  valuable  commodity  to  the  selling  screenwriter  but  can  
cost  far  more  significant  sums  for  the  buyer  to  develop.  In  The  Hollywood  
Economist,  Epstein  (2012:63)  reports  that  the  budget  line  for  ‘Story  and  
Rights’  for  Terminator  3:  Rise  of  the  Machines  (2003,  Jonathan  Mostow)  was  
$19.6  million,  which  presumably  included  the  $5.2  million  he  claims  was  
spent  developing  the  initial  script  (2012:  51).  Although  movie  budgets,  as  
reported  by  studios,  are  notoriously  difficult  to  validate  and  verify.  

A  screenplay  is  an  even  more  valuable  commodity  in  an  industry  that  makes  
significant  financial  bet  on  every  screenplay  given  the  green-­‐light  to  proceed  
into  production.  We  can  get  some  idea  of  how  valuable  from  these  estimates  
provided  by  the  Hollywood  tracking  website  The  Numbers  (http://www.the-­‐
numbers.com    -­‐  accessed  27  July  2009).  

Until  Avatar  (2009,  James  Cameron),  the  highest  grossing  movie  ever  was  
Titanic  (1997,  James  Cameron)  with  an  estimated  worldwide  gross  of  $1.849  
billion  against  an  estimated  budget  of  $200  million.  The  highest  grossing  
movie  of  2008  was  thought  to  be  The  Dark  Knight  (2008,  Christopher  Nolan)  
with  an  estimated  worldwide  gross  of    $1  billion  against  an  estimated  

Stewart  McKie   -­‐  108  -­‐   As  of  15/08/2014  

budget  of  $185  million.  However  Box  Office  Mojo  claims  that  Avatar  has  now  
become  the  world’s  highest  grossing  movie,  having  grossed  $2.075  billion  
worldwide  so  far.  (http://boxofficemojo.com/news/?id=2667  -­‐  accessed  30  
August  2012).  

Clearly  giving  the  green  light  to  produce  a  movie  from  a  screenplay  is  
massively  risky  and  has  the  potential  to  return  either  stellar  profits  or  
generate  losses  that  could  threaten  to  bankrupt  studios  or  make  them  a  
takeover  target  (cf.  United  Artists  and  Michael  Cimino’s  Heaven’s  Gate  
(1980)).  That’s  why  perhaps  the  only  well-­‐established  branch  of  screenplay  
analytics  is  focused  on  the  use  of  predictive  analytics  to  use  pre-­‐defined    
criteria  to  try  to  predict  a  box  office  hit  to  help  potential  screenplay  
investors  to  minimize  their  risk.      

The  idea  of  defining  rules  to  be  applied  to  screenplays  for  evaluating  the  
quality  of  the  content  is  not  new.  One  of  the  first  sets  of  rules  published  
specifically  to  guide  screenwriters  can  be  found  in  chapter  XI  of  the  Palmer  
Plan  Handbook  (1919).  William  C.  DeMille’s  rules  comprise  ten  qualitative  
rules  governing  ‘Story  Requirements’  and  a  further  ten  qualitative  rules  to  
avoid  screenplay  rejection.  A  few  decades  later,  Vale  (1944:282-­‐287)  
identified  142  ‘of  the  most  common  mistakes  from  which  the  questions  
necessary  for  analysis  can  easily  be  derived’.  Rules  can  be  found  in  many  
screenplay  writing  manuals  such  as  Flinn’s  How  Not  to  Write  a  Screenplay:  
101  Common  Mistakes  Most  Screenwriters  Make  and  frequently  popup  on  the  
web,  for  example  the  “22  Rules”  proposed  by  Pixar  storyboard  artist  Emma  
Coats  (see  http://slacktory.com/2012/07/pixar-­‐story-­‐rules-­‐illustrated-­‐by-­‐
icanlegothat/  -­‐  accessed  3  September  2012).    

However,  it  should  be  noted  that  no  screenwriting  applications  today  
actually  codify  and  apply  any  of  these  kinds  of  rules  as  a  means  to  evaluate  
screenplay  quality.  

Stewart  McKie   -­‐  109  -­‐   As  of  15/08/2014  

Predictive  Analytics  

Siegel  (2013:3)  makes  the  point  that  data  ‘embodies  a  priceless  collection  of  
experience  from  which  to  learn’  and  predictive  analytics  is  all  about  
analyzing  data  to  make  predictions  based  primarily  on  discerning  patterns  
in  the  evidential  data,  often  based  on  the  application  of  domain-­‐specific  rules  
to  test  the  data.  

ScriptReader  (http://scriptreaderonline.com  -­‐  accessed  3  September  2012)  

provides  a  detailed  checklist  of  questions/rules  that  claims  to  provide  ‘a  
quantitative  approach  to  screenplay  analysis’  based  on  a  proprietary  
‘evaluation  algorithm’  developed  as  a  ‘software  analysis  tool  for  a  major  
Entertainment  Agency’  so  writers  can  find  out  ‘if  your  screenplay  will  get  a  
Pass,  Consider  or  Recommend’.  However  this  is  a  manual  process  that  is  in  
effect  a  self-­‐assessment  by  the  writer,  which  is  then  scored  to  indicate  the  
“grade”  the  screenplay  could  get  from  an  industry  reader,  rather  than  an  
automated  process  whereby  the  screenplay  content  is  subjected  to  software  
algorithms  that  ask  and  answer  these  kinds  of  grading  questions  by  
generating  analysis  reports  and  visualizations  at  the  click  of  a  button.    

However,  according  to  The  Guardian  (July  13,  2007:3),  a  British  firm  called  
Epagogix  says  it  has  ‘designed  a  computer  programme  to  assess  a  proposed  
movie’s  likelihood  of  success…’  and  ‘claims  it  can  estimate  80%  of  projects’  
likely  US  box  office  to  within  $10m  of  the  final  figure.’  Apparently  it  does  this  
by  breaking  down  a  script  into  ‘hundreds  of  constituent  elements…assigning  
each  one  a  commercial  value.’  Over  time,  if  this  process  results  in  high  levels  
of  predicative  success,  then  the  element  mix  and  value  assignments  will  be  
further  optimized,  as  the  program  (or  its  designers)  learn  more  about  what  
makes  a  movie  a  “surefire”  commercial  success.    

However  the  target  market  for  the  Epagogix  commercial  valuation  service  is  
not  the  screenwriter  but  the  studio  or  production  company.  The  sort  of  fees  
that  studios  might  be  prepared  to  pay  for  this  kind  of  information  can  only  
be  guessed  at.  In  The  New  York  Times  article  mentioned  above  relating  to  the  

Stewart  McKie   -­‐  110  -­‐   As  of  15/08/2014  

Worldwide  Motion  Picture  Group,  the  amount  paid  for  script  analysis  
services  could  be  ‘as  much  as  $20,000  per  script’.  On  this  basis,  the  Epagogix  
business  model  is  unlikely  to  accommodate  individual  screenwriters  
wishing  to  gain  a  better  understanding  of  the  commercial  potential  of  their  
screenplay.  Yet  the  fact  that  Epagogix,  Worldwide  Motion  Picture  Group  and  
others  exist  at  all  and  studios  are  apparently  willing  to  pay  for  their  services  
is  an  indication  that  predictive  screenplay  analysis  is  a  viable  and  valuable  
service  proposition.    

Epagogix  and  other  service  providers  –  such  as  Dave  Kelly  Entertainment,  
which  claims  to  use  a  package  called  Gold  Light  to  perform  proprietary  
analytics  on  screenplays  (See  
http://www.davekellyentertainment.com/marketsolution.html  -­‐  accessed  
11  August  2007)  -­‐  are  almost  certainly  leveraging  software  tools  and  
techniques  known  collectively  as  “predictive  analytics”.    

Eliashberg,  Hui  and  Zhang  (2006)  discuss  the  use  of  predictive  analytics  as  a  
means  to  determine  a  movie’s  potential  success.  In  an  appendix  (2006:27),  
the  authors  propose  22  questions  that  can  be  used  to  help  gauge  the  
potential  success  of  a  movie  based  on  the  characteristics  of  the  script.  These  
questions  are  interesting  in  the  context  of  this  thesis  so  they  are  reproduced  
below  in  full:  

1) Clear Premise (CLRPREM): The story has a clear premise that is important to
audiences.

2) Familiar Setting (FAMSET): The setting of the story is familiar to you.

3) Early Exposition (EAREXP): Information about characters comes very early in the
story.

4) Coincidence Avoidance (COAVOID): Story follows a logical, causal relationship.

Coincidences are avoided.

5) Inter-Connected (INTCON): Each scene description advances the plot and is closely
connected to the central conflict.

Stewart  McKie   -­‐  111  -­‐   As  of  15/08/2014  

6) Surprise (SURP): The story contains elements of surprise, but is logical within context
and within its own rules.

7) Anticipation (ANTICI): Keep readers trying to anticipate what would happen next.

8) Flashback Avoidance (FLHAVOID): The story does not contain flashback sequences.

9) Linear Timeline (LINTIME): The story unfolds in chronological order.

10) Clear Motivation (CLRMOT): The hero of the story has a clear outer motivation (what
he/she wants to achieve by the end of the movie).

11) Multi-dimensional Hero (MULDIM): Many dimensions of the hero are explored.

12) Strong Nemesis (STRNEM): There is a strong nemesis in the story.

13) Sympathetic Hero (SYMHERO): Hero attracts your sympathy because he/she
exhibits courage AND belongs to one of the followings: -good/nice, funny, good at what
he does OR has power.

14) Logical Characters (LOGIC): Actions of main characters are logical considering their
characteristics. They sometimes hold surprises but are believable.

15) Character Growth (CHARGROW): Conflict is important enough to change the hero.

16) Important Conflict (IMP): The story has a very clear conflict, which involves high
emotional stakes

17) Multi-Dimensional Conflict (MULCONF): The central conflict is explained in many

different points of view.

18) Conflict Build-up (BUILD): The hero faces a series of hurdles. Each successive hurdle
is greater and more provocative than the previous ones.

19) Conflict Lock-in (LOCKIN): The hero is locked into the conflict very early in the movie.

20) Unambiguous Resolution (RESOLUT): Conflicts is unambiguously resolved through

confrontation between the hero and nemesis at the end.

Stewart  McKie   -­‐  112  -­‐   As  of  15/08/2014  

21) Logical Ending (LOGICEND): The ending is logical and believable.

22) Surprise Ending (SURPEND): The ending carries surprise and is unexpected.

In  the  opinion  of  Eliashberg  et  al.  positive  answers  to  these  questions  are  
likely  to  indicate  good  potential  for  commercial  success,  therefore  
purchasers  of  screenplays  may  have  a  clear  interest  in  analyzing  screenplays  
by  leveraging  this  methodology.    

In  their  2010  paper,  Green-­‐lighting  Movie  Scripts:  Revenue  Forecasting  and  

Risk  Management,  Eliashberg,  Hui  and  Zhang  state  that:  

As  a  script  is  the  very  foundation  of  a  movie,  a  sophisticated  analysis  of  
the  textual  information  and  hidden  story  structures  in  the  script  help  
us  better  predict  box  office  revenues.”  (2010:3)  

By  analyzing  the  genre/content,  words  and  semantics  of  a  script  and  using  
them  as  predictors  in  a  Bayesian  Additive  Regression  Tree  for  Quasi-­‐Linear  
model  (BART-­‐QL),  the  authors’  claim  that:  

…our  model’s  capability  to  generate  predictive  distribution  of  the  box  
office  revenue  not  only  allows  a  studio  to  assess  the  risk  associated  
with  a  point  forecast,  but  also  opens  new  doors  for  a  studio  to  optimize  
its  portfolio  choice  and  manage  its  risk  exposures.  

-­‐  a  model  one  can  imagine  that  every  production  studio  would  want  to  apply  
before  investing  further  funds  in  script  development.  

Sentiment  Analytics  

 
Other  than  predictive  analytics,  another  kind  of  analytics  that  could  be  
applied  for  screenplay  analytics  is  sentiment  analysis.  Sentiment  analysis  is  
used  to  deduce  positive,  neutral  and  negative  sentiment  from  a  body  of  text.  
It  has  become  a  popular  way  to  indicate  the  sentiment  of  the  ‘crowd’  –  for  
example  to  determine  what  is  trending  on  Twitter  -­‐  by  analyzing  data  
streams  on  social  networks.  For  example,  it  has  been  used  to  analyze  
sentiment  about  products  from  tweets  posted  to  Twitter  (see  Go,  Bhayani  
and  Huang  at  

Stewart  McKie   -­‐  113  -­‐   As  of  15/08/2014  

http://cs.stanford.edu/people/alecmgo/papers/TwitterDistantSupervision0
9.pdf  -­‐  accessed  11  June  2013)  and  to  analyze  responses  to  movies  via  
reviews  posted  to  Digg    (see  Yessenov  and  Misailovic    at  
http://people.csail.mit.edu/kuat/courses/6.863/report.pdf  -­‐  accessed  11  
June  2013)  and  via  movie  reviews  posted  on  discussion  boards  (see  Thet,  Na  
and  Khoo  at  http://jis.sagepub.com/content/36/6/823.full.pdf+html  -­‐  
accessed  11  June  2013).  

Sentiment  analytics  can  be  applied  to  the  text  of  an  individual  screenplay,  
rather  than  a  data  stream,  in  order  to  indicate  its  overall  ‘positivity’  or  
‘negativity’  at  the  script,  sequence/storyline  or  scene  levels.    It  can  be  used  
for  a  number  of  purposes:  

• To  identify  positive  or  negative  action  or  dialog  snippets  in  order  to  
help  the  writer  to  increase/decrease  this  positivity/negativity  in  the  
script  

• To  confirm,  from  the  data,  a  tough  start  to  a  script  or  a  happy  ending.    

• To  differentiate  conflict  (negative)  from  resolution  (positive).    

• As  the  basis  to  visualize  a  positive/negative  ‘rollercoaster’  chart  from  

the  script  (see  figures  2.5a/b  below).    

Of  course  it  could  also  be  used  to  determine  the  sentiment  ‘heartbeat’  
characteristic  of  a  corpus  of  screenplays  within  a  genre  in  order  to  
understand  if  a  specific  screenplay  conforms  to  or  subverts  genre  sentiment  
conventions.  

The  positive/negative  aspect  of  sentiment  analysis  is  reflected  in  some  ‘story  
chart’  visualizations  from  James  Dai  and  Robert  McKee.  In  both  cases,  the  
visualizations  are  based  on  multiple  plotlines  within  the  screenplay  and  both  
also  highlight  structural  plot  points.  

James  Dai’s  Story  Charting  (see  

http://www.storycharts.ca/pages/casablanca/  -­‐  accessed  11  June  2013)  

Stewart  McKie   -­‐  114  -­‐   As  of  15/08/2014  

combines  some  interesting  structural  propositions  with  useful  visualizations  
of  plot  lines  and  turning  points  In  his  analysis  of  Casablanca,  his  Story  Chart  
visualization  (see  figure  2.5a)  charts  three  plotlines:  

1.  External  plot  –  Escape  Quest  (Laszlo  and  Ilsa)  

2.  Relationship  plot  –  Love  between  Rick  and  Ilsa  

3.  Internal  plot  –  Rick’s  patriotism  

The  circles  represent  the  key  turning  points  in  those  plotlines  with  a  positive  
(above  the  line)  or  negative  (below  the  line)  impact  over  time.  

Figure  2.5a  –  Story  Chart  for  Casablanca    

Figure  2.5b  is  another  kind  of  story  chart,  this  time  of  Tarantino’s  Pulp  
Fiction,  and  is  attributed  to  Robert  McKee  (see  
http://artscienceblog.blogspot.co.uk/2012/04/should-­‐good-­‐brand-­‐stroy-­‐
chart-­‐like-­‐good.html  -­‐  accessed  11  June  2013).  Here  the  positive  and  

Stewart  McKie   -­‐  115  -­‐   As  of  15/08/2014  

negative  ‘planes’  are  superimposed  on  the  acts  and  sequences  and  the  
inciting  incidents,  turning  points  and  climaxes  more  clearly  indicated  along  
the  plotlines.  

Figure  2.5b  –  McKee’s  Story  Chart  of  Pulp  Fiction.  

There  are  clearly  viable  commercial  reasons  for  analyzing  screenplay  content  
and  people  already  delivering  it  as  a  business  offering.  So  what  is  worth  
analyzing?  

2.2  What  is  Worth  Analyzing?  

Corrigan  (2007:36-­‐81),  discussing  analyzing  and  writing  about  film,  

suggests  a  number  of  topics  for  film  analysis  and  writing  that  are  equally  
applicable  to  screenplay  analysis  including:  Themes,  narrative,  characters  
and  point  of  view.  A  cursory  review  of  the  table  of  contents  from  almost  any  
“how-­‐to”  screenwriting  manual  may  also  act  as  a  guide  to  what  is  worth  
analyzing.  For  example,  table  2.1  lists  a  representative  range  of  topics  from  a  
random  set  of  ten  screenwriting  manuals  to  show  how  some  topics  may  be  
obvious  targets  for  analysis  and  visualization  (an  X  indicates  that  the  topic  is  
either  explicitly  mentioned  in  the  contents  or  has  an  appropriate  level  of  
referencing  in  the  index).  

Stewart  McKie   -­‐  116  -­‐   As  of  15/08/2014  

  

Screenwriting Manual

Sequence
Character

Structure
Narrative
Genre
Dialog

Scene
Plot

Theme
Advanced  Screenwriting  (Seger,  2003)   X X X X X X X

(The)  Art  and  Science  of  Screenwriting  (Parker,  1999)   X X X X X X

Crafty  Screenwriting  (Epstein,  2002)   X X X X

Screenplay  (Field,  2005)   X X X X X X

Screenwriting  Updated  (Aronson,  2001)   X

Story  (McKee,  1998)   X X X X

Writing  a  Great  Movie  (Kitchen,  2006)   X X X X

Writing  Drama  (Lavandier,  2005)   X X X X X

Writing  Movies  (Gotham  Writers  Workshop,  2006)   X X X X X

Writing  the  Screenplay  (Armer,  1993)   X X X X

Table  2.1  -­‐  Potential  Screenplay  Analysis  Topics      

Based  on  this  table,  some  obvious  potential  analytic  subjects  are:  

• character  (and  dialog)  

• genre  (and  theme)  
• narrative  (and  plot)  
• structure  (including  sequences  and  scenes)  

So  in  order  to  narrow  the  scope  of  screenplay  analysis  and  visualization  
discussed  here,  I  will  focus  my  attention  on  these  two  topics:  Character  and  
Narrative,  because  I  believe  that  the  interplay  of  character  and  narrative  
represent  the  essential  DNA  of  a  screenplay,  they  are  easier  for  me  to  apply  
analytics  to  and  do  not  require  access  to  or  the  creation  of  a  script  corpus  as  
genre  analysis  would  do.  My  discussion  of  potential  character  analytics  will  
embrace  both  action  (what  characters  do)  and  dialog  (what  characters  say)  
elements  of  the  screenplay.  My  discussion  of  narrative  will  embrace  plot,  
structure  and  theme.    

Stewart  McKie   -­‐  117  -­‐   As  of  15/08/2014  

I  have  decided  not  to  include  discussion  of  genre  analysis  for  the  following  
reasons.  Firstly,  there  is  no  clear  agreement  about  what  the  range  of  genres  
actually  is  or  what  exactly  defines  a  specific  genre.    In  other  words  there  is  
no  universally  accepted  standard  to  work  from,  for  as  Dancyer  suggests  in  
The  flexibility  of  genre  (in  Analysing  the  Screenplay,  2011:122),  ‘Genres  are  
not  fixed  forms.  They  are  no  more  than  templates  for  audience  recognition  
and  interest.’  This  is  further  reflected  in  De  Haan’s  comment  (2004:7)  in  The  
Consulting  Process  as  Drama  relating  to  client  and  consultant  that  could  
equally  well  apply  to  an  audience  and  a  movie  perceived  to  be  of  a  specific  
genre,  ‘Experience  shows  that  usually  the  entry  “starts  before  it  starts”:  both  
the  client  and  the  consultant  have  images  and  expectations  of  the  other,  even  
before  they  meet.’    And  Edgar’s  reference  to  Terence  Hawkes  (2009:66)  that  
“genre  enables  the  reader  ‘to  decode  literature  in  the  same  mode  as  it  was  
encoded  by  the  writer’”.  This  is  as  it  should  be  because  genre  is  a  dynamic  
and  developing  notion  that  reflects  the  times,  culture,  audience  whims  and  
so  on.    

Secondly,  I  do  not  believe  that  screenwriters  writing  a  spec  script  set  out  to  
“write  to  genre”  -­‐  unless  they  specifically  decide  to  do  so  or  are  specifically  
commissioned  to  do  so  and  then  the  genre  conventions  that  they  pay  
attention  to  will  be  modeled  on  a  specific  exemplar  movie  or  have  been  
defined  for  them  as  specific  instructions  (e.g.  “go  write  Jaws  in  space”).  
Thirdly  I  suggest  that  genre  is  largely  a  marketing  decision  that  is  either  the  
driving  genesis  of  the  screenplay  or  grafted  onto  a  screenplay  later,  
triggering  some  kind  of  “adapt  to  genre”  rewrite.  Fourthly  it  is  not  entirely  
clear  that  any  movie  is  necessarily  representative  of  a  single  genre.  For  
example,  is  Ridley  Scott’s  Alien  (1979)  a  representative  of  horror,  sci-­‐fi,  
feminist  or  even  western  genre  or  is  it  an  example  of  some  kind  of  composite  
genre?  For  all  these  reasons,  I  view  screenplay  genre  analysis  as  probably  
the  least  useful  to  focus  on,  especially  in  terms  of  using  analytics  to  improve  
the  quality  of  an  individual  script.  

However,  this  is  not  to  say  that  genre  analysis  is  not  an  interesting  area  for  

Stewart  McKie   -­‐  118  -­‐   As  of  15/08/2014  

screenplay  analytics.  Walker,  Lin  and  Sawyer  analyzed  a  corpus  of  862  film  
scripts  from  the  Internet  Movie  Script  Database  (IMSDb)  in  order  to  learn  
about  and  characterize  character  style  from  the  dialog  spoken  in  the  scripts  
so  that  the  results  can  be  used  as  parameters  to  feed  into  Mairesse  and  
Walker’s  PERSONAGE  (personality  generator)  tool.  Blackstock  and  Spitz  also  
analyzed  399  scripts  in  order  to  determine  they  could  successfully  classify  
the  genre  of  a  script,  Their  techniques  proved  to  be  ‘up  to  55%  accurate  at  
nailing  every  single  genre  for  a  given  movie  exactly.’    Genre  analysis,  as  a  
niche  of  screenplay  analytics,  clearly  has  much  potential  for  future  research.    

In  each  sub-­‐section  that  follows,  wherever  possible,  I  will  first  outline  the  
scope  of  analysis  and  visualization  topic,  then  suggest:    

• why  a  specific  kind  of  analysis  may  be  useful  and  benefit  the  
screenwriter  

• how  the  analytical  subject  in  question  may  be  identified  within  the  
screenplay  content  programmatically  

• what  form  any  visualization  of  the  analysis  may  take  

2.3  Character    

Aristotle  (1996:12)  believed  that  character  takes  second  place  to  plot  and  
that  people  can  be  differentiated  in  only  two  ways  –  by  defect  or  excellence  
(1996:5).  Whereas  in  The  art  of  fiction  Henry  James  (1963:80)  was  more  
ambivalent,  asserting:  ‘What  is  character  but  the  determination  of  incident?  
What  is  incident  but  the  illustration  of  character?’  which  sounds  remarkably  
like  George  Eliot  (from  Adam  Bede  as  quoted  by  Chatman  1978:96),  ‘Our  
deeds  determine  us,  as  much  as  we  determine  our  deeds’.  Culler  (cited  in  
Rimmon-­‐Kenan  2002:31)  offers  the  perspective  that  ‘The  notion  of  
character,  structuralists  would  say,  is  a  myth.’  Whether  mythical  or  not,  
Rimmon-­‐Kenan  (2002:31)  identifies  two  problems  with  characters:  Are  they  
about  people  or  words  or  about  being  or  doing?  Or  put  another  way,  in  
screenplay  as  data  terms,  are  they  about  dialog  or  action  snippets?  
Stewart  McKie   -­‐  119  -­‐   As  of  15/08/2014  
Some  popular  screenwriting  gurus  regard  effective  characters  as  an  
essential  part  of  every  screenplay  and  critical  to  the  quality  and  commercial  
potential  of  every  script.  In  Screenplay:  The  Foundations  of  Screenwriting  
Field  (2005:46)  says,  ‘Character  is  the  essential  internal  foundation  of  your  
screenplay.  The  cornerstone.  It  is  the  heart  and  soul  and  nervous  system  of  
your  screenplay.’  Corrigan  (2007:44)  states  that  characters  ‘…normally  
focus  the  action  and,  often,  the  themes  of  a  movie.’  Lavandier  (2005:110)  
believes  that  ‘Drama  consists  of  imitating  human  actions  therefore  of  
creating  characters…’  McKee’s  Story  (1998:106)  asserts  that,  ‘The  function  
of  CHARACTER  is  to  bring  to  the  story  the  qualities  of  characterization  
necessary  to  convincingly  act  out  choices.’  In  screenplay  element  terms,  
characters  are  usually  involved  in  both  a  series  of  action  (or  scene  
description)  blocks  and  in  dialog  blocks  -­‐  either  as  messenger  or  receiver.  
This  section  will  focus  primarily  on  improving  quality  of  characters  in  a  
screenplay  by  appropriate  analysis  of  what  a  character  does  and  says.  

A  movie  without  strong  characters  we  believe  in  and  can  empathize  with  is  
almost  certain  to  fail.  This  is  because  it  is  difficult  to  care  about  a  movie  as  a  
whole,  if  there  are  no  characters  that  we  can  relate  to,  that  we  can  root  for,  
that  we  can  despise  or  love,  or  that  we  can  identify  with  in  some  way.  We  
want  to  be  shocked  or  disgusted  by  characters,  to  be  comforted  or  turned-­‐on  
by  characters,  to  be  interested  in  what  happens  to  them  and  why  they  are  
the  way  they  are  and  do  the  things  they  do.  Much  of  the  rationale  for  
audiences  ignoring  the  fatigue  of  cinema  movie-­‐watching  -­‐  concentrating  on  
a  darkened  screen  for  around  two  hours  -­‐  and  staying  to  the  end  of  a  movie  
is  tied  up  with  finding  out  what  happens  to  specific  characters  or  how  
specific  character  development  or  relationships  will  play-­‐out,  or  as  Indick  
(2004:  xi)  puts  it:  ‘Through  the  unconscious  process  of  “identification,”  the  
people  in  the  audience  actually  become  the  characters  that  they  identify  
with  in  the  script.’  

There’s  no  doubt  that  it’s  not  only  audiences  that  value  characters,  so  too  
does  the  marketplace.  As  Elberse  relates  in  Blockbusters  (2013:  pp.48-­‐55)  

Stewart  McKie   -­‐  120  -­‐   As  of  15/08/2014  

characters  that  can  drive  blockbuster  movies  are  worth  a  fortune.  In  2009  
Disney  paid  $4  billion  to  purchase  Marvel  Entertainment  (2013:48),  a  deal  
that  has  already  resulted  in  billions  of  dollars  of  revenues  from  subsequent  
movie  releases  involving  just  a  handful  of  Marvel  characters  (2013:55).    
Elberse    (2013:51)  quotes  Bob  Iger  (Disney’s  CEO  at  the  time)  as  saying  
‘This  treasure  trove  of  over  5000  characters  offers  Disney  the  ability  to  do  
what  we  do  best’  i.e.  make  money  from  movies.    Assuming  a  few  dud  
characters  in  the  bunch  it’s  fair  to  conclude  that  Disney  paid  around  $1  
million  for  each  Marvel  character  in  the  comic  book  publishers  stable  
(including  their  linked  merchandising  capability).    

However,  what  is  it  that  defines  characters  in  a  screenplay?  Is  it  only  what  
Egri  in  The  Art  of  Dramatic  Writing  calls  the  “bone  structure”  of  physiology,  
sociology  and  psychology  (2004:37-­‐38)?  Certainly  one  differentiator  is  their  
exterior  characteristics  –  their  physical  and  sartorial  makeup.  Another  is  
their  habits  and  “tics”  –  what  Chatman  (1978:127)  called  a  set  of  “traits”.  But  
these  obvious  surface  characteristics  serve  primarily  as  “identifiers”  to  
differentiate  characters  visually  for  the  audience,  create  some  kind  of  
expectation  as  to  how  these  characters  might  behave  as  the  movie  
progresses  and  use  to  remind  the  audience  of  a  prior  event  (e.g.  a  facial  scar  
from  an  important  backstory  confrontation).    

Character  is  also  more  than  the  ‘figural’  aspects  of  characterization  
identified  by  Pfister  in  his  Theory  and  Analysis  of  Drama  (1991:  184).  Pfister,  
in  reference  to  a  ‘dramatic  text’,  defines  ‘explicit’  techniques  as  including  
‘self-­‐commentary’  and  ‘commentary  by  others’  and  ‘implicit’  techniques  as  
non-­‐verbal  and  verbal.  If  this  were  all  a  character  comprised  of  in  a  
screenplay,  the  character  would  likely  be  viewed  in  a  stereotypical  way  that  
is  unlikely  to  excite  much  interest  or  empathy  from  a  movie  audience.    

But  a  more  qualitative  differentiator  comprises  their  interior  characteristics  

–  their  emotional  depth,  maturity  and  psychological  profile  for  example.  
While  it  is  possible  that  a  character  can  change  their  physical  appearance  or  
traits  over  the  course  of  a  movie,  it  is  more  likely  that  changes  in  their  

Stewart  McKie   -­‐  121  -­‐   As  of  15/08/2014  

internal  characteristics  that  will  be  of  more  interest  to  an  audience,  since  
these  are  “deep”  changes  in  nature  rather  than  “shallow”  and  more  obvious  
surface  changes.  Just  as  it  is  more  interesting  to  see  how  a  business  changes  
its  internal  culture  than  simply  watch  it  move  from  one  office  block  to  
another.  

Another  key  aspect  to  character  depth  is  how  they  relate  to  each  other  –  the  
relational  aspect  of  character.  Imagine  how  much  less  interesting  the  Travis  
Bickle  (Robert  De  Niro)  character  in  Martin  Scorcese’s  Taxi  Driver  (1976)  
would  be  if  all  we  ever  saw  and  heard  of  him  was  as  a  lonely  cab  driver  
cruising  the  New  York  streets  narrating  his  view  of  life  and  the  city  by  
constant  voiceover.    Bickle’s  character  becomes  interesting  as  it  is  developed  
through  his  relationships  with  the  two  lead  female  roles  of  Iris  (Jodie  Foster)  
and  Betsy  (Cybill  Shepherd).      

Screenwriters  may  choose  to  use  a  character  type  or  personality  model  to  
base  their  characters  on.  These  include  simple  archetypes  such  as  Indick’s  4-­‐
dimensional  ‘Quatrains’  (2004:139)  for  female  and  male  characters  up  to  the  
particularly  rich  Enneagram  model  with  its  nine  styles  that  Lee  describes  in  
The  Psychology  of  Screenwriting  as  ‘a  workable,  accurate,  and  highly  nuanced  
psychological  tool  for  developing  characters  and  understanding’  (2013:87).  
The  Enneagram  also  benefits  from  the  styles  themselves  being  visualized,  as  
shown  in  figure  2.6  below  (see  http://www.9types.com/epd/1.php  -­‐  
accessed  02  July  2013).    

Stewart  McKie   -­‐  122  -­‐   As  of  15/08/2014  

  

Figure  2.6  -­‐  The  Reformer  Enneagram  Style  

The  fundamentals  of  screenplay  character  analysis  (via  technology)  assume  

the  following  can  be  easily  identified  from  the  screenplay  data:  

• The  character’s  entrance  (first  appearance)  and  exit  (last  

appearance)  in  the  script.  

• What  dialog  the  character  speaks  and  to  whom.  

• What  action  the  character  takes  part  in,  where  and  with  whom.  

• Where  the  character  is  included  in  scenes  that  are  part  of  the  
narrative  line  of  specific  plots/sub-­‐plots/storylines  in  the  
screenplay.  

The  character’s  entrance  and  exit  is  used  to  establish  the  extent  of  their  
“throughline”  or  participation  in  the  script.  Within  this  throughline  will  be  
blocks  of  dialog  spoken  by  the  character  and  blocks  of  action  that  the  
character  participates  in.  Note  that  other  characters  may  refer  to  a  character  

Stewart  McKie   -­‐  123  -­‐   As  of  15/08/2014  

in  dialog  or  action  that  takes  place  outside  the  actual  extent  of  the  
character’s  physical  throughline  in  the  script.  And  in  structural  terms,  
characters  participate  in  or  are  referred  to  in  scenes.  Individual  scenes  may  
form  a  component  part  of  a  plot  or  sub-­‐plot  thereby  creating  a  plot-­‐related  
(vs.  a  script  related)  throughline  for  the  character  also.  

Role  and  Relationships  

Characters  in  a  screenplay  may  conform  to  a  specific  role  dynamic  and  
participate  in  a  defined  set  of  role  relationships.  The  most  obvious  of  these  is  
the  hero/villain  or  protagonist/antagonist  roles  and  relationship.  These  
roles  and  the  effectiveness  of  this  relationship  are  often  considered  to  be  
central  to  the  success  of  any  screenplay.  McKee  (1998:379)  claims  that  ‘In  
essence  the  protagonist  creates  the  rest  of  the  cast’  and  visualizes  the  
protagonist  at  the  center  of  what  he  calls  ‘cast  design’.  

Stewart  McKie   -­‐  124  -­‐   As  of  15/08/2014  

Figure  2.7  -­‐  Protagonist-­‐Driven  Cast  Design  (McKee  1998:380)  

Baboulene’s  The  Story  Book  emphasizes  the  story  impact  of  the  two  key  
character  roles  through  a  basic  Freudian  analysis  whereby  the  protagonist  
represents  the  ‘super-­‐ego’  and  the  antagonist  the  ‘id’.  And  it  is  the  interplay  
of  the  two,  in  the  form  of  conflict,  which  creates  the  resulting  ‘ego’  or  story:  
‘The  true  character  of  the  protagonist  and  antagonist,  revealed  through  the  
actions  they  take  in  response  to  the  conflict  between  them.’  [2010:  20].  In  
this  sense,  story  is  neither  exclusively  character  nor  plot,  but  requires  
characters  and  their  actions  to  create  and  drive  plot.    

Figure  2.8  -­‐  Character  conflict  as  story  (adapted  from  Baboulene  2012:20)  

Because  of  their  relative  importance  in  most  screenplays  (except  say  those  
with  an  ensemble  cast),  the  protagonist/antagonist  roles  are  worthy  of  
particular  attention  for  analysis.  For  example,  who  is  the  ‘driver’    -­‐  the  
protagonist  or  the  antagonist?  Is  it  the  protagonist  or  the  antagonist  who  
drives  the  action  by  being  proactive  and  making  choices  that  ‘pull’  the  other  
along  with  them?  Are  antagonist  scenes  triggered  by  a  causal  chain  of  
preceding  protagonist  scenes  or  vice  versa?  Character  analysis  may  be  able  
to  help  a  writer  to  better  understand  who  really  is  the  main  character  in  
their  script  and  therefore  who  this  script  is  really  about.  

But  there  are  many  other  roles  and  relationships  that  add  colour  or  depth  to  
the  central  roles/relationships.  These  include  ally/adversary  roles,  mentor  

Stewart  McKie   -­‐  125  -­‐   As  of  15/08/2014  

and  shapeshifter  roles,  love-­‐interest  and  sibling  roles    -­‐  among  those  roles  
identified  in  Vogler’s  Hero’s  Journey  paradigm.  These  roles  and  relationships  
are  often  more  interesting  if  certain  “twists”  are  applied,  for  example  if  one  
is  unknown  to  the  other,  if  they  are  siblings  at  odds  with  one  another  or  if  
one  role  acts  as  the  “shadow”  of  the  other  (e.g.  doppelgänger  or  ‘mirror’  
roles  –  for  example  the  Richard  Gere  and  Andy  Garcia  roles  in  Internal  
Affairs  (Mike  Figgis,  1990)  or  the  Ethan  (John  Wayne)  and  Scar  (Henry  
Brandon)  characters  in  The  Searchers  (John  Ford,  1956).  

Identifying  

An  indicator  of  role  importance  or  at  least  “richness”  could  be  how  a  specific  
role  relates  to  others  –  either  quantitively  or  qualitatively  –  or  put  another  
way,  the  scope  and  depth  of  the  social  network  of  the  character.  A  wider,  
deeper  social  network  may  indicate  a  more  complex  or  sophisticated  
character.  Here  “wider”  means  interacting  with  more  other  characters  and  
“deeper”  means  more  interactions  either  in  dialog  or  action.  A  wider,  deeper  
social  network  also  has  a  commercial  dimension  because  these  are  the  kinds  
of  challenging  roles  that  major  acting  talent  is  likely  to  be  more  interested  in  
playing  so  a  character  analysis  visualization  could  be  used  to  help  to  ‘sell’  the  
role  to  an  actor.  

Visualizing  

One  way  to  analyze  and  visualize  this  aspect  of  a  character  in  a  screenplay  is  
by  using  a  node-­‐based  social  network  aka  a  friendship  graph  using  clique  
detection  algorithms.  The  user-­‐selected  “focal”  character  is  centralized  (see  
Hamlet  in  figure  2.2  above  and  2.9  below  (see  
http://www.jibble.org/shakespeare/images/hamlet.xml-­‐00000384.png,  
accessed  6  July  2008)  and  the  nodes  that  radiate  from  the  focal  character  
represent  relationships  with  other  characters.  The  relative  size  of  each  
character  node  reflects  the  size  of  their  action  and  or  dialog  elements  in  the  
script  (i.e.  bigger  means  more).  The  thickness  of  the  connecting  lines  
between  characters  reflects  how  often  they  interact  together  i.e.  involved  in  

Stewart  McKie   -­‐  126  -­‐   As  of  15/08/2014  

scenes  or  action  or  dialog  together  (i.e.  thicker  means  more).  The  colour  of  
the  nodes  reflects  the  “group”  the  character  belongs  to  (e.g.  protagonist  and  
allies,  antagonist  and  allies  or  “unaffiliated”).  Social  network  visualizations  
would  usually  be  whole-­‐script  based  but  could  also  be  role,  scene  or  
storyline  based  for  example.  

Figure  2.9  –  Social  Network  of  Hamlet  (Scene  VI)  

Franzosi’s  network  analysis  of  violence  in  fascist  Italy  also  demonstrates  
another  useful  way  to  use  this  kind  of  analysis  when  combined  with  a  time  
series.  His  2  figures  (2010:113)  show  the  ‘star’  network  graph  for  the  Sphere  
of  Action  of  Violence,  first  for  1919-­‐1920  and  then  for  1921-­‐22  that  clearly  
show  (based  on  his  source  data)  that  the  ‘nexus’  of  violence  shifted  from  the  
police  at  the  centre  to  the  fascists  at  the  center  for  the  times  series  analyzed.    
In  a  screenplay  context  this  could  be  used  to  show  the  shifting  ‘power’,  say,  
of  the  protagonists  vs.  the  antagonist  as  the  screenplay  progresses,  based  on  
their  actions  or  dialog.  

Stewart  McKie   -­‐  127  -­‐   As  of  15/08/2014  

Moretti  (2011:3-­‐5)  also  makes  a  number  of  useful  points  about  using  
network  models  to  analyze  ‘the  hierarchy  of  centrality’,  particularly  in  
relation  to  the  protagonist  (he  discusses  plays  but  the  same  principles  apply  
to  screenplays).  By  making  a  network  of  a  play  you  create  a  ‘model’  that  you  
can  experiment  with.  Moretti  proposes  using  this  model  to  test  the  
‘significance’  or  ‘centrality’  of  the  protagonist  by  viewing  the  impact  on  the  
model  when  you  remove  the  protagonist  from  it  or  when  you  remove  other  
characters  from  it  (e.g.  the  antagonist).    

Gibes  and  Anderson  (see  http://clearcongressproject.com/charviz/  -­‐  

accessed  11  June  2013)  have  produced  a  different  kind  of  character  
visualization  that  is    

‘…accomplished  by  encoding  each  typed-­‐character  in  the  screenplay  with  small  
bit  of  visual  information  -­‐  a  tiny  colored  rectangle,  the  color  encoding  which  
character  owns  that  particular  portion  of  text.’  

Figure  2.10  shows  how  this  technique  can  be  used  to  visualize  the  ‘weight’  of  
the  key  characters  across  episodes  of  season  one  of  Buffy  the  Vampire  Slayer.  

Figure  2.10  –  Character  Weights  in  Buffy  

Conflict  and  Challenge  

We  care  about  characters  partly  because  we  are  interested  in  how  they  cope  
with  conflict  and  challenge  –  especially  if  these  characters  are  in  some  way  
“like  us”  or  like  we  would  aspire  to  be.  The  more  conflict  and  challenges  a  

Stewart  McKie   -­‐  128  -­‐   As  of  15/08/2014  

character  is  involved  in,  the  more  likely  that  character  will  retain  our  
interest  –  assuming  the  conflicts  and  challenges  are  realistic  and  balanced  by  
other  kinds  of  character  action.  Indick  (2004)  identifies  many  kinds  of  
character  conflicts  exemplified  in  various  movies,  including  neurotic,  
normative  and  existential,  and  discusses  them  in  psychoanalytic  terms  in  
reference  to  the  work  of  Adler,  Erikson,  Freud,  Jung  and  May.    

Identifying  

One  way  of  identifying  conflict  and  challenge  in  a  screenplay  might  be  by  
linguistic  analysis  of  both  action  and  dialog  elements.  In  the  action  blocks  for  
example,  we  might  look  for  the  propensity  of  certain  words  and  terms  
associated  with  conflict  and  challenge  –  verbs  such  as  “hit”,  “shouted”,  
“pushed”.  In  the  dialog  blocks  we  may  look  for  sequences  of  dialog  focused  
on  two  characters  to  the  exclusion  of  others  where  the  tone  is  analyzed  as  
confrontational  because  it  contains  more  statements  (e.g.  “leave  me  alone”,  
”get  lost”)  than  questions.    

Visualizing  

One  way  to  visualize  this  aspect  of  a  character  in  a  screenplay  is  to  use  a  line  
chart  visualization:    

• Scenes  in  which  the  character  is  not  in  conflict  or  challenged  are  
represented  as  a  straight  line  or  equilibrium.  

• Scenes  involving  the  character  in  conflict  or  being  challenged  and  
responding  positively  or  resolving  positively  (e.g.  high-­‐points”)  are  
represented  by  more  or  less  “static”  above  the  line.    

• Scenes  involving  the  character  in  conflict  or  being  challenged  and  
responding  negatively  or  resolving  positively  (e.g.  low-­‐points”)  are  
represented  by  more  or  less  “static”  below  the  line.    

What  one  might  look  for  are  patterns  of  static  that  also  help  to  visualize  the  

Stewart  McKie   -­‐  129  -­‐   As  of  15/08/2014  

 relative  levels  of  conflict  and  challenge  across  the  duration  of  the  
screenplay.  

Choices  and  Change  

The  concept  of  the  transformational  arc  of  a  character  is  fundamental  to  
Vogler’s  Hero’s  Journey,  in  which  the  hero  is  transformed  –  usually  
positively  (unless  the  screenplay  is  a  tragedy)  -­‐  by  a  series  of  experiential  
stages  and  encounters.  How  the  character  responds  to  an  experience  or  
encounter  in  terms  of  the  choice(s)  made  and  the  change(s)  that  are  
manifested  in  the  character  as  a  result  of  that  choice  is  another  aspect  of  
character  that  may  determine  whether  or  not  an  audience  remains  engaged  
with  a  character  throughout  the  movie,    

A  choice  and  any  resulting  change  that  occurs  may  be  developmentally  
ongoing  or  a  one-­‐off  event  that  literally  defines  much  of  the  way  the  
character  behaves  going  forward.  When  Michael  Corleone  chooses  his  family  
business  over  his  own  wife’s  wish  that  he  eschews  their  way  of  life,  he  
makes  a  choice  that  profoundly  impacts  both  his  own  ongoing  development  
as  a  character  and  the  ensuing  narrative  of  Francis  Ford  Coppola’s  The  
Godfather  (1972).  Similarly  when  Josh  is  temporarily  transformed  in  Big  
(1988,  Penny  Marshall)  from  being  a  child  into  an  adult  with  a  child’s  
worldview,  this  single  event  acts  as  fundamental  change  that  drives  the  rest  
of  the  movie.  

Identifying  

Murtagh  et  al.  (2008)  applied  Euclidean  Embedding,  correspondence  

analysis  and  clustering  to  analyze  the  feature  film  script  Casablanca  (1942)  
and  a  series  of  scripts  from  the  US  TV  show  CSI  (Crime  Scene  Investigation)  
to  attempt  to  reveal  some  aspects  of  the  “deep  structure”  of  the  screenplay  
content.  In  the  case  of  Casablanca  this  resulted  in  the  identification  of  
pivotal  scenes  that  were  indicative  of  some  kind  of  change  in  the  direction  
of  the  narrative.  In  the  case  of  CSI,  Murtagh’s  analysis  confirmed  that:  

Stewart  McKie   -­‐  130  -­‐   As  of  15/08/2014  

 ‘There  is  occasionally  a  very  strong  link  between  commercial  breaks  
and  change  in  thematic  content  as  evidenced  by  the  hierarchy.  In  other  
cases  we  find  continuity  of  content  bridging  the  gap  of  the  commercial  
breaks’  (2008:17).    

Murtagh’s  approach  was  guided  by  story  precepts  proposed  by  Robert  
McKee,  who  claims  that  text  ‘means  the  sensory  surface  of  a  work  of  art’  
(1998:252)  whereas  subtext  is  ‘an  inner  life  that  contrasts  with  or  
contradicts  the  text’  (1998:255).    

Visualizing  

If  it  is  possible  to  identify  moments  of  choice  and  a  subsequent  change  in  
behavior  then  this  aspect  of  character  may  also  be  mapped  in  some  way,  
perhaps  through  a  familiar  flowchart  visualization  using  the  decision  
diamond  to  indicate  scenes  in  which  a  change  point  is  identified  or  an  
‘organization  tree’  representation  of  possible  choices.  This  kind  of  
‘branching  analysis’  could  in  turn  provide  guidance  to  screenwriters  as  to  
alternate  directions  that  the  narrative  may  take  so  these  can  be  more  fully  
understood  and  explored.    

Dynamism  

Characters  that  are  merely  differentiated  by  their  physical  appearance  and  a  
collection  of  traits  are  likely  to  be  viewed  by  an  audience  as  “flat”  or  “static”.    

Movie  audiences  don’t  necessarily  want  their  characters  to  be  wholly  
realistic,  in  fact  they  may  prefer  them  to  be  larger  than  life  in  some  way:  
More  decisive,  more  evil,  more  lucky.  This  suggests  that  dynamic  characters  
who  dominate  the  screen  are  more  likely  to  have  higher  audience  appeal  
than  more  static  characters,  who  appear  infrequently  and  do  less.  A  dynamic  
character  is  likely  to  have  a  wide  social  network,  to  experience  relatively  
more  conflict  and  challenge,  to  make  more  choices  and  exemplify  various  
kinds  of  change.  She  is  also  likely  to  be  more  active  generally  and  speak  
more  influential  dialog.  

Stewart  McKie   -­‐  131  -­‐   As  of  15/08/2014  

Identifying    

Being  more  active  implies  that  the  character  is  associated  with  more  verbs  
and  more  active  verbs  in  particular,  in  scene  action  descriptions.  Speaking  
more  influential  dialog  implies  that  he  may  voice  more  orders  (imperatives)  
or  ask  more  questions  or  that  more  other  characters  listen  to  what  he  says  
or  refer  to  him/her  more  often  themselves  in  their  own  dialog.  This  kind  of  
analysis  requires  that  the  scene  action  the  character  is  involved  in  and  the  
dialog  he  speaks  be  analyzed  in  more  depth.    

Visualizing  

One  way  to  visualize  character  dynamism  is  to  show  the  relative  “footprint”  
of  the  character  in  the  screenplay  using  a  bubble  graph.  A  more  conventional  
way  is  to  show  a  range  of  characters  relative  to  their  ‘position’  against  two  
axes  and  four  quadrants  (see  figure  2.11).  

• Action  Axis  -­‐  more  or  less  active  

• Dialog  Axis  -­‐  more  or  less  commanding/questioning/listened  to  

The  more  active  and  commanding/questioning  characters  are  located  in  the  
upper  right  quadrant,  which  is  where  one  might  expect  to  find  the  main  
protagonist/antagonist.    

Stewart  McKie   -­‐  132  -­‐   As  of  15/08/2014  

Figure  2.11  -­‐  Dynamic  Character  Matrix  

Cohesive  Characters  

Characters  who  do  not  behave  consistently  in  the  course  of  a  movie  –  
assuming  inconsistent  behavior  is  not  a  defining  trait  of  the  character  -­‐  may  
be  viewed  with  suspicion  by  an  audience  and  eventually  cause  them  to  
dismiss  the  character  as  unbelievable.  This  could  have  a  direct  commercial  
impact  by  diminishing  the  “word-­‐of-­‐mouth”  about  a  movie  in  a  negative  way.  
A  consistent  character  is  a  cohesive  character.  Rimmon-­‐Kenan  (2002:39)  
defines  the  main  principles  of  cohesion  as  repetition,  similarity,  contrast  and  
implication.  

2.4  Narrative  

A  story  consists  of  a  series  of  related  events,  whether  that  series  is  ordered  
forwards  or  backwards  in  time  or  consists  of  separate  sequences  of  events  
assembled  so  as  to  constitute  a  viable  story.  For  example,  a  story  about  a  
murder  may  start  with  the  killing  and  then  look  back  over  the  events  that  led  
up  to  it,  to  “explain  it”  as  a  “whydunnit”  or  progress  from  the  killing  event  
forward  to  discover  the  killer  and  unravel  the  “whodunnit”.  Or  like  
Rashomon  (1950,  Akiro  Kurosawa)  may  tell  the  story  from  the  perspective  of  
different  participants  involved  in  the  event.  In  this  case  the  same  essential  
story,  that  of  a  murder,  is  organized  by  the  writer  in  a  different  way  by  using  
a  different  narrative  approach.    

By  definition,  narrative  is  about  a  sequential  succession  of  events.  Metz  

(1974:24)  says  ‘A  narrative  is  a  sum  of  events;  it  is  these  events  which  are  
ordered  into  a  sequence…  A  narrative  is  not  a  sequence  of  closed  events  but  
a  closed  sequence  of  events.’    According  to  Rimmon-­‐Kenan  (citing  Chatman  
and  Barthes  2002:16):  

Events  can  be  classified  into  two  main  kinds:  those  that  advance  the  
action  by  opening  an  alternative  (‘kernels’)  and  those  that  expand,  
amplify,  maintain  or  delay  the  former  (‘catalysts’).  

Stewart  McKie   -­‐  133  -­‐   As  of  15/08/2014  

She  goes  on  to  state  that:  

Structural  descriptions  show  how  events  combine  to  create  micro-­‐

sequences  which  in  turn  combine  to  form  macro-­‐sequences  which  
jointly  create  the  complete  story.    

In  the  case  of  Rashomon,  it  is  remembrances  from  a  specific  point-­‐of-­‐view  of  
a  single  event,  that  create  the  micro-­‐sequences  that  in  turn  create  the  macro-­‐
sequence  of  the  complete  story  as  constructed  from  the  different  
perspectives.  

She  cites  temporal  succession  and  causality  as  the  two  main  principles  of  
combination  and  defines  a  storyline  as  a  succession  of  events  restricted  to  
one  set  of  individuals.  In  general  screenplay  terms,  her  micro-­‐sequences  can  
be  mapped  to  scenes  and  macro-­‐sequences  to  sequences  of  scenes;  and  a  
specific  character’s  participation  in  a  storyline  represents  part  (or  all)  of  
their  narrative  throughline,  which  itself  functions  as  a  macro-­‐sequence.    

So  can  story  events  be  formalized  or  functionalized?  According  to  Propp,  in  
the  world  of  the  Russian  folktale,  they  can.  Metz  considers  Propp  to  be  a  
structuralist  interested  in  structural  analysis  and  comments  (1974:16)  that  
‘it  might  be  said  that  the  main  interest  of  structural  analysis  is  only  in  being  
able  to  find  what  was  already  there,  of  accounting  with  much  more  precision  
for  what  a  naïve  consciousness  had  “picked  up”  without  analysis.’  Propp’s  
Morphology  of  the  Folktale  calls  the  constant  elements  of  a  folktale  
“functions”  (1968:21)  and  states  that,  ‘Function  is  understood  as  an  act  of  a  
character,  defined  from  the  point  of  view  of  its  significance  for  the  course  of  
the  action.’  Propp  proposed  that  these  functions  could  be  combined  into  31  
folktale  patterns.  But  this  rather  mechanistic  analysis  was  later  challenged  
by  Bremond’s  (1966:75)  existentialist,  process-­‐centric  perspective  based  on  
the  idea  that  three  functions  combine  to  form  a  sequence  of  logical  steps:  
potentiality,  process  and  outcome.  This  potential  could  result  in  either  an  
actualization  or  non-­‐actualization  process  and  the  outcome  of  the  
actualization  process  result  in  success  (improvement)  or  failure  
(deterioration).      

Stewart  McKie   -­‐  134  -­‐   As  of  15/08/2014  

Greimas  (Structural  Semantics  (1966)  referenced  in  Edgar  2009:22)  also  
applied  reductive  techniques  to  characters.  He  defined  six  plot  functions  or  
‘actants’:  Sender/receiver,  subject/object  and  helper/opponent.  Actants  
truly  represent  ‘roles’  that  carry  out  a  function  in  the  plot  rather  than  
individual  characters  i.e.  the  role  has  an  arc  rather  than  the  character  
representing  the  role.  On  this  basis,  Greimas  found  three  plot  forms  in  
folktales:  Contractual,  performative  and  disjunctive.  All  this  indicates  that  
structural  analysis  of  creative  works  has  a  long  pre-­‐Fieldian  history.  

The  fundamentals  of  screenplay  narrative  analysis  assume  the  following  can  
be  easily  identified  from  a  screenplay.  

• The  extent  (start  and  end)  of  the  screenplay.  

• What  plots/sub-­‐plots/storylines  interweave  to  create  the  

narrative  line.  

• When  these  storylines  start  and  end  (their  extent).  

• What  sequences  of  scenes  are  represented  in  the  narrative  as  a  
whole,  a  plot  or  sub-­‐plot.  

• Where  plot  intersects  with  scenes  that  are  part  of  the  throughline  
of  specific  characters  in  the  screenplay.  

Plots,  storylines  and  sequences  are  not  usually  “flagged”  in  the  text  of  the  
screenplay;  in  fact  in  initial  drafts  a  screenwriter  may  not  even  be  aware  
they  exist  –  only  identifying  them  and  rejecting  or  refining  them  in  the  
revision/rewrite  process  when  some  kind  of  structure  is  applied  to  the  text  
manually  or  surfaced  from  the  data  automatically.  Therefore  identifying  
plots  and/or  sequences  within  a  script  requires  some  work  by  the  human  
analyst  or  software  analytical  engine,  first  to  identify  them  and  then  to  
visualize  them.  

Linear  and  Non-­‐Linear  

Stewart  McKie   -­‐  135  -­‐   As  of  15/08/2014  

Most  screenplays  have  a  linear  narrative,  although  non-­‐linear  narrative  
became  fashionable  following  the  critical  and  box-­‐office  success  of  
Tarantino’s  Pulp  Fiction  (1994).  Linear  narratives  follow  a  sequential,  time-­‐
based  start-­‐to-­‐end  flow  (or  end-­‐to-­‐start  flow)  whereas  a  non-­‐linear  
narrative  jumbles  up  the  narrative  so  that  the  narrative  may  jump  from  past,  
to  present,  to  future  in  a  non-­‐sequential  way  (e.g.  The  Machinist  (2004,  Brad  
Anderson)).  However,  a  linear  narrative  may  also  include  either  flashback  
and  flash-­‐forward  scenes  or  sequences  that  disturb  but  do  not  essentially  
alter  the  linearity  of  the  narrative.  For  example  Titanic  (1997)  that  
comprises  largely  one  big  flashback,  bookended  by  scenes  located  in  the  
present.  But  a  non-­‐linear  narrative  does  not  have  to  start  at  the  beginning  of  
the  narrative  in  question  e.g.  Memento,  (2000,  Christopher  Nolan)  and  may  
backtrack  or  jump  ahead  in  ways  that  disrupt  the  sequential  flow  of  a  
traditional  linear  narrative.    

Non-­‐linear  narrative  form  is  becoming  more  important  as  the  ‘gamification’  
of  stories,  to  create  online  interactive  narratives  that  leverage  the  full  
potential  of  in-­‐story  hyperlinking,  and  interactive  documentaries  or  ‘i-­‐docs’  
become  more  popular.  Tools  that  support  the  creation  of  non-­‐linear  
narratives  like  Florian  Thalhofer’s  Korsakow  system  (korsakow.org),  Mozilla  
Popcorn  Maker  (popcorn.webmaker.org)  and  Storyplanet  (storyplanet.com)  
for  example,  enable  creatives  to  combine  text,  links,  images  and  audio/video  
clips  in  ways  that  call  into  question  the  traditional  concept  of  a  screenplay  or  
even  the  need  for  one  at  all.  In  effect  the  screenplay  is  not  used  to  drive  the  
production  of  the  interactive  narrative  but  needs  to  extrapolated  from  it,  
after-­‐the-­‐fact.  

However,  visualizing  the  narrative  line  in  linear  narratives  is  useful  because  
it  may:    

• indicate  gaps  or  breaks  in  the  story    

• expose  narrative  overlap  or  lack  of  it  

Stewart  McKie   -­‐  136  -­‐   As  of  15/08/2014  

 • highlight  scenes  that  need  to  be  moved,  removed  or  added  

Any  of  which  may  cause  an  audience  to  become  confused  or  irritated  by  the  
resulting  movie,  reducing  the  potential  for  positive  word-­‐of-­‐mouth  
recommendations.  

Identifying  

The  narrative  line  is  identified  by  the  order  in  which  the  scenes  in  the  
screenplay  are  presented.  Individual  scenes  may  be  part  of  a  single  narrative  
line  or  multiple  narrative  lines.  But  in  the  latter  case,  it  is  not  necessarily  
easy  to  identify  where  one  narrative  line  ends  and  another  begins  (or  
continues).  Although  this  may  be  more  evident  because  each  narrative  line  
takes  place  in  the  same  milieu  or  within  a  specific  timeslot  or  involves  a  
specific  set  of  characters  who  do  not  cross  narrative  lines  (e.g.  in  a  
portmanteau  movie).  

Visualizing  

Visualizing  a  linear  narrative  is  relatively  straightforward  to  analyze  and  

visualize.  Often  the  analysis  merely  needs  to  identify  scenes  linked  to  a  role,  
sequence  or  storyline  and  visualization  takes  the  form  of  scenes  listed  
sequentially  along  a  timeline  that  references  the  anticipated  running  time  of  
the  movie  based  on  the  page  count  of  the  screenplay.  If  there  are  multiple  
narrative  lines  then  these  can  be  visualized  using  “swim  lanes”  whereby  
each  narrative  line  and  the  scenes  that  represent  it  are  contained  within  
their  own  lane.  

But  any  visualization  becomes  more  complex  when  the  screenplay  involves  
multiple  narrative  lines  organized  non-­‐linearly  or  makes  extensive  use  of  
flashback/flash-­‐forward  over  the  course  of  the  screenplay.  In  these  cases  
there  may  be  multiple  narrative  “strands”,  which  may  overlap  each  other  
and  narrative  “jumps”  both  backwards  and  forwards.    

Figure  2.12  shows  Pope’s  (1998:44)  analysis  of  Pulp  Fiction  with  multiple  

Stewart  McKie   -­‐  137  -­‐   As  of  15/08/2014  

character  throughlines  being  reorganized  by  director  Tarantino  to  
emphasize  rising  emotional  tension  rather  than  the  linear  way  that  events  
actually  unfolded.  

Figure  2.12  -­‐  Pulp  Fiction  Narrative  Lines  

Figure  2.13  shows  Aronson’s  (2001:113)  analysis  of  the  narrative  structure  
of  Shine  involving  the  interlocking  connections  between  the  “film  in  the  
present”  and  the  “film  in  the  past”.  

Stewart  McKie   -­‐  138  -­‐   As  of  15/08/2014  

  

Figure  2.13  -­‐  Flashback  as  Thwarted  Dream:  Shine  

Plot  Points  

Many  screenwriting  “gurus”  advocate  the  use  of  defined  plot  points  such  as  
“turning  points”,  “inciting  incidents”  “midpoints”  and  “climaxes”  to  
punctuate  and  guide  the  screenplay  narrative.  These  plot  points  often  signal  
a  significant  action  or  change  in  direction  within  the  narrative.  Apparently  
some  script  readers  actually  look  for  these  plot  points  and  expect  them  to  
occur  on  or  around  specific  pages  in  a  script.  So  a  visualization  of  these  plot  
points  would  essentially  present  this  information  automatically  before  the  
script  is  read,  saving  valuable  script-­‐reading  time  to  find  them  manually  
(except  of  course  if  readers  are  so  focused  on  this  approach  that  they  simply  
skip  to  certain  pages  in  a  script  to  find  them).  

Stewart  McKie   -­‐  139  -­‐   As  of  15/08/2014  

Identifying  

Currently  screenwriting  programs  generally  do  not  provide  a  way  to  identify  
plot  points  in  the  script.  So  again,  identifying  the  plot  points  may  require  the  
writer  to  specifically  tag  them  in  the  content  either  as  they  are  written  or  
after  the  fact,  in  the  absence  of  screenplay  parsing  software  sophisticated  
enough  to  find  these  plot  points  and  tag  them  automatically  as  one  kind  of  
screenplay  metadata.  

Visualizing  

Plot  points  can  be  visualized  as  a  series  of  peaks  and  troughs,  rollercoaster  
fashion.  One  might  expect  at  least  3  such  peaks  in  the  average  3-­‐act  movie  
namely:  The  inciting  incident  (in  act  one),  the  mid-­‐point  (in  act  two)  and  the  
climax  (in  act  three).  The  kind  of  structure  reflected  in  Freytag’s  Pyramid  
(see  figure  2.1)  where  (a)  is  the  introduction,  (b)  is  the  rise,  (c)  is  the  climax,  
(d)  is  the  return  or  fall  and  (e)  is  the  catastrophe.  The  difficulty  is  identifying  
them,  without  the  writer  signposting  them  by  explicitly  tagging  their  
occurrence  in  the  script.    

2.5  Structure  

Acts,  Sequences,  Scenes  and  Beats  

All  screenwriters’  practice  scene  writing  and  many  will  break  a  scene  down  
into  “beats”  and/or  combine  scenes  into  “sequences”  and  “acts”.  These  may  
be  regarded  as  the  core  structural  components  of  the  screenplay  narrative  
and  represent  a  hierarchy  of  increasing  detail  such  as:  

3-­‐5  acts  comprising  

     10-­‐20  sequences  comprising  

             50-­‐100  scenes  comprising  

                   150-­‐300  beats  

Stewart  McKie   -­‐  140  -­‐   As  of  15/08/2014  

Some  writers  may  go  as  far  as  to  write  scenes  where  each  scene  itself  has  an  
identifiable  beginning,  middle  and  end  and  could  also  employ  this  triad  
structure  again  when  composing  sequences  of  scenes  to  represent  a  logical  
dramatic  unit  within  the  screenplay.  In  this  sense  the  whole  screenplay  
becomes  an  example  of  a  fractal  design.  

Identifying  

Whilst  scenes  are  always  identified  in  a  screenplay  –  by  means  of  the  scene  
heading  or  slug  line  -­‐  acts,  sequences  and  beats  are  not.  Unless,  that  is,  the  
screenwriting  package  used  by  the  writer  provides  the  ability  to  organize  
scenes  and  the  content  of  a  scene  in  this  way.  So  identifying  the  scope  of  
these  structural  components  may  require  the  writer  to  specifically  tag  them  
in  the  content  either  as  they  are  written  or  after  the  fact.  

Visualizing  

This  kind  of  visualization  is  better  suited  to  navigation  of  screenplay  content  
rather  than  analysis.  It  is  useful  to  see  the  “forest  and  the  trees”  to  navigate  
and  jump-­‐around  the  script  content  and  to  use  as  a  means  to  “shuffle”  and  
reorganize  the  script  content,  say  by  dragging  a  scene  from  one  sequence  
and  dropping  it  into  another.  It  is  also  useful  for  understanding  if  the  script  
is  “well-­‐balanced”  and  does  not  deviate  too  far  from  accepted  norms  –  a  
script  with  20  acts  and  500  scenes  would  be  perceived  as  unbalanced  for  
example.    

Balanced  Acts  

Proponents  of  the  three-­‐act  structure,  including  script  readers  who  are  
anecdotally  supposed  to  look  for  the  three-­‐act  structure  as  a  basis  for  
commercial  movie  scripts,  are  quite  exact  about  the  “balance”  between  the  
content  of  the  acts.  Field  (see  figure  2.12  below)  divides  a  screenplay  into  
roughly  one  quarter  (act  1),  two  quarters  (act  2)  and  one  quarter  (act  3).  In  
this  context  an  unbalanced  act  structure  may  make  a  screenplay  less  
commercially  acceptable.  

Stewart  McKie   -­‐  141  -­‐   As  of  15/08/2014  

Also  relevant  here  is  Yanno’s  concept  (2006:  15)  of  act  1  asking  a  question  
and  act  3  answering  it  -­‐  an  act-­‐based  setup  and  payoff.  If  act  1  asks  a  
question  that  act  3  does  not  answer  or  if  act  3  answers  a  question  that  act  1  
does  not  ask,  the  screenplay  may  also  be  unbalanced.  

Identifying  

Balance  depends  on  the  ability  to  detect  act  breaks  that  are  unlikely  to  be  
specifically  defined  by  the  screenwriter  in  the  script.  The  question/answer  
balance  depends  on  identification  of  attribute  in  the  part  of  the  script  
identified  as  act  1  that  can  be  found  to  have  been  referenced  again  somehow  
in  the  part  of  the  script  identified  as  act  3.  

Visualization  

 A  linear  representation  of  script  balance  seems  the  best  solution  to  show  
balance  between  the  three  acts  but  visualizing  the  question  and  
answer/setup  and  payoff  balance  between  act  1  and  act  3  is  more  difficult.  
One  simple  option  is  a  two  column  display  headed  with  potential  questions  
listed  in  the  left  column  (act  1)  and  potential  answers  listed  in  the  right  
column  (act  3).  Questions  with  no  answers  or  answers  with  no  questions  
would  then  be  easy  to  identify,  allowing  the  writer  to  revisit  the  appropriate  
act  and  rewrite  to  put  the  screenplay  back  in  balance.    

Scenes  –  Similarity  and  Difference  

The  scene  is  the  key  structural  entity  within  a  screenplay  because  it  can  both  
be  decomposed  into  shots  and  aggregated  into  sequences  or  storylines.  An  
interesting  approach  to  scene  analysis  is  to  look  for  similarities  and  
differences  between  a  set  of  scenes,  or  all  scenes  in  the  screenplay,  and  what  
this  might  tell  us.  Scene  analysis  can  refer  to  either  data  or  metadata  or  both.  
One  obvious  similarity/difference  is  whether  a  scene  is  internal  (INT)  or  
external  (EXT)  and  whether  it  is  DAY  or  NIGHT  -­‐  others  include:  

• Do  the  scenes  take  place  in  the  same  location?  

Stewart  McKie   -­‐  142  -­‐   As  of  15/08/2014  

 • Do  the  same  set  of  characters  take  part  in  the  scenes?  

• Does  one  specific  character  appear  in  more/less  scenes  than  other  
characters?  

• Are  the  scenes  action  only  or  action  and  dialog?  

• Is  the  linguistic  “footprint”  of  the  scenes  similar?  

Consider  Ridley  Scott’s  Alien  (1979)  in  the  light  of  the  above.    

• Most  scenes  take  place  in  the  spaceship  except  around  the  beginning  
(on  the  beacon-­‐sending  planet)  and  at  the  end  (in  the  escape  shuttle).  
These  differences  may  be  considered  important  in  leading  to  and  
from  the  core  “middle”  of  the  movie.  The  propensity  of  scenes  in  the  
same  place  is  also  something  that  is  likely  to  be  characteristic  of  a  
specific  genre  e.g.  horror  or  a  sitcom.  

• Many  scenes  that  involve  Ripley  (Sigourney  Weaver)  peripherally  

involve  other  crewmembers  being  killed  by  the  Alien;  whereas  Ripley  
is  present  in  the  majority  of  scenes.  From  this  one  might  deduce  that  
she  is  probably  the  protagonist,  as  she  not  only  has  a  throughline  
from  the  beginning  of  the  movie  until  the  end,  she  also  has  a  “lifeline”.  

• Many  scenes  that  involve  Ripley  and  one  other  character  –  even  if  
that  character  is  “mother”  the  ship’s  computer  –  are  usually  scenes  
where  Ripley  challenges  the  other  character  in  some  way.  Again,  
indicative  of  protagonist  (or  antagonist)  behavior.  

• In  the  last  third  of  the  movie,  Ripley  appears  in  many  “action-­‐only”  
scenes  –  unsurprising,  as  the  rest  of  the  crew  is  dead.  But  this  is  a  big  
difference  from  earlier  scenes  that  are  relatively  dialog-­‐heavy  as  the  
whole  crew  discuss  their  situation  and  argue  about  how  to  kill  the  
monster.  By  identifying  the  change  in  scene  type,  we  can  locate  some  
kind  of  “tipping  point”  that  probably  occurs  around  the  time  that  the  

Stewart  McKie   -­‐  143  -­‐   As  of  15/08/2014  

 last  crewmember  is  killed  by  the  Alien.    

• The  linguistic  footprint  of  the  scenes  up  until  the  last  crewmember  
dies  may  possibly  indicate  rising  tension  or  fragmentation  of  the  
team  under  pressure.  Perhaps  this  could  be  reflected  in  more  
statements  than  questions,  more  expletives,  more  action  than  dialog.  

Wherever  the  Alien  appears  in  a  scene,  death  often  occurs  also.  This  could  be  
indicative  of  an  antagonist  role,  especially  as  towards  the  end  Ripley  
(possible  protagonist)  and  the  Alien  (possible  antagonist)  are  essentially  the  
only  characters  left  in  the  screenplay  (apart  from  the  cat  that  is).  

2.6  Analytic  Archetypes  

Significant  academic  attention  has  been  focused  on  using  “intangible”  

screenplay  text  as  an  input  to  automatically  generate  some  kind  of  “tangible”  
animated  output  or  as  the  raw  material  to  auto-­‐manufacture  a  computer-­‐
generated  visual  product.  

Zhi-­‐Qiang  Liu’s  ScriptViz  takes  well-­‐formed  sentences  from  a  script  as  input  
and  converts  these  into  agents  and  plans  for  those  agents  that  are  then  
rendered  by  a  “scene  generator”  in  the  form  of  computer-­‐generated  
animations  as  the  story  evolves.  Coyne  and  Sproat’s  WordsEye  system  
converts  text  input  into  scenes  by  tagging  and  parsing  sentences  to  convert  
them  into  a  dependency  structure  that  is  converted  into  a  semantic  
representation.  This  representation  is  subject  to  depiction  rules  that  
generate  a  series  of  depicters  to  which  transduction  rules  are  applied  to  
create  the  refined  depicters  that  are  used  to  manipulate  the  3D  objects  used  
to  finally  render  the  scene  to  the  user  as  a  computer  generated  animation.    

Ramirez  proposed  using  a  film  language  (FL)  based  on  Arijon’s  Grammar  of  
the  Film  Language  (1976)  to  generate  an  ontology  from  a  given  script  input  
that  could  then  be  interrogated  by  a  reasoning  engine  as  the  basis  for  
analysis  and  generation  of  a  computer-­‐generated  cinematic  version  of  the  
story.    Whereas  Johsi,  Wang  and  Li’s  Story  Picturing  Engine  is  more  of  an  

Stewart  McKie   -­‐  144  -­‐   As  of  15/08/2014  

automated  storyboarding  application  that  first  identifies  keywords  in  the  
text,  then  searches  an  annotated  image  database  for  suitable  images  to  
match  with  the  text  and  then  determines  which  images  are  the  “elite”  images  
to  illustrate  a  story  using  a  mutual  reinforcement  ranking.  

The  rise  of  computer  gaming  as  an  interactive  rather  that  traditional  
narrative  has  focused  attention  on  interactive  storytelling  and  ways  of  
generating  visually  interactive  products  from  texts  or  using  existing  stories  
and  narrative  paths  to  generate  new  stories  or  alternative  narrative  paths  
within  a  story.  

Gervás,  Díaz-­‐Agudo,  Peinado  and  Hervás  utilized  case-­‐based  reasoning  

based  on  an  ontology  that  leverages  Propp’s  Morphology  of  the  Folktale,  to  
generate  a  new  plot  sketch  based  on  the  plot  structures  extracted  from  
existing  stories  submitted  by  the  user.  

However,  surprisingly  few  screenwriting  manuals  either  advocate  or  

use  analytics  and  visualization  as  a  means  to  improve  a  screenplay.  But,  as  
suggested  above,  even  the  limited  visualizations  of  screenplay  content  that  
are  presented  are  almost  certainly  not  generated  from  analyzing  the  script  
itself  (the  input)  but  reverse-­‐engineered  from  watching  the  movie  that  
resulted  from  the  script  (the  output).    

A  review  of  screenwriting  literature  indicates  that  there  is  only  a  handful  of  
content  visualization  archetypes  represented,  defined  here  as:  Linear,  
Circular,  Nodal  and  Rollercoaster.  Samples  of  each  archetype  are  displayed  
and  discussed  below.    

Linear  Archetype    

The  linear  archetype  can  help  to  visualize  the  narrative  structure  and  the  
chronology  of  the  screenplay.  Linear  visualizations  are  useful  for  viewing  
timelines  and  throughlines  or  the  narrative  threads  that  relate  to  characters  
or  plots/subplots  in  the  screenplay.  

Stewart  McKie   -­‐  145  -­‐   As  of  15/08/2014  

The  simple  example  in  figure  2.14  below  visualizes  Field’s  (2005:21)  3-­‐act  
“paradigm”  against  the  timeline  of  a  120-­‐minute  feature-­‐film.  Field  maps  
Aristotle’s  Beginning,  Middle  and  End  structure  to  his  own  Set-­‐Up,  
Confrontation  and  Resolution.  He  even  provides  suggested  page  extents  for  
each  of  the  three  acts.  

Figure  2.14  -­‐  Field’s  Three-­‐Act  Structure  

Figure  2.15  below  indicates  how  the  linear  archetype  can  also  be  used  to  
visualize  parallel,  converging  or  intersecting  "throughlines"  as  in  the  
screenplay  of  The  Day  of  the  Jackal  (1973,  Fred  Zinnemann)  that  help  to  
generate  the  tension  that  reflects  both  the  ‘closing  of  the  net’  around  the  
Jackal  and  the  Jackal  getting  ever  closer  to  his  assassination  target.    

Stewart  McKie   -­‐  146  -­‐   As  of  15/08/2014  

  

Figure  2.15  -­‐  Converging  Plot  Lines  in  The  Day  of  the  Jackal  (Pope,  1998:173)  

Linear  archetypes,  that  structure  screenplays  into  acts  and  sequences  are  
not  well  represented  in  today’s  screenwriting  software,  perhaps  indicating  
that  this  archetype  is  not  considered  to  have  a  practical  use  for  writers.  

Circular  Archetype  

The  circular  archetype  is  less  suitable  for  chronological  timelines  and  more  
suitable  for  visualizing  emotional  or  psychological  "journeys"  with  a  holistic  
dimension  that  includes  a  return  to  the  start,  implying  psychological  
resolution  or  re-­‐integration.  Homer’s  Odyssey  is  the  story  archetype.  The  
character  whose  journey  the  circle  reflects  may  in  fact  remain  in  or  around  
the  same  physical  place  throughout  this  kind  of  journey.  Each  stage  of  the  
journey  could  "map"  to  a  particular  sequence  of  scenes,  or  a  key  individual  
scene  in  the  screenplay.  The  circular  archetype  is  more  suitable  for  
visualizing  character  development  and  is  exemplified  by  Vogler's  Hero's  
Journey  (from  his  1992  The  Writer’s  Journey  see  figure  2.16),  which  may  
have  been  borrowed  from  Murdock's  1990  The  Heroine's  Journey  (see  figure  
2.17).  

Stewart  McKie   -­‐  147  -­‐   As  of  15/08/2014  

  

Figure  2.16  -­‐  Vogler’s  Hero’s  Journey  (Vogler,  1994:194)  

Figure  2.17  -­‐  Murdock’s  Heroine’s  Journey  (Murdock,  1990:5)  

Some  useful  examples  of  using  the  hero’s  journey  version  of  the  circular  
archetype  for  movie  analysis  are  found  in  Voytilla  (1999).  Figure  2.16  shows  
his  analysis  of  the  movie  Die  Hard  (1998),  which  maps  plot  points  onto  the  

Stewart  McKie   -­‐  148  -­‐   As  of  15/08/2014  

timeline  of  the  movie  to  show  how  McClane  moves  from  his  ordinary  world  
to  the  special  world  of  the  terrorists  and  back  to  his  ordinary  world.  

Figure  2.18  -­‐  Die  Hard  Analysis  (Voytilla,  1999:41)  

Rollercoaster  Archetype  

The  “rollercoaster”  archetype  (see  figure  2.19a  below)  reflects  that  fact  that  
the  narrative  of  a  screenplay  often  delivers  a  series  of  peaks  (or  emotional  
climaxes)  and  troughs  (or  releases  from  tension)  for  the  movie  audience;  
involves  creating  escalating  tension  that  demands  resolution  in  a  satisfying  
climax;  and  as  the  movie  progresses,  more  story  information  is  provided  and  
the  characters  developed  so  that  the  "bulk"  of  the  movie  "mountain"  
increases  until  it  reaches  its  “summit”  at  the  climax  of  the  movie.  Cooper  
(1997)  uses  the  rollercoaster  term  whereas  Aronson  (2001:42)  prefers  the  
term  “mountain”  and  Karetnikova  (1990:32)  simply  states,  “It  is  always  
helpful  to  view  the  development  of  a  plot  graphically  as  a  kind  of  triangle,  in  
which  the  angles  represent  the  opening,  the  climax  and  the  resolution.”    

Stewart  McKie   -­‐  149  -­‐   As  of  15/08/2014  

Throughout  her  book,  Cooper  discusses  a  variety  of  rollercoaster  examples  
that  offer  an  interesting  way  to  visualize  plot  points  in  particular,  with  both  
the  “peaks”  and  “troughs”  providing  a  useful  way  to  indicate  “highs”  and  
“lows”  in  the  flow  of  the  script  and  the  bulk  of  the  script  text  providing  the  
means  of  showing  the    “information  mountain”  contained  in  the  script  
building  over  time  in  the  mind  of  the  audience.  She  also  discusses  the  use  of  
rollercoasters  as  a  tool  for  writers  to  proactively  design  a  specific  kind  of  
narrative  experience  for  the  audience.  

Figure  2.19a  -­‐  Rollercoasters  of  Escalating  Tension  (Cooper,  1997:71)  

However,  as  Aronson  herself  points  out  (2010:  293)  the  mountain  
visualization  can  break  down  when  applied  to  characters  within  movies  

Stewart  McKie   -­‐  150  -­‐   As  of  15/08/2014  

with  an  unconventional  structure  such  as  Paul  and  Jack’s  stories  in  21  Grams  
(2003,  Alejandro  González  Iñárritu),  resulting  in  a  structure  that  she  terms  a  
‘fractured  tandem’.  

Figure  2.19b  -­‐  Paul  and  Jack’s  Fractured  Tandem  in  21  Grams  (Aronson,  2010:293)  

Nodal  Archetype  

The  nodal  archetype  offers  a  number  of  implementations.  A  common  form  is  
a  top-­‐down  or  left-­‐to-­‐right,  tree-­‐and-­‐branch  visualization  to  show  
hierarchies  or  parent-­‐child  dependencies  (vertical  or  horizontal  
dendrograms).  For  example,  figure  2.20  shows  an  “organization  chart”  view  
of  the  tree  structure  of  a  screenplay  to  visualize  the  relationship  within  a  
single  script  between  acts,  sequences,  and  scenes.  This  archetype  is  also  
useful  as  a  way  to  visualize  parallel  or  multiple  storylines,  especially  where  
these  storylines  are  nested  within  each  other.  Each  storyline  becomes  a  
branch  with  its  own  set  of  dependent  nodes  that  may  or  may  not  be  nested  
within  another  branch.  Equally  the  branches  could  be  literally  that  in  a  
dendrogram  of  an  interactive  narrative  where  each  branch  represents  a  
different  path  through  the  narrative  that  the  viewer/player  can  take.    

Stewart  McKie   -­‐  151  -­‐   As  of  15/08/2014  

  

Figure  2.20  -­‐  Screenplay  Structure  as  Node  Tree  (Kitchen,  2006:162)  

Another  nodal  implementation  relates  to  character  analysis,  whereby  each  

character  acts  as  a  node  in  the  relationship  network  of  the  script’s  
characters.  Figure  5.21  shows  a  visualization  of  the  character  relationships  
or  social  network  of  The  Third  Man  (1949).  

Stewart  McKie   -­‐  152  -­‐   As  of  15/08/2014  

Figure  2.21  -­‐  Characters  in  The  Third  Man  (MacKendrick,  2005:19)  

2.7  Generic  Data  Analysis  and  Visualization  Tools  

The  screenplay  visualization  examples  above,  drawn  largely  from  

screenwriting  ‘how-­‐to’  manuals,  were  almost  certainly  all  created  manually  
by  the  authors  -­‐  as  opposed  to  by  clicking  a  button  in  a  screenwriting  
application.  But  there  are  generic  analysis  and  visualization  tools  available  
that  can  generate  interesting  visualizations  when  provided  with  screenplay  
content  as  a  data  input.    

There  are  dozens  of  chart  formats  that  could  be  used  to  display  screenplay  
as  data  content  visually  (see  http://mbostock.github.io/protovis/ex/  -­‐  
accessed  11  June  2013)  for  over  50  chart  types  that  can  be  produced  from  
the  Protovis  tool.  This  section  discusses  some  software  tools  that  have  been  
designed  for  generic  data  analysis  and/or  visualization  that  interesting  to  
apply  specifically  to  screenplay  content.  The  first  three  of  these  
visualizations  utilize  tools  are  found  on  Jeff  Clark’s  web  site  at  
www.neoformix.com.  The  example  visualizations  are  produced  from  my  
own  script  Carpini.  

Arc  Diagram  

According  to  Clark:    

Document  Arc  Diagrams  illustrate  the  similarity  structure  within  a  text  

document  by  drawing  arcs  connecting  segments  of  a  document  that  
share  similar  vocabulary…I  have  written  before  about  Martin  
Wattenberg's  Arc  Diagrams  for  visualizing  structure  within  strings.  
They  are  an  intriguing  way  of  visualizing  repetition  at  varying  scales  
within  a  linear  sequence.  

Clark  explains  the  process  of  generating  an  arc  diagram  as  follows:  

1. Break  the  document  up  into  a  stream  of  words.  

2. Throw  away  any  'stop  words'  (a,  at,  of,  the...).  

Stewart  McKie   -­‐  153  -­‐   As  of  15/08/2014  

 3. Divide  the  remaining  stream  of  more  interesting  words  into  50  equal  
segments  based  on  linear  position.  

4. Calculate  a  similarity  metric  between  each  pair  of  segments  based  on  
the  amount  of  overlapping  words.  

5. Draw  a  diagram  where  the  document  segments  are  connected  by  arcs  
with  the  transparency  determined  by  the  similarity  between  the  
segments.  Use  a  threshold  so  that  weakly  connected  arcs  don't  get  
drawn  at  all.  

6. Show  the  top  two  words  for  each  arc  drawn  at  both  segment  endpoints.  

This  visualization  could  be  more  interesting  when  applied  to  a  screenplay,  
assuming  the  following  modifications:  

1. If  the  squares  (see  step  3  above)  represented  each  scene  in  the  script,  
rather  than  an  arbitrary  50  segments,  the  diagram  would  have  a  
more  screenplay  friendly  structure.  

2. If  the  character  names  included  in  each  scene  and/or  scene  headings  
were  listed  against  each  square  consistently,  this  would  be  more  
useful.  

3. If  the  text  included  could  be  selected  as  just  dialog  elements,  just  
action  elements  or  both,  rather  than  all  text  it  might  be  more  
interesting.  

If  the  language  content  could  be  defined  in  some  way  e.g.  conflict  language,  
the  arcs  could  be  colored  to  reflect  this  and  used  to  connect  scenes  of  similar  
linguistic  type.  I  used  one  of  my  own  scripts,  Carpini,  to  convert  into  a  
suitable  data  input  in  order  to  generate  the  three  diagrams  shown  in  figure  
2.22-­‐2.24  below.    Combinations  of  character  names  are  listed  to  the  left  of  
the  ‘scene  square’  and  the  curves  represent  links  between  scenes  deemed  to  
be  ‘similar’  (the  orange  line  indicates  a  stronger  similarity  than  the  blue  

Stewart  McKie   -­‐  154  -­‐   As  of  15/08/2014  

line).  

Figure  2.22  -­‐  Arc  Diagram  Applied  to  my  Screenplay  Carpini  

Topic  Flower  

The  topic  flower  presents  an  interesting  and  artistic  tonal  impression  of  a  
body  of  text  that  reflects  the  occurrence  of  certain  defined  categories  of  
language  in  the  text.  Here  the  topic  flower  is  designed  to  reflect  the  content  
Stewart  McKie   -­‐  155  -­‐   As  of  15/08/2014  
in  terms  of  the  relative  weight  of  language  relating  to  Art,  Recreation,  
Society,  Technology,  Science  and  Economy  –  categories  clearly  more  
relevant  to  news  or  factual  content  rather  than  a  screenplay.  

Clark  explains  the  process  of  generating  a  topic  flower  visualization  as  
follows:  

1. The  same  text  will  always  generate  the  same  flower.  

2. More  text  will  generate  more  layers  of  petals.  

3. The  primary  topic  will  be  shown  using  the  associated  colour  on  the  
outermost  two  layers  of  petals.  

4. If  there  is  a  secondary  topic  it  will  be  shown  on  the  third  layer  of  petals.  
This  pattern  repeats,  two  layers  using  the  primary,  then  one  with  the  
secondary.  

5. If  there  exists  a  tertiary  topic  its’  colour  is  used  to  accent  the  edges  of  
some  of  the  primary  coloured  petals.  

6. The  number  of  little  'hairs'  on  the  flower  is  indicative  of  the  number  of  
personal  pronouns  used  in  the  text.  

7. Rounder  petal  shapes  are  suggestive  of  emotionally  positive  terms  

(love,  yes,  peace),  and  more  elongated  terms  indicate  negative  terms  
(death,  murder,  idiot).  

This  visualization  could  be  interesting  if  applied  to  a  screenplay,  assuming  
the  following  modifications:  

1. If  the  topics  were  changed  to  reflect  more  screenplay  relevant  factors  
such  as  emotion  or  sex  or  violence  or  conflict  –  the  linguistic  categories  
would  be  more  relevant.  

Stewart  McKie   -­‐  156  -­‐   As  of  15/08/2014  

  

Figure  2.23-­‐  A  Topic  Flower  Applied  to  a  Screenplay  

Double  Document  Shared  Word  Diagrams  

 
According  to  Clark:    

Double  Document  Shared  Word  Diagrams  compare  and  contrast  two  

documents  by  showing  both  the  unique  and  shared  vocabulary  and  it’s  
distribution  across  the  two  documents  of  interest.    

The  two  columns  of  squares  represent  the  two  documents.  The  leftmost  
column  of  word  circles  shows  the  highest  frequency  non-­‐trivial  words  found  
in  document  1  but  not  document  2.  The  rightmost  column  of  word  circles  
shows  those  words  unique  to  2  and  the  central  column  shows  the  words  that  
are  common  to  both.  

This  visualization  could  be  interesting  if  applied  to  a  screenplay,  assuming  
the  following  modifications:  

1. If  the  squares  represented  scenes  in  the  script,  the  diagram  would  have  
a  more  screenplay  friendly  structure.  

2. If  character  names  common  to  both  drafts  were  listed  exclusively  in  the  
centre  column,  this  would  be  more  useful.  

Stewart  McKie   -­‐  157  -­‐   As  of  15/08/2014  

 3. If  screenplay  specific  noise  words  such  as  “INT”  and  “EXT”  were  
filtered  out  automatically.  

 
 

Figure  2.24  -­‐  A  Shared  Word  Diagram  Applied  to  Two  Drafts  of  a  Screenplay  

Dialog  Analysis  

Of  course  it  is  now  very  easy  to  search  any  digital  screenplay  for  specific  
words  or  phrases  to  locate  potential  product  placement  opportunities  or  to  
identify  profanity  and  sexual  terms  in  dialog,  which  could  impact  the  
potential  movie  rating.  But  dialog  analysis  is  also  interesting  in  the  context  
of  the  interplay  between  specific  characters  in  a  screenplay  or  in  relation  to  
what  is  realistic  for  the  time-­‐period  the  script  is  set  in.    

Stewart  McKie   -­‐  158  -­‐   As  of  15/08/2014  

Some  interesting  visualizations  have  been  produced  to  analyze  the  U.S.  
presidential  primaries  in  the  run  up  to  the  2008  election.  In  screenplay  
terms,  most  of  these  have  been  dialog-­‐centric,  that  is  they  have  focused  on  
the  spoken  content  of  the  candidate  debates  and  speeches.    

(http://www.nytimes.com/interactive/2007/12/15/us/politics/DEBATE.html accessed on 06 July2008)  

Figure  2.25  -­‐  Naming  Names  

Figure  2.25  was  produced  by  The  New  York  Times  (15  December  2007)  and  
shows  the  names  referenced  by  major  presidential  candidates  in  the  series  
of  Democratic  and  Republican  debates  leading  up  to  the  Iowa  caucuses.  
What  this  visualization  seems  to  show  is  that  at  this  stage  the  Republican  
and  Democrat  candidates  are  primarily  concerned  with  name-­‐checking  the  
rival  candidates  within  their  own  parties,  rather  than  attacking  the  
opposition.  This  is  in  fact  a  form  of  social  network  visualization  that  could  
equally  well  be  applied  to  a  selected  set  of  main  characters  in  a  screenplay  
based  on  the  name-­‐checking  of  other  characters  within  their  dialog  blocks  in  
the  script.  Mrs.  Clinton  provoked  the  largest  number  of  name  checks  
indicating  that  she  was  perhaps  considered  the  candidate  to  beat  at  that  

Stewart  McKie   -­‐  159  -­‐   As  of  15/08/2014  

time.  

Figure  2.26  -­‐  Transcript  Analyzer  

Jeff  Clark’s  visualization  (figure  2.26,  see  

http://www.neoformix.com/Projects/TranscriptAnalyzer/index.html  -­‐  
accessed  06  July  2008)  of  the  transcript  of  the  Democratic  debate  of  April  16  
2008  in  Pennsylvania  has  3  layers  of  content,  as  Clark  explains:

The  top  section  shows  the  distribution  of  some  selected  words  within  
the  text  across  a  'timeline',  which  goes  from  left  to  right.  Each  speech  
segment  is  the  same  width  and  the  height  of  the  small  white  bars  show  
the  number  of  occurrences  of  that  word  for  that  segment.  You  can  add  
new  words  with  the  text  box  in  the  top  right  corner.  You  can  remove  
existing  words  by  clicking  on  them.  Right  below  the  word  distribution  
graphs  is  a  similar  coloured  set  showing  a  spectral  decomposition  of  
the  text  based  on  who  spoke  and  how  much  was  said.  In  this  case  the  
bar  heights  give  the  amount  of  text  for  each  segment.  Click  and  drag  

Stewart  McKie   -­‐  160  -­‐   As  of  15/08/2014  

 the  mouse  left  to  right  to  move  along  the  timeline  and  show  the  actual  
text  for  3  consecutive  segments.  

This  kind  of  visualization  is  also  useful  for  screenplay  analysis.  The  timeline  
could  represent  the  running  order  of  a  series  of  scenes.  The  speakers  
become  characters  in  the  script,  the  text  bars  representing  dialog  blocks  and  
the  popup  text  the  actual  dialog  spoken  in  a  given  text  bar.    

Figure  2.27  –  Transcript  Analysis  

Figure  2.27  (see  

http://www.nytimes.com/interactive/2008/04/16/us/politics/20080416_
DEBATE_GRAPHIC.html#  -­‐  accessed  6  July  2008) is  another  New  York  Times  
visualization  from  April  16  2008  that  provides  an  analysis  of  the  
visualization  of  the  televised  democratic  debate  of  October  30th  2007.  It  
shows  the  word  count  and  speaking  times  of  each  participant  and  pattern  of  
their  contributions,  the  selected  speaker’s  dialog  is  coloured  in  blue  and  
replies  in  grey.  Clicking  one  of  the  blue/grey  blocks  reveals  the  actual  text  
from  the  transcript.  Again  this  could  be  used  as  is  on  a  screenplay  as  an  

Stewart  McKie   -­‐  161  -­‐   As  of  15/08/2014  

indication  of  the  size  of  a  characters  part  and  as  a  means  to  navigate  the  
character’s  dialog  throughout  the  script.    

Dialog  analysis  can  also  be  used  to  deliver  a  ‘reality-­‐check’  on  the  dialog  
content  of  a  script  by  identifying  anachronisms  that  are  out  of  place  in  the  
script  given  its  subject  matter  or  time  period  for  example.    Benjamin  
Schmidt,  a  fellow  at  the  Harvard  Cultural  Institute,  has  focused  attention  on  
uncovering  anachronisms  in  screenplays  such  as  those  by  Julian  Fellowes  for  
the  UK  TV  series  Downton  Abbey,  Tony  Kusner  for  Lincoln  (2012)  and  Chris  
Terrio’s  Argo  (2012).  Schmidt  uses  a  ‘big  data’  resource,  such  as  Google’s  
Ngrams  corpus,  to  compare  the  dialog  words  used  in  a  single  screenplay  to  
works  from  around  the  same  time  period  in  the  corpus  to  identify  words  or  
phrases  that  look  out  of  place  such  as  ‘peace  talks’  in  Lincoln  or  ‘lifetime  
achievement  award’  in  Argo.  Schmidt  also  makes  use  of  word  cloud  
visualizations  to  understand  how  well  a  script’s  dialog  ‘fits’  the  time  period  
as  shown  in  figure  2.28  for  Argo.  

Stewart  McKie   -­‐  162  -­‐   As  of  15/08/2014  

  

Figure  2.28  –  Argo  Word  Cloud  

(see  Nobody Said 'Racial Equality' in 1865: The Anachronistic English of

'Lincoln'    in  The  Atlantic  Jan.  10,  2013  and  
http://www.prochronism.com/2013/02/current-­‐events-­‐argo-­‐
anachronisms.html,  accessed  March  2013).  

Storyline  Analysis  

The  New  York  Times  also  produced  storyline  visualizations  (figures  2.29a-­‐c,  
see  http://www.nytimes.com/2005/04/24/magazine/24TV.html  -­‐  all  3  
figures  accessed  from  the  online  article  on  6  July  2008)  to  compare  the  
relative  narrative  complexity  of  selected  US  TV  shows  overtime.  In  the  
magazine  article,  Watching  TV  Makes  You  Smarter  (April  24,  2005),  Steven  

Stewart  McKie   -­‐  163  -­‐   As  of  15/08/2014  

Johnson  provides  storyline  analysis  of  episodes  from  US  TV  shows  Starsky  
and  Hutch,  Hill  Street  Blues  and  The  Sopranos  to  show  relative  narrative  
complexity  based  on  the  number  and  depth  of  the  interweaving  storylines.  
Clearly  Starsky  and  Hutch  is  the  most  simplistic  with  just  one  main  storyline  
and  a  bookending  start  and  finish  sub-­‐plot  whereas  both  Hill  Street  Blues  has  
more  storylines  and  The  Sopranos  is  the  most  complex  with  multiple,  denser  
storylines  and  considerable  overlapping  between  them.  

Figure  2.29a  -­‐  Starsky  and  Hutch  Storylines

Figure  2.29b  -­‐  Hill  Street  Blues  Storylines  

Stewart  McKie   -­‐  164  -­‐   As  of  15/08/2014  

Figure  2.29.c  -­‐  The  Sopranos  Storylines  

2.8  Using  ‘Big  Data’  Tools  for  Screenplay  Analysis  

As  mentioned  in  chapter  1,  the  datafication  of  the  screenplay  makes  the  data  
available  to  a  new  generation  of  analytic  tools  designed  to  work  with  ‘big  
data’  sets  and  automatically  derive  certain  analytics  with  minimal  user  
intervention.  Here  I  briefly  mention  three  such  tools,  all  of  which  were  
tested  with  data  derived  from  the  Scenepad  demo  script  Alien  (1979).    
Because  this  script  was  already  datafied  in  Scenepad  the  dataset  needed  to  
feed  these  analytic  tools  was  sourced  from  the  output  of  a  single  query,  
another  reason  why  datafication  of  content  helps  to  make  analytics  easier  
and  increases  the  range  of  options  available  to  the  analyst.  

Splunk  

Some  rather  prosaic,  but  easy  to  generate  analytics  of  screenplay  data  are  
provided  by  the  Open  Source  data  analysis  tool  Splunk  (downloaded  from  
splunk.com  –  accessed  April  28,  2013).  To  experiment  with  the  kind  of  
analytics  that  tools  targeted  at  big  data  problems  can  deliver,  I  imported  a  
version  of  the  Alien  (1979)  script  into  Splunk  for  analysis.    

The  core  value  of  Splunk  lays  in  its  ‘search  processing  language’  –  SPL.  With  
the  screenplay  data  in  Splunk,  I  generated  some  basic  ‘top  n’  analyses  of  the  
data,  which  are  discussed  below.  

Stewart  McKie   -­‐  165  -­‐   As  of  15/08/2014  

  

Figure  2.30  –  Splunk  Snippet  Analysis  

Here  we  see  that  almost  two-­‐thirds  of  the  Alien  script  consists  of  dialog  
snippets  and  one-­‐third  action  snippets.  At  first  glance  this  might  give  the  
impression  that  the  script  is  dialog-­‐heavy  and  action-­‐light  but  in  reality  this  
is  hardly  the  case.  Dialog  content  in  Alien  is  of  far  less  importance  than  
action  content  to  both  the  narrative  flow  and  the  characters  –  since  much  of  
the  action  consists  of  the  characters  being  killed  off  by  the  Alien.  

Figure  2.31  –  Splunk  Word  Analysis  

Splunk  highlights  ‘right’  as  the  most  frequent  word  in  the  snippet  content  
(15  occurrences),  a  word  that  many  might  regard  as  a  ‘noise’  word  rather  
than  anything  that  provides  any  meaningful  analysis  value.  

Stewart  McKie   -­‐  166  -­‐   As  of  15/08/2014  

  

Figure  2.32  –  Splunk  Role  Analysis  

Splunk  tells  us  that  the  ‘Alien’  role  is  referenced  in  523  snippets.  This  at  least  
indicates  that  the  Alien  –  conventionally  the  ‘antagonist  ‘  –  is  probably  the  
most  important  role  in  the  screenplay  with  almost  3  times  the  mentions  of  
the  ‘protagonist’  Ripley.    From  this  it  would  be  realistic  to  suppose  that  the  
‘causality’  for  this  might  be  that  the  Alien  is  in  fact  the  ‘driving’  character  in  
the  screenplay  –  the  villain  rather  than  the  hero(ine).  

Stewart  McKie   -­‐  167  -­‐   As  of  15/08/2014  

  

Figure  2.33  –  Splunk  Location  Analysis  

Splunk  tells  us  that  the  ‘bridge’,  ‘mess’  and  ‘infirmary’  locations  are  where  
the  most  action  and  dialog  snippets  take  place.    This  makes  sense  when  you  
consider  that  the  ‘causality’  for  this  is  that  these  are  three  main  communal  
areas  on  the  ship  where  multiple  members  of  the  crew  are  likely  to  spend  
time  and  to  converse  with  each  other.  

These  simple  analyses  make  minimal  use  of  the  SPL  and  other  Splunk  
features  but  they  are  almost  ‘one  click’  in  terms  of  analytical  effort  and  what  
they  illustrate  is  that  even  these  simple  correlations  in  the  data  may  be  
helpful  in  determining  ‘causality’  –  i.e.  why  the  data  is  what  it  is.  

BigML  

Compared  to  Splunk,  BigML  (see  https://bigml.com) positions  itself  as  

‘machine  learning  made  easy’  and  also  provides  predictive  analytics  tools  
within  its  analysis  toolkit.  As  with  Splunk  I  loaded  in  data  sourced  from  the  
Scenepad  demo  script  Alien.  

Stewart  McKie   -­‐  168  -­‐   As  of  15/08/2014  

  

Figure  2.34  –  BigML  Script  Analysis  

Figure  2.34  shows  the  basic  ‘immediate’  analysis  you  get  from  BigML.  

1.  The  title  histogram  simply  confirms  all  data  is  from  a  single  script.  
 
2.  The  heading  histogram  shows  the  distribution  of  action  and  dialog  
‘instances’  by  scene  (the  top  3  locations  are  the  bridge,  the  mess  and  the  
infirmary).  

3.  The  int/ext  histogram  shows  that  most  scenes  are  INT  (they  take  place  
within  the  Nostromo).  

4.  The  day/night  histogram  shows  that  most  scenes  are  NIGHT  (although  this  
is  not  that  helpful  as  they  all  take  place  in  space).  

5.  The  cue  histogram  shows  that  the  top  3  characters  in  terms  of  dialog  
instances  are  Dallas,  Ripley  and  Ash  but  also  that  the  instances  are  fairly  well  
distributed  across  the  cast  from  Dallas  (top)  to  the  Alien  (bottom).  

6.  The  gender  histogram  shows  there  is  just  over  twice  as  many  instances  
involving  male  characters  than  female  (despite  a  female  protagonist).  

7.  The  type  histogram  shows  there  are  about  six  times  as  many  action  

Stewart  McKie   -­‐  169  -­‐   As  of  15/08/2014  

instances  than  dialog  instances  so  this  likely  to  be  an  action-­‐heavy  movie.  

8.  The  text  histogram  shows  that  Ripley  leads  the  number  of  text  instances  
(i.e.  both  action  and  dialog)  over  Dallas  perhaps  reflecting  the  fact  that  she  
lasts  longer  than  he  does  in  the  script.  

Figure  2.35  –  BigML  Scene  Prediction  

Figure  2.35  shows  a  sample  predictive  analysis  ‘path’  based  on  the  scene  
data  provided.    Essentially  all  this  says  is  that  if  the  scene  is  INT  and  DAY,  
contains  KANE  and  not  ASH  or  DALLAS  it  is  93.98%  likely  to  be  located  in  
the  MESS.  It’s  easy  to  see  how  this  prediction  is  largely  based  on  which  
characters  are  most  often  associated  with  which  scene  locations  (because  
they  speak  dialog  at  that  location  or  take  part  if  action  at  that  location).    

Figure  2.36  shows  another  predictive  analysis  path  using  the  character  cue  
and  a  gender  attribute  to  predict  that  if  the  cue  is  not  Ripley  or  Lambert  or  
Mother’s  Voice  then  the  gender  of  the  character  is  99.89%  likely  to  be  male.    
In  reality  it’s  100%  likely  as  these  are  the  only  3  female  characters  identified  
in  the  dataset  –  all  the  rest  are  male.  But  the  point  is  that  BigML  is  predicting  
this  outcome  from  the  data.  

Stewart  McKie   -­‐  170  -­‐   As  of  15/08/2014  

  

Figure  2.36  –  BigML  Gender  Prediction  

Dataseed  

Dataseed  (http://getdataseed.com)  is  another  ‘in  beta’  online  data  analysis  

application  that  accepts  a  file  upload  and  generates  some  analyses  from  the  
dataset  immediately  after  upload.  As  with  BigML  I  uploaded  the  Alien  
screenplay  data  in  a  .CSV  format  and  generated  the  analysis  shown  in  figure  
2.37  below.  What  the  dataseed  visualization  tells  us  is:  

1. There  are  1339  action  &  dialog  ‘snippet’  rows  in  the  dataset.  
 
2. 100  snippets  are  in  EXT  scenes  and  1239  are  in  INT  so  we  know  we  
are  inside  most  of  the  time  (actually  inside  the  Nostromo).  
 
3. 174  snippets  are  in  DAY  scenes  and  1165  are  in  NIGHT  scenes  
(actually  the  ‘darkness’  of  space  on  board  ship).  
 
4. Most  of  the  action  and  dialog  takes  place  on  the  Bridge,  the  Mess  or  
the  Infirmary.  
 
5. There  are  514  action  snippets  and  825  dialog  (so  is  it  an  action-­‐heavy  
or  dialog-­‐heavy  script  –  we  don’t  know.).  
 
6. RIPLEY  and  DALLAS  have  the  most  important  roles  (so  one  or  other  is  
likely  to  be/should  be  the  Protagonist).  
 

Stewart  McKie   -­‐  171  -­‐   As  of  15/08/2014  

 7. Males  dominate  the  dialog  but  female  roles  have  a  good  deal  of  screen  
time  (540  vs.  285  dialog  snippets  respectively)  especially  considering  
there  are  only  2  of  them.  
 
8. The  Protagonist  role  has  a  significant  number  of  dialog  snippets  (186  
vs.  631  for  all  the  other  ‘Main’  characters  –  indicating  significant  
screen  time).  

Figure  2.37  –  Dataseed  Analysis  

What  these  three  simple  ‘experiments’  in  uploading  screenplay  data  to  data  
analytics  tools  tells  us  is  that  you  don’t  even  need  a  ‘Screenwriting  2.0’  
application  like  Scenepad  with  built-­‐in  analytics,  you  just  need  an  easy  way  
to  get  screenplay  content  delivered  in  the  form  of  a  data  file  (as  opposed  to  a  
document  text  file)  to  enable  these  kinds  of  tools  to  deliver  a  ‘starter’  set  of  
analytic  visualizations  from  the  content.    

The  data  file  provided  to  both  BigML  and  Dataseed  was  a  simple  .CSV  file  like  
the  sample  shown  in  figure  2.38  below.  

Stewart  McKie   -­‐  172  -­‐   As  of  15/08/2014  

  

Figure  2.38  –  Sample  of  Uploaded  Data  File  

Summary  

Screenplay  analytics,  the  analysis  and  visualization  of  screenplay  content,  is  
a  very  underdeveloped  area,  both  for  individual  screenplay  analytics  and  for  
screenplay  corpus  (e.g.  genre)  analytics.  Screenwriting  ‘how-­‐to’  manual  
writers,  media  publications  and  even  the  odd  academic  are  among  the  few  
sources  for  a  small  number  of  screenplay  content  visualizations  derived  
from  analysis  of  the  screenplay  text.  Some  generic  data  analysis  tools  can  be  
used  to  generate  analyses  and  visualizations  of  screenplay  content,  once  it  
has  been  ‘datafied’  into  an  appropriate  format  for  use  by  the  tool.  Despite  
the  lack  of  screenplay  analytic  capabilities  in  most  screenwriting  
applications,  screenplay  content  can  be  easily  imported  into  generic  data  
analysis  tools  like  Splunk  in  order  to  drive  some  level  of  analytics  from  the  
text.  

Much  of  this  lack  of  screenplay  analytic  activity  is  a  result  of  the  prevailing  
orthodoxy  of  working  with  screenplays  as  textual  documents  rather  than  
screenplays  as  data.  As  more  screenwriting  software  datafies  the  screenplay  
text,  and  especially  if  structured  and  standardized  formats  are  used,  it  is  
very  likely  that  new  kinds  of  analysis  and  visualization  of  screenplay  content  
will  emerge,  as  applying  analysis  and  visualization  to  the  ‘datafied’  text  will  
be  easier  to  do  and  the  text  will  become  more  easily  accessible  to  a  wider  
range  of  analytical  and  visualization  software.    

Stewart  McKie   -­‐  173  -­‐   As  of  15/08/2014  

But  screenwriting  2.0  is  about  more  than  screenplays  as  data  and  leveraging  
that  data  for  content  analysis  and  visualization.  It  is  also  about  the  
screenplay  as  social  network  and  collaborative  business  process  and  about  
new  perspectives  of  the  way  screenplays  are  made  or  screenplay  as  design.  
These  are  the  topics  discussed  in  my  next  chapter.  

Stewart  McKie   -­‐  174  -­‐   As  of  15/08/2014  

3  Screenplay  as  Social  Network  

Writing  and  realizing  a  screenplay  as  an  audio-­‐visual  work  is  not  a  discrete  
event  but  a  creative  and  potentially  business  process  that  often  reflects  a  
long-­‐duration  lifecycle,  sometimes  extending  across  many  years,  with  many  
different  activities  involving  many  different  stakeholder  roles  as  participants  
in  the  process.    Therefore  screenwriting  implies  a  social  process  with  the  
screenplay  text  itself  providing  the  context  or  ‘anchor’  for  a  social  network  
to  collaborate  around.  

A  single  writer  working  on  a  ‘closed’  screenplay  (i.e.  stored  on  their  local  PC  
and  not  shared  with  others)  is  not  a  social  network  but  a  writing  partnership  
with  just  one  other  partner  corresponds  to  the  minimum  dyadic  relationship  
needed  for  the  basis  of  a  social  network,  especially  if  this  relationship  
includes  use  of  an  ‘open’  screenplay  that  is  stored  online  and  accessible  via  
the  Web  for  sharing  with  other  collaborators.    

Although  a  screenplay  shared  with  a  large  group  might  benefit  from  the  
‘collective  intelligence’  gained  from  what  James  Surowiecki  popularized  as  
‘The  Wisdom  of  Crowds’  in  his  2005  book  of  the  same  name,  it  is  much  more  
likely  to  benefit  from  exposure  to  a  more  finite  and  smaller  group  of  
collaborators  where,  according  to  Kao  and  Couzin  (2014):  

…it  is  the  noise  inherent  in  these  small  groups  that  enhances  their  
accuracy,  allowing  individuals  in  such  groups  to  avoid  the  detrimental  
effects  of  correlated  information  while  exploiting  the  benefits  of  
collective  decision-­‐making.  

A  simple  example  of  an  early  form  of  online  social  network  is  the  blog.  
Typically,  the  owner/producer  posts  news,  reviews  or  opinions  and  the  
reader/consumer  posts  comments,  rates  the  content  or  reblogs  the  content  
to  co-­‐create  a  conversation  or  debate  around  the  post  topic.  Here,  the  social  
network  paradigm  we  are  more  interested  in  is  that  of  the  online  social  
networking  site/application  (SNS/SNA),  such  as  Facebook  or  Twitter,  
primarily  designed  to  function  as  social  networks  that  encourage  the  

Stewart  McKie   -­‐  175  -­‐   As  of  15/08/2014  

crowdsourcing  of  user  generated  content  (UGC)  that  typically  takes  the  form  
of  text,  images  or  audio/video  clips.    I  refer  to  social  network  applications  
(SNAs)  below,  not  social  network  sites  (SNS),  since  in  my  view  the  term  ‘site’  
implies  primarily  a  content-­‐delivery  purpose  whereas  the  term  ‘application’  
implies  both  delivering  content  and  the  functionality  needed  to  manage  that  
content.    

According  to  boyd  and  Ellison  (2007),  who  trace  SNS  development  over  the  
decade  1997-­‐2007,  social  network  sites  are  defined  as:    

…web-­‐based  services  that  allow  individuals  to  (1)  construct  a  public  or  
semi-­‐public  profile  within  a  bounded  system,  (2)  articulate  a  list  of  
other  users  with  whom  they  share  a  connection,  and  (3)  view  and  
traverse  their  list  of  connections  and  those  made  by  others  within  the  
system.  

However  this  is  a  rather  technical  definition  focused  on  online  SNSs.  For  our  
purposes  here,  I  define  a  social  network  as  comprising  actors  with  
relationships  between  them.    In  this  context  those  relationships  may  be  
developed  and  managed  both  face-­‐to-­‐face  and  online.  The  exchange  of  data  
between  participants  engaged  in  a  social  network  creates  social  capital.    The  
relative  engagement  of  the  network  actors  is  what  enriches  this  social  
capital.  This  engagement  is  reflected  in  the  amount  of  UGC  aggregated  
around  a  specific  actor  and  their  individual  content  items  (e.g.  blog  posts,  
Facebook  pokes,  Twitter  tweets).  This  UGC  can  take  various  forms  including  
commenting  or  rating  (e.g.  1-­‐5  star  ratings  or  Facebook  likes)  or  referencing  
(e.g.  blog  backlinks  or  Twitter  retweets).  The  social  network  is  anchored  by  
a  focal  point  around  which  a  specific  actor’s  social  capital  revolves.  

SNA   Actors   Anchor   Items   Engagement  

Facebook   Friends   Wall   Pokes   Likes  

Stewart  McKie   -­‐  176  -­‐   As  of  15/08/2014  

Twitter   Followers   Timeline   Tweets   Retweets  

Wordpress   Contributors   Blog   Posts   Comments  

Table  3.1  –  Social  Network  Characteristics  

Another  way  to  look  at  a  social  network  is  as  an  instance  of  a  creative  system  
model  as  proposed  by  J  T  Velikovsky’s  blog  post,  CREATIVE  PRACTICE  
THEORY:  The  Model,  of  15,  Dec  2012  (see  
http://storyality.wordpress.com/creative-­‐practice-­‐theory/  -­‐  accessed  09  
July  2013)  

that  identifies  some  characteristics  of    a  systems  model  that  could  equally  be  
applied  to  a  social  network:  The  ‘domain’  (e.g.  screenwriting),  the  ‘field’  (e.g.  
screenwriters)  and  the  ‘agents’  (e.g.  individual  writers)  and  arguably  the  
‘symbolic  capital’  at  the  centre  of  all  three  is  the  screenplay  content  itself.  

In  fact,  some  screenplays  could  involve  multiple  social  networks  that  

overlap  each  other  to  reflect  the  different  domains,  fields  and  agents  that  are  
involved  with  the  script  over  its  process  lifecycle  as  shown  in  figure  3.1  
below:  

Stewart  McKie   -­‐  177  -­‐   As  of  15/08/2014  

Figure  3.1  -­‐  The  3  principal  screenplay-­‐anchored  social  networks  

The  writing  network  involves  the  writers,  readers/analysts  and  others  who  
are  involved  in  writing,  rewriting  and  ‘polishing’  the  script  content.  The  
business  network  involves  those  agents,  producers  and  others  involved  in  
financing,  selling  and  marketing  the  screenplay.  The  production  network  
involves  those  involved  in  converting  the  content  into  a  movie  including  the  
director,  talent,  cinematographers,  production  managers  and  others.  In  each  
case  the  ‘domain’  and  ‘field’  is  different,  the  ‘agents’  reflect  a  different  
stakeholder  group  and  the  ‘symbolic  capital’,  while  fundamentally  the  
screenplay  content,  is  changing  all  the  time  and  presented  in  different  ways  
depending  on  the  network  using  it  (e.g.  production  shooting  script  vs.  
business  selling  treatment).  What  this  triple  domain  perspective  indicates  is  
that  the  screenplay  can  be  at  the  heart  of  a  rich  social  network.  

In  this  chapter  I  will  discuss  the  functionality  expected  of  an  online  social  
network,  the  range  of  screenplay  stakeholders  and  screenwriting  as  a  
process.  

3.1  Functionality  of  Online  Social  Networking    

We  recognize  an  online  SNA  by  its  provision  of  specific  functionality  that  
promotes  social  networking  and  fosters  ongoing  engagement  with  the  other  
network  participants  and  their  content.  

Typically,  SNAs  operate  in  both  public  and  private  modes  in  terms  of  their  
content  visibility.  The  ability  to  flag  content  as  public  or  private  (either  to  
you  or  a  specific  group  of  friends  or  followers  within  your  social  network)  is  
integral  to  most  SNAs.  However  a  SNA  anchored  by  a  screenplay  that  is  
commercially  ‘in-­‐play’  is  more  likely  to  operate  as  a  ‘closed’  network  where  
the  user  community  and  access  to  the  content  is  prescribed  by  the  
screenplay  owner  or  copyright  holder.  Participants  in  the  network  must  be  
invited  to  join  so  it  is  not  ‘open’  in  the  sense  that  any  online  user  can  view  
the  screenplay  content  and  interact  with  it  (as  is  the  case  with  a  typical  blog  

Stewart  McKie   -­‐  178  -­‐   As  of  15/08/2014  

for  example).  

A  SNA  is  by  definition  multi-­‐user.  Note  that  many  screenwriting  applications  
are  not  since  they  often  assume  a  single  user  working  with  the  screenplay  
content  –  namely  the  owner-­‐writer.  Multi  user  access  also  means  that  many  
SNAs  recognize  the  need  for  user  roles  (e.g.  editor,  author,  viewer)  to  
effectively  manage  content  access  and  usage  rights.  However  single  user  
screenwriting  applications  do  not  support  user  roles  since  they  are  not  
necessary.  

In  general,  most  online  SNAs  support  the  functionality  outlined  below:  

Create  and  grow  their  network  by  sharing  content    

In  an  ‘open’  social  network  the  aim  is  to  expand  an  individuals  network  as  
widely  as  possible  by  making  it  easy  to  share  content  between  users.  On  
Facebook  this  is  known  as  ‘friending’,  on  Twitter  ‘following’  and  on  LinkedIn  
‘linking’.  Sharing  activities  such  as  re-­‐tweeting,  re-­‐blogging  etc.  allow  
specific  tweets  or  blog  posts  to  go  ‘viral’  very  quickly,  reaching  many  more  
participants  in  the  social  network,  if  the  content  happens  to  catch  the  
imagination  of  the  crowd.  This  is  the  power  behind  tweets  that  ‘trend’  on  
Twitter  or  YouTube  videos  that  record  millions  of  views.    

Contribute  content    

All  social  networks  depend  on  user-­‐generated  content  (UGC)  to  function:  
Facebook  wall  pokes  and  Twitter  tweet  posts  are  both  examples  of  UGC,  as  
are  photos  posted  to  Flickr  or  video  clips  posted  to  YouTube,  and  it  is  this  
UGC  that  comprises  much  of  the  core  ‘data’  of  the  online  social  network  (the  
rest  being  content  added  by  other  users  specifically  intended  to  enrich  that  
core  content).  In  most  cases  the  initial  UGC  (blog  post,  tweet,  photo  etc.)  is  
itself  the  trigger  for  generating  new  UGC  around  it.  It  is  this  content  
enrichment  process  that  creates  the    ‘conversation’  that  in  turn  has  the  
potential  to  add  value  to  the  original  content.    

Stewart  McKie   -­‐  179  -­‐   As  of  15/08/2014  

Rate  content  

In  a  social  network,  rating  content  is  a  way  of  crowdsourcing  an  indication  of  
the  quality  of  that  content  and  also  a  means  for  filtering  network  content  
programmatically  so  that  ‘cream  rises  to  the  top’  in  the  form  of  an  
automated  ranking  based  on  user  ratings.  The  Facebook  ‘like’  is  a  simple  
rating  system  for  ranking  leading  and  lagging  content.  The  Twitter  ‘retweet’  
is  another  way  of  saying  you  value  content  by  actively  sharing  it.  The  most  
liked  and  most  retweeted  content  is  likely  to  indicate  popularity  that  is  
perhaps  a  reflection  of  content  quality.    

Crowdsourced  ratings  are  highly  effective  in  the  e-­‐commerce  domain,  for  
example  the  1-­‐5  start  rating  system  for  both  products  and  sellers  on  Amazon  
that  may  have  a  direct  impact  on  the  buying  decision  of  a  customer.  Usually  
anyone  can  rate  content  and  it  is  up  to  content  consumers  to  decide  how  
relevant/realistic  those  ratings  are  to  them  individually.  

Comment  on  Content  

Providing  feedback  in  the  form  of  comments  is  an  integral  part  of  every  
social  network  and  a  simple  example  of  a  peer  review  process  in  action.  
Commenting  on  content  is  a  key  way  to  encourage  the  original  author  to  
review  and  revise  their  content  based  on  the  multiple  perspectives  reflected  
in  a  comment  stream.  Comments  are  generally  viewed  in  context  i.e.  
alongside  or  below  the  content  being  commented  on.  The  number  and  
quality  of  comments  is  generally  considered  to  be  an  indication  of  the  
original  ‘stimulating’  content’s  quality.  

A  problem  with  many  SNA  comment  streams  is  that  they  contain  a  great  deal  
of  ‘chaff’  e.g.  of  the  ‘awesome  dude’  or  ‘this  sucks’  variety.  This  chaff  is  
seldom  very  helpful  to  the  content  creator.  That  is  why  it  can  help  to  force  
comment  senders  to  categorize  their  comments,  for  example  as  ‘suggestions’  
or  ‘problems’,  which  in  turn  helps  the  comment  receiver  to  better  analyze  
these  comments  and  take  action.  Categorizing  comments  will  not  prevent  

Stewart  McKie   -­‐  180  -­‐   As  of  15/08/2014  

chaff  but  may  help  to  reduce  it.  

Aggregate  content  from  multiple  sources    

Content  aggregation  becomes  important  when  there  are  a  multitude  of  

places  that  content  can  be  posted  to  or  sourced  from.  This  has  become  more  
important  given  the  pervasive  adoption  of  mobile  devices,  especially  
smartphones,  that  can  capture  content  anywhere,  anytime  and  then  post  it  
to  various  online  SNAs.  Content  aggregation  will  become  more  important  as  
more  devices  become  part  of  the  Internet-­‐of-­‐Things  and  can  communicate  
data  wirelessly  to  a  specific  collection  application.  For  example  when  a  
camera  used  to  shoot  a  script  in  production,  a  ‘thing’,  can  automatically  
communicate  metadata  about  the  filming  via  the  Internet  to  a  screenwriting  
application  where  it  can  be  stored  along  with  other  scene  data  and  
metadata.  

Early  forms  of  online  content  aggregation  included  online  newsreaders  that  
aggregated  RSS  (Really  Simple  Syndication)  feeds  from  multiple  source  sites.  
The  use  of  RSS  in  this  way  facilitated  easy  syndication  of  content  to  multiple  
content  consumer  target  applications  from  a  single  publisher  source  
application.    

Other  than  via  RSS,  the  usual  way  to  share  and  aggregate  content  
programmatically  by  use  of  the  source  and/or  target  application’s  API  
(Application  Programming  Interface).  An  API  generally  provides  a  secure  
way  for  a  consuming  application  to  request  and  get  specific  content  from  a  
publishing  application.  All  leading  SNAs  –  for  example  Twitter  and  Tumblr  -­‐  
offer  such  APIs  to  promote  posting  and  querying  their  SNA  data  
programmatically.    

Activity  Updates  and  Alerts  

A  SNA  highlights  the  engagement  of  its  users  by  making  much  of  their  
engagement  activity  transparent  to  other  users  in  the  form  of  activity  
streams  and  alerts  (e.g.  ‘X  posted  Y’  or  ‘A  is  now  linked  to  B’).  An  activity  

Stewart  McKie   -­‐  181  -­‐   As  of  15/08/2014  

stream  visualizes  this  activity  in  the  form  of  an  activity  list  or  a  posting  
timeline  that  constantly  updates  to  reflect  the  latest  user  activity.  Alerts  may  
be  triggered  to  alert  specific  users  to  specific  activity  based  on  a  set  of  
alerting  preferences  managed  by  the  user.  The  alert  (‘C  likes  your  D’)  is  
usually  delivered  automatically  to  the  interested  users  by  email  or  SMS  text  
message.  The  surfacing  of  activity  streams  and  automated  alerting  help  to  
increase  engagement  with  the  SNA  by  prompting  users  to  respond  in  some  
way  –  even  if  only  to  login  to  the  SNA  rather  than  to  elicit  any  specific,  formal  
response.  

Gamification  and  PBLs  

Points,  Badges  and  Leaderboards  (PBLs)  have  also  become  an  aspect  of  
social  networks,  borrowed  from  computer  gaming.  According  to  Werbach  
and  Hunter  (2012),  ‘gamification’  of  social  networks  can  be  a  way  to  
increase  engagement  with  and  feedback  from  a  social  network.  This  
gamification  often  uses  PBLs  by  awarding  points  for  activities  such  as  
posting  content  comments  or  rating  ‘likes’,  leading  to  award  of  badges  as  
participants  ‘level-­‐up’  to  pre-­‐defined  award  levels.  High  scoring  participants  
are  then  showcased  on  leaderboards  e.g.  Top  10  Commentators  (on  the  
network)  to  provide  recognition  for  their  efforts.    Gamification  and  PBLs  
probably  make  less  sense  in  a  relatively  small  social  network  focused  on  a  
single  screenplay’s  content  but  might  make  sense  in  a  larger  network  
focused  on  a  corpus  of  screenplays  for  a  use  case  such  as  genre  analysis  and  
are  certainly  more  applicable  to  sites  focused  on  delivery  or  discussion  of  
movies  for  example,  rather  than  screenplays  

3.2  Screenplay  Stakeholders  

Most  of  the  current  generation  of  screenwriting  applications  does  not  
support  this  social  networking  perspective  of  screenwriting.  This  is  largely  
because  many  were  conceived  or  developed  before  the  rise  of  social  
networking  as  an  Internet  phenomenon  and  so  tend  to  focus  on  
screenwriting  as  an  individual  creative  process  rather  than  as  a  

Stewart  McKie   -­‐  182  -­‐   As  of  15/08/2014  

collaborative  business  process  that  eventually  can  involve  a  social  network  
of  both  creative  and  commercial  stakeholders.  

That  professional  screenwriting  and  script  development  is  a  business  

process  is  well  understood  outside  the  world  of  screenwriting  software.    For  
example,  the  Creative  Skillset’s  overview  of  the  script  department  
(http://www.skillset.org/film/jobs/script/  -­‐  accessed  04  September  2012)  
states:  

Although  the  Screenwriter  or  Screenwriters  are  central  to  the  script  
writing  process,  like  every  other  aspect  of  film  making,  development  is  
collaborative  work  and  typically  requires  the  creative  input  of  a  
number  of  other  film  professionals  including:  Script  Readers,  Writer's  
Agents,  Producers,  Development  Executives,  Script  Editors,  and,  
eventually,  Directors,  who  are  often  involved  in  the  final  versions  of  
the  screenplay  and  shooting  script.  When  projects  are  developed  as  
part  of  a  Screen  Agency,  Broadcaster,  or  National  Lottery  funded  film  
scheme,  Development  Executives  from  these  agencies  and  script  
development  staff  may  also  be  involved  in  the  process.  

This  process  reflects  the  journey  that  the  script  makes  from  an  initial  idea  to  
the  final  shooting  script  from  which  a  movie  is  more  or  less  directly  shot.    

In  fact  screenwriting  as  a  collaborative  activity  has  much  in  common  with  
the  idea  management  activity  within  the  business  domain  of  innovation  
management.  The  progression  of  the  screenplay  development  process  from  
ideation  to  realization  has  many  steps  that  include  genuine  commercial  
‘stage-­‐gates’  at  which  business  decisions  are  made  –  optioning,  developing,  
green-­‐lighting  -­‐  just  like  any  other  commercial  product  development  
process.  And  like  the  manufacture  of  a  physical  object,  a  screenplay  can  
generate  many  versions  and  prototypes  in  the  form  of  script  drafts  and  
treatments,  storyboards  and  pre-­‐viz  visualizations.  

A  script  may  begin  its  journey  owned  by  an  individual  screenwriter  or  a  
partnership  or  team  of  screenwriters.  The  former  is  a  common  paradigm  for  
either  a  speculative  or  commissioned  feature  film  screenplay  and  the  latter  
for  a  screenplay  destined  for  TV  –  especially  if  written  as  an  episode  within  
an  established  series  leveraging  an  established  writing  team  with  a  

Stewart  McKie   -­‐  183  -­‐   As  of  15/08/2014  

showrunner  and  a  series  bible.  

In  terms  of  collaborative  screenwriting,  what  is  of  interest  here  are  the  
similar  yet  different  workflows  the  individual  and  team  scripts  follow,  the  
common  phases  within  these  workflows  and  the  needs  and  activities  of  the  
various  roles  who  participate  in  the  workflows.  In  practice,  the  original  or  
‘initiating’  screenwriter  may  not  in  fact  participate  in  the  process  from  start  
to  finish.  This  originator  may  in  fact  be  ‘cut  loose’  from  the  moment  the  
script  copyright  ownership  is  purchased,  replaced  by  other  writers  while  the  
script  is  ‘in  development’  or  simply  marginalized  by  more  powerful  
production  or  commercial  roles  such  as  the  Director  or  Producer  even  if  
they  remain  on  board  for  the  duration  of  the  project.  

Screenplays  are  either  commissioned  (i.e.  paid-­‐for  by  a  third  party)  or  
produced  speculatively  (i.e.  without  advance  payment)  by  a  screenwriter  
who  hopes  to  sell  his  work.  Kawin  (1992:302-­‐326)  provides  a  thorough  
explanation  of  these  two  paths  of  screenplay  genesis.  From  the  work  of  
MacDonald  (2010)  on  the  screen  idea  and  its  development,  and  anecdotal  
descriptions  of  working  with  a  writing  partner  (e.g.  
http://spitball.com/2009/04/how-­‐we-­‐collaborate-­‐on-­‐a-­‐screenplay,  
accessed  28  August  2012)  it  is  clear  that  the  original  author(s),  or  any  other  
writer  involved  with  the  script,  is  unlikely  to  be  individually  largely  
responsible  for  or  in  control  of  every  draft  of  a  script  from  pre  to  post  
production  (unless  the  screenplay  is  created  and  produced  by  a  
writer/director).  In  other  words  most,  if  not  all,  screenplays  are  co-­‐created  
that  are  eventually  realized  cinematically.  

As  a  collaborative  process  it  may  be  that  the  script  is  initially  written  and/or  
developed  by  a  writing  partnership  or  team  –  this  is  often  the  case  with  
scripts  written  for  television  (as  the  classic  US  television  series  The  Dick  Van  
Dyke  Show  and  more  recently  30  Rock  portray).  But  whatever  the  genesis  of  
the  script,  the  process  of  writing  and  rewriting,  combined  with  the  way  that  
scripts  are  marketed,  developed  and  produced  will  almost  certainly  mean  
that  any  "final’  shooting  script  will  reflect  input  from  a  wide  range  of  

Stewart  McKie   -­‐  184  -­‐   As  of  15/08/2014  

stakeholders  in  addition  to  the  original  screenwriter,  who  may  in  fact  have  
been  removed  from  the  project  a  long  time  ago.    

There  are  a  number  of  stakeholder  roles  involved  in  the  screenwriting  
process.  Here  I  identify  some  key  roles,  their  involvement  in  the  process  and  
their  potential  interest  in  being  able  to  analyze  the  screenplay  content  (as  
facilitated  by  the  datafication  of  the  content).  
 
Writers  
 
The  writers  include  anyone  who  is  involved  in  writing  or  rewriting/revising  
screenplays.  The  original  author(s)  and  subsequent  ‘rewriters’  and  
‘polishers’  attached  to  the  script  are  included  in  this  role.  This  is  the  primary  
role  involved  in  conceiving  and  writing  the  initial  screenplay  content  (data  
and  metadata)  re-­‐writing  it  and  being  officially  credited  with  the  writing.  
This  role  is  motivated  by  the  desire  to  write  a  better  script  to  increase  its  
commercial  viability  or  to  respond  to  the  needs  of  the  producer/director  
who  may  be  subject  to  their  own  budgetary  or  artistic  forces.    

Investors/Producers/Directors  
 
Investors,  Producers  and  Directors  include  anyone  who  is  involved  in  
participating  in  the  financing,  buying  or  optioning  of  a  screenplay  and  who  
may  require  the  work  to  be  formally  evaluated  prior  to  any  investment  or  
purchasing  decision.  This  group  is  motivated  by  the  need  to  identify  scripts  
that  are  judged  likely  to  succeed  commercially  and  therefore  worth  
investing  in  for  an  acceptable  financial  return.    
 
Script  Readers/Story  Analysts  
 
Script  readers  and  story  analysts  include  anyone  who  professionally  (i.e.  as  a  
job  role  or  as  freelancer  for  a  fee)  reads  and  analyzes  screenplays.  This  
group  often  works  on  behalf  of  investors/producers/directors  or  writers,  in  

Stewart  McKie   -­‐  185  -­‐   As  of  15/08/2014  

order  to  produce  ‘coverage’  or  any  other  kind  of  report  or  analysis  that  
evaluates  the  artistic  merit  and/or  commercial  potential  of  a  screenplay.  
These  roles,  like  Agents,  often  act  as  ‘gatekeepers’  in  terms  of  their  influence  
on  the  continued  progression  of  script  into  the  production  or  realization  
process.  
 
Production  Managers  
 
Production  managers  include  anyone  who  is  involved  in  managing  the  
production  of  a  movie  from  a  screenplay.    This  group  is  interested  in  the  
implications  of  a  screenplay  from  a  production  perspective,  for  example  the  
script  editor  for  continuity  or  specialist  crew  and  managers  focused  on  
locations,  props,  makeup,  costumes  or  special  effects.  The  screenplay  acts  as  
the  primary  source  for  the  production  breakdown  reporting,  shot  lists  and  
daily  call  sheets  that  are  essential  to  the  mechanics  of  movie  production.  

Talent  
 
Talent  refers  to  the  actors  who  are  potentially  or  contractually  attached  to  a  
project.  Their  motivation  for  engaging  with  the  screenplay  content  may  be  in  
better  understanding  the  screenplay  to  perform  their  role  or  for  deciding  
whether  or  not  to  buy  or  option  the  screenplay  for  their  own  production  
companies  or  whether  to  attach  themselves  to  an  existing  production  
project  because  of  the  quality  of  the  role  or  its  ‘Oscar’  potential  for  example.    
 
Agents  
 
Agents  sit  between  screenwriters  and  producers/directors/talent  to  
perform  a  “filtering”  and  selling  role  that  may  also  require  the  engagement  
of  script  readers/analysts  to  pre-­‐qualify  screenplays  before  the  agent  can  
pitch  them  to  potential  investors  in  the  script  (e.g.  
producers/directors/talent/angels).  They  may  be  interested  in  using  
screenplay  content  analysis  and  visualization  to  cut  their  slush  pile  of  

Stewart  McKie   -­‐  186  -­‐   As  of  15/08/2014  

unsolicited  “spec”  scripts  from  unknown  writers  or  to  help  them  apply  
additional  due  diligence  to  solicited  scripts  or  those  from  existing  clients.      

CGI/SFX/VFX  

The  increasing  use  of  green/blue  screen  techniques  and  computer  generated  
images  (CGI)  to  create  special  and  virtual  effects  (SFX/VFX)  in  movie  
production  means  that  a  relatively  recent  addition  to  the  stakeholder  group  
are  the  technicians/programmers  and  others  required  to  deliver  these  
effects.  Here  the  script  functions  as  a  functional  input  into  the  software  
development  project  required  to  realize  this  aspect  of  the  movie  production.  

This  range  of  roles  indicates  that  individual  screenplay  data  ‘consumers’  will  
need  to  interact  differently  with  a  screenplay  at  different  stages  of  the  script  
development  and  production  process  depending  on  their  point-­‐of-­‐view.  

Note  that  here,  we  are  not  concerned  with  the  audience  of  the  finished  
movie  as  a  screenplay  stakeholder.  The  audience  of  a  finished  movie  is  
seldom  aware  of  or  interested  in  the  screenplay  after  it  has  changed  form  i.e.  
from  written  artefact  to  produced  movie.    Nor  are  other  stakeholders  
involved  the  post-­‐production  sale  and  distribution  of  the  movie  relevant  
here.  

3.3  Screenwriting  as  Business  Process  

Screenwriting  as  a  business  process  can  be  discussed  in  at  least  three  ways:    

1.  As  having  pre-­‐production  and  production  phases  

2.  As  having  pre-­‐sale  and  post-­‐sale  phases  

3.  As  an  idea  development  process  

Again,  here  we  are  not  concerned  primarily  with  the  post-­‐production  
business  process,  when  the  screenplay  has  changed  form  (from  text  to  
movie).  Once  a  movie  is  made,  the  screenplay  itself  has  changed  form  from  a  

Stewart  McKie   -­‐  187  -­‐   As  of  15/08/2014  

fluid,  ‘living’  document  of  a  potential  movie  to  a  ‘frozen’  transcript  of  a  
realized  movie.  We  are  only  concerned  with  the  business  process  that  
relates  directly  to  the  development  of  the  screenplay  as  potential  movie.  

Pre-­‐Production  

Pre-­‐production  is  the  process  phase  that  occurs  before  the  actual  production  
(or  shooting)  of  a  movie  begins.  At  this  point  the  script  may  not  even  have  
been  written,  if  commissioned  by  a  production  company  or  not  bought  but  
only  optioned,  if  a  speculative  script.  The  stakeholders  involved  in  pre-­‐
production  could  include:  

 -­‐The  Screenwriter(s)  primarily  responsible  for  the  initial  script  draft(s).  

-­‐  The  Script  Analyst(s)  who  may  help  the  screenwriter(s)  to  refine  the  script  
by  providing  analysis  or  coverage  of  the  script.    

-­‐  The  Script  Reader(s)  responsible  for  assessing  the  commercial  worth  of  a  
script  in  terms  of  its  production  potential  for  a  specific  production  company.  

-­‐  The  Script  Rewriter(s)  and  Polisher(s)  who  may  be  hired  to  rewrite  or  
otherwise  improve  specific  aspects  of  the  script  pre-­‐production.  

-­‐  Producers,  Directors  and/or  the  Talent  who  may  provide  input  resulting  
from  their  reading  and  interpretation  of  the  script  once  they  become  
‘attached’  to  a  project.  

At  this  stage,  the  content  of  the  script  can  be  highly  fluid  since  it  does  not  yet  
have  any  direct  production  (i.e.  financial)  impact.        

Production  

Production  is  the  process  phase  in  which  shooting  the  movie  takes  place  and  
the  bulk  of  the  movie  budget  is  spent.  At  this  point,  a  substantive  script  will  
normally  exist  (unless  improvisational  techniques  are  being  used  by  the  
director  and  the  script  is  being  created  largely  ‘on-­‐the-­‐fly’  on  set)  and  the  

Stewart  McKie   -­‐  188  -­‐   As  of  15/08/2014  

production  company  will  have  bought  the  rights  to  the  script  from  the  
original  screenwriter  and  assumed  ownership.  The  stakeholders  involved  
during  production  include:  

-­‐  The  original  screenwriter  (but  only  if  their  contract  with  the  production  
company  continues  their  involvement).  

-­‐  The  Producer(s)  who  may  influence  the  script  to  meet  production  
timing/budgeting  demands.  

-­‐  The  Director(s)  who  may  influence  the  script  to  reflect  their  personal  
directorial  "vision".  

-­‐  The  Talent  who  may  influence  the  script  to  reflect  their  personality  or  
acting  preferences.  

-­‐  Various  production  managers,  who  may  influence  the  script,  say  to  make  it  
easier  or  more  cost-­‐effective  to  make  into  a  movie  from  their  production  
perspective.  

-­‐  The  Script  Doctors(s)  and  Polisher(s),  the  ‘hired  guns’  who  rewrite  or  
otherwise  improve  the  script  content  as  required  during  the  shoot.  

The  screenplay  content  may  change  significantly  during  this  development  

process  but  any  revisions  are  usually  clearly  identified,  not  only  for  writing  
credit  purposes,  but  also  as  they  may  have  a  production/financial  impact.  
Traditionally,  at  this  stage,  revisions  are  added  to  the  script  on  different  
coloured  paper  to  reflect  different  revision  cycles.    These  revisions  may  
include  line-­‐by-­‐line  edits  or  wholly  new  scenes  or  cause  whole  scenes  to  be  
deleted  altogether.  Characters  and  locations,  special  effects  and  sounds  may  
also  be  developed,  dropped  or  added.  Dialog  in  particular  may  be  
substantially  revised  as  it  is  successively  polished  for  optimum  impact.  
Locations  may  be  dropped  and  action  scaled  back  to  reflect  budgetary  
concerns.  

Stewart  McKie   -­‐  189  -­‐   As  of  15/08/2014  

Of  course,  it  is  possible  for  a  script  to  change  post-­‐production  during  the  
editing  process  -­‐  say  to  dub  on  a  new  piece  of  dialog  –  and  clearly  an  editor  
who  cuts  a  movie  is  in  effect  cutting  the  shooting  script  content.  But  by  the  
time  a  movie  is  in  post-­‐production,  the  original  screenplay  is  essentially  a  
historical  document  and  it  is  the  transcription  of  the  final  movie  that  
becomes  the  truly  representative  and  actual  ‘final  draft’  screenplay  version.    

Pre  and  Post  Sale  

Another  way  of  viewing  the  screenwriting  process,  specifically  from  the  
screenwriter’s  perspective,  is  to  divide  it  into  pre-­‐sale  and  post-­‐sale  phases.  
As  the  discussion  above  indicates,  a  pre-­‐production  draft  may  bear  little  
resemblance  both  in  form  and  content  to  the  final  shooting  script.  In  fact  
they  are  being  used  for  different  purposes.  The  purpose  of  a  pre-­‐production  
script  is  to  sell  the  story  vision  of  the  writer,  whereas  the  purpose  of  the  
production  shooting  script  is  to  efficiently  convert  the  script  into  shots  that  
can  be  edited  into  a  movie.  

During  this  selling  process,  the  screenplay  also  is  represented  by  a  variety  of  
documents  that  have  different  purposes  within  the  process:  Treatments,  
synopses,  contracts,  coverage  reports,  ‘the  first  10  pages’  and  so  on  that  
reflect  the  fact  that  not  every  stakeholder  involved  in  this  process  is  
interested  in  the  actual  screenplay  content  itself  but  only  a  version  of  it,  
appropriate  to  their  needs  as  a  stakeholder.  The  screenplay  itself  still  
functions  as  the  ‘anchor’  for  all  these  spin-­‐off  documents  but  it  does  not  
stand-­‐alone  it  is  part  of  a  document-­‐set.  

Pre-­‐sale,  the  script  has  latent  rather  than  actual  value  and  the  script  is  
essentially  totally  controlled  and  owned  by  the  original  writer(s).  Pre-­‐sale,  
the  job  of  the  screenwriter  is  to  improve  the  script  so  that  its  latent  value  
can  be  realized,  resulting  in  either  the  script  being  optioned  or  bought  or  at  
least  winning  some  kind  of  award  or  competition  that  may  help  to  improve  
its  chances  of  being  optioned  or  bought.    

Stewart  McKie   -­‐  190  -­‐   As  of  15/08/2014  

The  way  this  is  done  is  by  an  ongoing  cycle  of  revisions  and  rewrites,  as  I  
outline  in  figures  3.2a  &  3.2b,  based  on  the  writer’s  own  reflection  or  
internal  feedback  combined  with  coverage  from  readers  and  analysts,  and  
more  or  less  external  feedback  from  co-­‐writers,  and  from  pitches  to  festivals,  
competitions,  production  companies  and  other  interested  parties.  Much  of  
this  cycle  involves  significant  reading  and  re-­‐reading  of  the  script,  which  is  
both  time-­‐consuming  and  repetitive.  

Figure  3.2a  -­‐  Write  –  Reflect  -­‐  Rewrite  

Production  company  readers  and  paid-­‐for  script  analysts  generally  produce  

a  formal  written  report  that  summarizes  their  personal  analysis  of  the  
quality  of  the  screenplay  content  and  its  commercial  viability.  Good  coverage  
is  valuable  feedback  to  help  a  writer  to  improve  the  script  or  to  facilitate  the  
marketing  of  their  screenplay  (for  example  a  script  that  has  coverage  from  a  
bona-­‐fide  production  company  is  by  definition  ‘in  play’  –  whatever  the  
perceived  positive  or  negative  nature  of  the  coverage).  A  useful  summary  of  
the  content  and  purpose  of  a  coverage  report  can  be  found  in  Garfinkel  
(2007:53-­‐95).  

Coverage  is  probably  the  only  formal  script  ‘analysis’  that  a  screenplay  will  
ever  get  pre-­‐sale.  But  all  coverage  inevitably  reflects  the  opinion  of  an  
individual  script  reader  or  analyst–  albeit  an  individual  likely  to  be  skilled  in  
providing  this  kind  of  analysis.  For  this  reason,  any  analysis  is  inevitably  

Stewart  McKie   -­‐  191  -­‐   As  of  15/08/2014  

broadly  subjective  based  on  intuition  and  market  experience  rather  than  
broadly  objective  based  on  the  evidence  of  the  script  data.  Subjectivity  may  
be  a  good  thing,  since  it  reflects  certain  requirements  and  biases  prevalent  in  
the  industry  at  the  time,  or  a  bad  thing,  since  it  reflects  certain  reader-­‐
specific  or  production  company  prejudices  or  preferences  that  may  ignore  
genuine  qualities  in  the  screenplay  content.  

Adding  a  new  step  into  the  screenplay  re-­‐writing  process,  that  of  some  kind  
of  script  content  analysis  and  visualization,  has  the  potential  to  significantly  
enrich  the  process  as  a  whole  by  creating  useful  new  feedback  loops,  as  
figure  3.2b  indicates.    

Figure  3.2b  -­‐  The  Enriched  Screenwriting  Process  

However,  it  is  unclear  what  constitutes  a  ‘high  quality’  script,  pre-­‐
production.    For  example  does  this  mean  a  script  that  is  judged  as  having  a  
satisfying  visual  narrative  or  one  that  is  judged  as  having  the  potential  for  
high  box-­‐office  (i.e.  commercial),  a  potential  Oscar  winner  or  what?    

Clearly,  from  a  commercial  perspective,  this  judgment  depends  largely  on  

what  a  particular  buyer  is  looking  for  in  order  to  satisfy  the  needs  of  the  
market  they  are  targeting  the  movie  at.  It  may  also  be  partly  determined  by  
the  availability  of  specific  funding  streams  or  tax  breaks  to  defray  

Stewart  McKie   -­‐  192  -­‐   As  of  15/08/2014  

production  costs.  Whereas  from  the  screenwriter’s  perspective  it  may  
depend  on  what  kind  of  screenplay  the  writer  set  out  to  write.  Is  the  writer  
satisfied  with  what  they  have  created  in  creative  terms?    

What  is  clear  is  that  it  is  impossible  to  tell  anything  about  the  quality  of  a  
script,  (that  has  not  been  ‘datafied’  in  order  to  facilitate  quantitative  
analytics)  without  reading  and  re-­‐reading  it,  either  as  a  whole  or  in  part,  
since  script  quality  is  a  value  judgment  based  on  the  specific  reader's  
interpretation  of  the  script  content  they  have  read  from  their  subjective  
perspective.  

But  if  certain  qualitative  characteristics  can  be  defined  and  a  script  
"processed"  to  look  for  matches  (or  not)  or  correlations  with  these  
qualitative  characteristics,  then  it  should  be  possible  to  make  meaningful,  
“first  cut”  qualitative  judgments  about  a  script  before  needing  either  to  read  
it  first  or  to  read  subsequent  revisions  made  based  on  the  qualitative  
characteristics  revealed.  At  the  very  least  this  new  process  step  would  allow  
the  script  analyzer  to  “see”  some  of  the  “unread”  content  and  “latent”  
potential  of  the  script  before  even  reading  the  actual  text.    

This  is  not  to  say  that  it  is  not  necessary  to  read  the  script,  far  from  it,  only  
that  much  could  be  learned  about  the  characters,  narrative  structure  and  
genre  conventions  of  the  script,  for  example,  before  reading  it  so  that  when  
it  is  read,  it  may  be  read  from  a  different  start-­‐point  -­‐  one  that  reflects  a  level  
of  evidential  analysis  that  has  already  been  done  in  advance  based  on  the  
data.    

Alternatively  the  script  could  be  read  first  and  then  analyzed/visualized  
retrospectively  using  a  tool  so  that  the  latter  step  could  be  used  to  more  or  
less  confirm  or  deny  aspects  of  the  “read”  analysis  or  prompt  the  
reader/analyst  to  go  back  and  review  certain  aspects  of  the  script  again  
based  on  what  the  tool  reveals.  In  this  sense  the  analysis  and  visualization  
step  can  function  both  as  a  “pre-­‐read”  and  a  “post-­‐read”  diagnostic  aid.  

Stewart  McKie   -­‐  193  -­‐   As  of  15/08/2014  

Idea  Development  

The  screenplay  business  process  can  also  be  viewed  as  an  idea  development  
process  or  workflow  whereby  an  original  idea  (whether  commissioned  or  
speculative)  is  developed  from  a  raw  idea  into  a  series  of  ‘deliverables’  in  the  
forms  of  a  treatment,  synopsis,  rough  draft,  final  draft,  shooting  script  and  
finally  transcript.  In  this  scenario,  the  ‘business’  of  the  process  is  that  of  the  
accretion  of  content  around  the  original  idea  kernel  together  with  some  kind  
of  content  transformation  step  at  each  of  the  process  stage  gates.  

Screenwriting/Screenplay   Innovation/Idea  Stage  Gates  

Activities  

Story  planning  (treatment,  synopsis)   Envisioning  idea  

Story  drafting   Expressing,  Defining  idea  

Story  rewriting   Enhancing,  Enriching  idea  

First  draft     Approved/Viable  idea  

Optioned  Script   Evaluated/Valuable  idea  

Script  polishing   Developed/Prototyped  idea  

Final/Shooting   Realized  idea  ready  for  

manufacture/release  

Transcript  of  movie   Product/service  in  use/practice  

This  idea  management  perspective  emphasizes  the  importance  of  the  first  

Stewart  McKie   -­‐  194  -­‐   As  of  15/08/2014  

stage  of  a  screenplay  lifecycle,  namely  ideation:  the  development  of  a  script  
idea  –  the  genesis  or  gestation  stage.  At  this  stage  there  are  not  necessarily  
any  screenplay  drafts  or  substantive  script  content  of  any  kind.  Just  an  idea  
of  what  the  script  is  intended  to  be  about.  MacDonald  (2004)  has  proposed  
the  concept  of  a  shared  screen  idea  and  a  work  group  that  engages  with  it,  
both  of  which  are  fundamental  to  the  ideation,  creation  and  realization  
stages  of  the  screenplay  lifecycle.  

…  I  have  argued  that  such  practice  congregates  around  a  shared  screen  

idea  rather  than  focusing  on  a  specific  written  text…I  have  also  
suggested  the  idea  of  a  Screen  Idea  Work  Group  (SIWG)  as  a  flexible  
and  semi-­‐formal  work  unit  that  congregates  around  the  screen  idea,  
and  whose  members  contribute  to  its  development.  

Director  Peter  Jackson  also  refers  to  a  3-­‐stage  screenwriting  process  that  
resembles  ideation  and  emphasizes  the  screenplay  beginning  as  an  
imaginative  work:  (http://gointothestory.blcklst.com/2011/04/peter-­‐
jackson-­‐on-­‐screenwriting-­‐process.html,  accessed  04  September  2012).  

We  always  find  there  are  three  distinct  phases  in  the  life  of  a  film  
script.  First,  it  exists  before  the  film  starts  shooting.  In  this  period,  
which  can  last  from  months  to  years,  the  script  is  a  theoretical  
document—an  imaginative  version  of  the  movie.  

Then  you  start  shooting  and  things  come  much  more  into  focus—
usually  in  a  very  positive  way.  We  now  have  actors  who  bring  their  
skill  to  the  roles  and  suddenly  we  see  the  characters  in  a  more  vivid  
and  tangible  way…  

The  final  writing  phase  comes  in  post-­‐production,  when  the  movie  is  
edited.  No  matter  what  you  were  imagining  when  you  wrote  the  script,  
and  what  you  imagined  during  the  shoot,  nothing  now  matters  beyond  
the  actual  cut  film.  We  often  find  that  script  work  continues  during  
post,  including  writing  and  shooting  new  scenes,  reorganising  the  
order  of  scenes,  or  recording  additional  dialogue  to  slip  into  shots.  We  
do  all  of  these  things,  and  the  writing  only  stops  when  the  film  is  finally  
finished.  

Jackson’s  observations  are  important  because  they  identify  the  dynamic  

nature  of  the  screenplay,  especially  during  production,  when  the  screenplay  
naturally  informs  the  filming  but  as  Jackson  makes  clear,  the  opposite  also  
applies.  In  this  sense,  both  screenplay  and  movie  are  participating  in  a  

Stewart  McKie   -­‐  195  -­‐   As  of  15/08/2014  

valuable  feedback  loop.  And  managing  this  kind  of  feedback  loop,  once  a  
screenplay  is  in  production,  is  important  functionality  that  many  
screenwriting  applications  simply  ignore.  

3.4  Screenplay  as  Social  Network  Anchor  

Clearly  the  process  view  of  screenwriting  and  the  recognition  of  the  
collaborative  involvement  of  many  different  types  of  stakeholders  in  that  
process  confirm  that  a  social  networking  perspective  of  screenwriting  and  
the  screenplay  has  validity.  Here,  the  network  anchor/context  is  the  
screenplay  text  itself  and  it  is  the  changing  text  of  the  screenplay  that,  in  
effect,  functions  as  the  analog  of  the  Facebook  ‘wall’  or  Twitter  Tweet  
‘stream’.  In  this  sense  the  screenplay  text  also  functions  also  as  a  content  
hub  to  which  other  UGC  can  be  attached  to  create  social  capital  and  further  
enrich  the  screenplay  content  and  context.    

A  use  case  of  a  screenplay  acting  as  the  anchor  for  a  social  network  and  able  
to  benefit  from  social  networking  activity,  is  in  the  educational  domain  in  
the  activity  of  learning  screenwriting,  where  each  student  is  creating  
screenplays    (or  scenes  within  a  screenplay)  as  a  member  of  an  
undergraduate  or  postgraduate  screenwriting  class.  Another  example  is  the  
screenplay  anchoring  the  movie  production  team  or  ‘on-­‐set’  social  network,  
where  each  practitioner  tends  to  approach  the  script  from  their  specific  
production-­‐role  point-­‐of  view  (POV).    

In  both  cases  there  are  a  range  of  social  networking  activities  that  we  can  
expect  members  of  these  social  networks  to  be  engaged  in,  in  these  and  
other  screenplay  anchor  use  cases.  One  key  activity  is  that  of  peer  review  for  
qualitative  assessment  of  the  content.  

Clayton  (2006)  discusses  the  importance  of  good  peer  review  practice  in  the  
design  of  her  MA  in  Screenwriting  for  TV  and  Film  (Retreat  Programme).  This  
Royal  Holloway  (London)  postgraduate  course  encourages  the  development  
of  dyadic  relationships,  leverages  collaborative  ‘group  boxes’  of  four  

Stewart  McKie   -­‐  196  -­‐   As  of  15/08/2014  

students,  and  demands  class  peer  review  of  student  work,  with  some  of  this  
review  work  itself  being  tutor  marked.  Clayton  points  out  that  peer  review  
can  deliver  ‘insights  for  both  donor  and  receiver’  but  she  also  highlights  the  
risks  of  ‘subjective’  comment,  the  impact  of  ‘negative’  opinions  and  handling  
the  ‘conflicting’  views  of  the  peer  group.    

A  screenplay  is  often  the  result  of  an  act  of  collaborative  writing  (CW),  which  
Henderson  and  De  Silva  (2006:3)  describe  as  ‘the  process  of  multiple  authors  
producing  one  document,  by  writing  together  and  soliciting  one  another’s  
opinions  about  their  writing.’  The  authors  propose  a  narrative  based  
business  process  model  for  CW,  founded  on  Rhetorical  Structure  Theory  
(RST)  so  that  the  narrative  can  be  divided  within  a  team  to  help  to  enable  
the  best  people  within  a  team  to  work  on  the  narrative  sections  they  are  best  
suited  to.  They  emphasize  the  importance  of  author  rights  and  roles  and  
more  emphasis  on  narrative  version  control  to  facilitate  the  CW  process.  It  
should  be  noted  that  neither  task  allocation  to  roles  nor  version  control  
capabilities  are  well  supported  in  any  current  screenwriting  application.    

Takach  (2006)  approaches  screenwriting  from  a  knowledge  management    

(KM)  perspective  proposing  (section  4.1)  that  screenplays  create  a  ‘context  
for  sharing’  and  that  the  screenplay  acts  as  a  means  for  the  screenwriter  to  
communicate  knowledge  ‘messages’  to  an  audience  so  members  can  ‘act’  
upon  them.  Sharing  knowledge  for  learning  is  the  primary  objective  of  a  
‘community  of  practice’  (cf.  Etienne  Wenger  at  
http://www.ewenger.com/theory/),  itself  another  way  of  describing  the  
kind  of  social  network  that  may  exist  around  a  screenplay,  although  it  
perhaps  better  described  as  community  of  ‘interest’  or  ‘purpose’.  At  different  
times  during  the  screenplay  development  process,  different  communities  of  
purpose  emerge,  comprising  different  participants  and  roles  that  have  
different  priorities  e.g.  to  sell  the  screenplay,  to  attract  talent  to  the  
screenplay  project  or  to  produce  a  movie  from  the  screenplay.  

Stewart  McKie   -­‐  197  -­‐   As  of  15/08/2014  

Summary  

A  screenplay  that  is  either  ‘in-­‐play’  commercially  or  ‘in-­‐production’  (i.e.  

principal  photography  has  started)  is  at  the  centre  of  a  social  network,  
maybe  even  multiple  overlapping  networks,  albeit  rather  smaller  ones  than  
a  popular  Facebook  page  or  Twitter  stream  involve.  As  the  screenplay  
progresses  through  its  development  and  realization  process,  more  and  
different  stakeholders  within  the  network  make  use  of  the  screenplay  for  
different  purposes.    

As  the  anchor  of  the  network,  the  screenplay  can  become  the  focus  for  a  
number  of  social  networking  activities,  like  rating  and  commenting,  and  acts  
as  a  content  hub  that  benefits  from  user  generated  content  linked  to  the  
screenplay  by  the  network  participants.    Today,  most  screenwriting  
applications  do  not  reflect  this  social  network  dimension  by  providing  the  
functionality  required  to  allow  a  screenplay  to  be  effectively  co-­‐created  and  
socialized.    

Datafication  of  screenplay  content  facilitates  the  perspective  of  screenplay  

as  social  network  by  making  it  easier  to  interact  or  engage  with  a  script  at  a  
more  granular  level  than  the  ‘whole-­‐script’  document  level.  For  example  
social  network  participants  can  rate  and  comment  on,  and  attach  
enrichment  content  to,  script  content  at  the  scene,  role,  location  or  even  
snippet  level.    This  enables  the  generation  of  more  social  capital  around  the  
script  which  may  lead  to  a  qualitative  improvement  either  in  the  screenplay  
itself  or  in  the  process  of  realizing  the  script  as  a  movie.  

Stewart  McKie   -­‐  198  -­‐   As  of  15/08/2014  

4  Screenplay  as  Design  

The  prevailing  design  of  screenplays  is  no  longer  fit  for  purpose.    

Screenplay  as  design,  as  both  a  concept  and  practice  is  in  need  of  a  complete  
overhaul.  Screenplay  design  is  steeped  in  20th  century  legacy  practices  of  
writing,  publishing  and  distributing  scripts  on  paper.  The  most  obvious  
‘legacy’  design  element  being  the  use  of  a  Courier  12  fixed  font  typeface  to  
simulate  the  production  of  a  paper  artefact  on  an  obsolete  device  -­‐  the  
typewriter.    The  presentation  of  screenplay  text  generally  assumes  a  printed  
rather  than  online  textual  artefact.    Kindles  and  other  mobile  devices  as  a  
form  factor  and  ebook-­‐like  formats  as  a  delivery  mode  have  largely  passed  
screenplays  by.  

Even  if  a  screenplay  is  only  conceived  as  more  like  a  web  page  where  the  
text  also  acts  as  an  anchor  for  linked  content  then  the  screenplay  has  the  
potential  to  become  an  interactive  rather  than  static  text.  This  interactivity  
can  already  allow  individual  scenes  to  be  ‘played’  in  different  ways  
depending  on  the  linked  content  (e.g.  video  and  audio  clips):  for  example  for  
dialog  to  be  spoken  out  loud  or  locations  viewed  via  video  clips  rather  than  
just  described  in  scene  headings.    

This  is  not  to  deny  the  utility  of  the  printed  screenplay  delivered  on  paper.  
Script  supervisor  Sabi  Paisa  offers  proof  of  this  in  her  blog  post  Director’s  
Homework  (see  http://scriptsuper.wordpress.com/2013/04/09/directors-­‐
homework/about  -­‐  accessed  17  June  2013)  about  the  use  of  a  printed  
screenplay  by  Director  Ian  Barry  on  the  set  of  House  Husbands  II.  As  figure  
4.1  shows,  Barry  uses  a  paper-­‐based  ‘script  book’  to  plan  shots  and  generally  
annotate  the  script  using  his  own  personal  colour-­‐coded  ‘markup  language.’    

Personally  I  find  this  approach  very  compelling  as  a  means  to  capture  a  
single  individual’s  ideas  around  a  script  using  offline  media  (i.e.  a  printed  
book).    Applications  such  as  Peter  Skarrat’s  Script  Supervisor  software  (see  
http://peterskarratt.com  -­‐  accessed  17  June  2013)  can  replicate  some  of  this  
Stewart  McKie   -­‐  199  -­‐   As  of  15/08/2014  
script  lining  capability  automatically  but  without  the  artistic  flexibility  that  
plain  paper  naturally  offers  (see  figure  4.2).  
 

Figure  4.1  –  Ian  Barry’s  Script  Book  

 
 

Stewart  McKie   -­‐  200  -­‐   As  of  15/08/2014  

Figure  4.2  -­‐  A  Lined  Script  in  Script  Supervisor  

While  a  paper  script  is  clearly  portable  and  easy  to  annotate  manually,  an  
online  script  can  offer  the  significant  benefit  of  hyperlinking,  like  any  web  
page.  The  interactive  screenplay  needs  only  the  ability  to  add  a  hyperlink  
within  the  text  or  some  other  means  to  link  content  to  screenplay  data  or  
metadata.  These  links  may  be  to  external  web  content  (you  can  do  this  in  
any  script  written  in  any  popular  word  processor  (e.g.  Microsoft  Word)  or  to  
‘internal’  content  based  on  relationships  inherent  in  the  data  structure  of  a  
datafied  screenplay.      

Examples  of  these  external  &  internal  links  include:  

• popup  links  e.g.  to  popup  a  character  image  from  a  character  cue  or  a  
location  image  from  a  scene  heading  
 
• branch  links  e.g.  to  popup  alternative  scene  choices  at  the  start  or  end  
of  a  scene  
 
• comment  links  e.g.  to  list  a  series  of  comments    or  Tweets  posted  to  a  
scene  
 
• storyboard  links  e.g.  to  view  a  series  of  board  images  linked  to  a  
scene  
 
• clip  links  e.g.  to  play  an  audio/visual  clip  linked  to  a  scene  

This  kind  of  screenplay  linking  creates  a  form  of  Augmented  Reality  (AR).  AR  
layers  ‘virtual’  information  on  top  of  a  physical  location.  For  example  if  you  
point  your  smartphone  camera  at  a  hotel  on  a  street,  AR  may  layer  a  booking  
form  over  the  hotel.  Your  physical  position,  as  determined  by  GPS,  is  what  
enables  the  layering  on  of  content  to  augment  the  view  you  see.  In  a  
screenwriting  2.0  screenplay,  the  physical  reality  is  the  on  screen  text  of  the  
script,  the  virtual  reality  is  the  variety  of  content  linked  to  the  script  

Stewart  McKie   -­‐  201  -­‐   As  of  15/08/2014  

depending  on  the  specific  content  location  the  reader  is  ‘at’  in  the  script.  The  
GPS  co-­‐ordinates  of  latitude  and  longitude  that  drive  the  availability  of  AR  
content  from  the  viewing  perspective  of  a  mobile  phone  camera  become  the  
scene  and  snippet,  location  or  character  metadata  identifiers  in  a  screenplay  
text.    As  a  result,  voice,  image  or  video  clip  content  could  be  layered  over  the  
text  to  create  a  more  immersive  screenplay  viewing  experience.  

4.1  Data-­‐Driven  Screenplay  Manufacturing  

It  is  becoming  increasingly  clear  that  data  can  drive  the  design  of  almost  
anything.    The  emergence  of  computer-­‐aided  design  (CAD)  laid  the  
groundwork  for  computer-­‐  aided  manufacturing  (CAM)  whereby  digitized  
design  data  is  used  to  drive  computer  numerically  controlled  (CNC)  tools.  
Automobile  manufacturing  is  just  one  industrial  process  revolutionized  by  
CAD/CAM  and  the  use  of  industrial  robots  to  carry  out  production  tasks.  

Now,  the  recent  emergence  of  3D  printing  shows  that  the  potential  of  CAD  
continues  to  expand  into  the  domain  of  direct  digital  manufacture  (DDM).  
From  gun  parts  to  the  bioprinting  of  living  tissue  (see  
http://www.explainingthefuture.com/bioprinting.html  accessed  03  January  
2013)  3D  printing  is  widely  expected  to  generate  a  new  wave  in  CAD  driven  
innovation  at  a  personal  rather  than  industrial  level.    

In  2012,  I  bought  a  Cubify  3D  printer  (http://www.cubify.com)  and  ran  a  

series  of  four  after-­‐school  class  at  Shaftesbury  School  (Dorset)  in  February  
2013  to  introduce  a  small  group  of  students  to  3D  printing.  We  used  Google  
Sketchup  (http://www.sketchup.com)  to  design  the  objects  we  printed  and  
Cubify’s  own  software  to  convert  the  output  from  Sketchup  into  the  input  
files  required  by  the  Cubify  printer.  

Anderson  (2012)  envisions  3D  printing  as  enabling  a  new  generation  of  
individual  craftspeople  to  participate  in  a  new  kind  of  creative  activity  –  
CAD/CAM  for  the  person  rather  than  the  process.    My  brief  exposure  to  3D  
printing  led  me  to  consider  whether  the  screenplay,  and  especially  

Stewart  McKie   -­‐  202  -­‐   As  of  15/08/2014  

screenplay  metadata,  cannot  also  be  subject  to  some  kind  of  ‘manufacturing’  
process  in  the  same  way  that  coded  CAD  instructions  can  drive  the  printing  
of  a  3D  physical  object.  In  the  next  two  sections  I  discuss  two  applications  of  
3D  printing  techniques  to  the  design  and  production  of  screenplays  (not  I  
emphasize,  the  output  from  them  i.e.  movies)  that  involves  creating  a  
slightly  different  kind  of  ‘prototype’  to  those  discussed  in  Analysing  the  
Screenplay  (2011)  by  Ganz  (127-­‐141)  and  Millard  (142-­‐157).  

Subtractive  Production  

Subtractive  production  uses  the  digitized  CAD  file  of  an  object  to  drive  a  
process  that  reduces  the  volume  of  a  source  material  in  order  to  fashion  it  
into  the  target  object.  For  example,  using  a  machine  (e.g.  a  Roland  Modela)  
to  carve  a  curvaceous  fruit  bowl  from  a  square  block  of  wood.  This  has  an  
application  to  screenwriting,  specifically  in  terms  of  the  process  of  
adaptation.  

As  the  term  suggests,  in  the  adaptation  process  the  screenwriter  takes  a  
source  text  –  such  as  a  novel  or  theatre  play  text  –  and  adapts  it  into  a  
screenplay  for  realization  as  a  movie.  So  how  can  subtractive  production  be  
applied  here?  

First  we  start  with  data,  in  this  case,  for  example,  the  digitized  text  of  the  
novel  or  play  to  be  adapted.    This  data  can  simply  be  in  the  form  of  a  plain  
text  (.txt)  file.  Now  when  this  data  is  fed  into  a  specific  type  of  content  
analytics  tool  and  a  specific  technique  (algorithm)  applied,  the  tool  can  
return  a  subset  of  data  found  in  the  text  that  is  relevant  to  a  screenplay  
adaptation  process.  In  effect  the  tool  takes  the  text  (cf.  the  block  of  wood)  
and  subtracts  this  screenplay-­‐irrelevant  data  from  the  rest  of  the  text  (cf.  the  
‘scrap’  wood  shaved  off  by  the  subtraction  process)  and  leaves  the  
screenplay  relevant  data  (cf.  a  prototype  fruit  bowl  object  –  not  the  finished  
object  –  I  am  not  suggesting  the  output  will  be  a  complete  screenplay).    

The  content  analytics  technique  that  applies  here  is  not,  for  example,  

Stewart  McKie   -­‐  203  -­‐   As  of  15/08/2014  

predictive  analytics  or  sentiment  analytics  (both  discussed  elsewhere  in  this  
thesis)  but  entity  analytics.    The  focus  of  entity  analytics  is  on  identifying  so-­‐
called  ‘entities’  in  the  text:  for  example  people,  places  and  things  as  outlined  
by  Seth  Grimes  in  12  Things  the  Semantic  Web  Should  Know  about  Content  
Analytics  (2011:  3-­‐4)  

Entity  extraction  is  a  process  that  starts  by  finding  entities  in  source  
materials,  whether  web  pages,  email,  audio  streams,  images,  or  some  
other  material  of  interest.  Once  discerned,  the  entity  is  disambiguated  
(Is  “Ford”  a  car,  an  industrial  company,  an  actor  [which?],  a  theater,  or  
a  place  to  cross  a  river?).  Then  it  is  typed  (Person,  organization  etc.),  
and  (perhaps)  mapped  into  a  canonical  form  according  to  a  controlled  
vocabulary.  It  may  be  designated  with  a  uniform  resource  identifier  
that  facilitates  associating  diverse  information  to  the  source  material.  

Now  it  just  so  happens  that  people,  places  and  things  have  screenplay  
relevance  since  they  map  to  (potential)  characters,  locations  and  props.    So  
it’s  easy  to  see  how  running  a  text  file  through  entity  analytics  creates  a  
‘framework’  for  a  screenplay  by  automating  the  population  of,  say,  the  
character,  location  and  prop  tables  in  a  screenwriting  application  that  
utilizes  a  database  to  manage  datafied  screenplay  content.  

Naturally  you  will  not  want  all  the  characters  and  locations  and  props  
written  automatically  to  the  tables.  But  as  entity  analytics  is  also  capable  of  
counting  the  number  of  times  these  entities  appear  in  the  text,  the  adapting  
screenwriter  can  easily  intervene  and  select  say  the  top  20  or  50,  or  
whatever,  of  each  entity  to  write  to  the  tables.    

By  using  a  subtractive  production  technique  via  entity  analytics,  a  

‘framework’  of  certain  screenplay  metadata  is  now  available  to  use  as  the  
basis  for  the  adaptation  without  the  need  to  even  read  the  novel  or  play  that  
is  being  adapted.    Furthermore  it  would  also  be  relatively  easy  to  link  some  
or  all  of  these  identified  entities  to  other  ‘rich  content’  located  on  the  Web  to  
kick-­‐start  the  research  process  and  link  this  additional  content  to  the  
entities.  

Stewart  McKie   -­‐  204  -­‐   As  of  15/08/2014  

Additive  Production  

Unlike  subtractive  printing,  ‘additive’  3D  printing  builds  3D  objects  in  layers:  
Very  thin  ‘slices’  of  material  are  layered  on  top  of  each  other  from  the  
bottom-­‐up  until  the  physical  object  is  fully  realized  from  the  CAD  plan  that  
has  been  preprocessed  into  a  stereolithography  file  (e.g.  .STL  file).  
Screenplays  can  be  also  considered  from  the  same  additive  perspective  (see  
figure  4.3).    

For  example,  using  a  structural  framework  such  as  Vogler’s  Hero’s  Journey  a  
screenplay  can  first  be  populated  with  a  ‘sequence’  layer  corresponding  to  
Vogler’s  12  sequences  mapped  across  the  three  acts  of  Vogler’s  ‘ordinary’  
and  ‘special’  worlds.    It  is  then  up  to  the  writer  to  ‘add’  the  locations  and  
scenes  to  ‘flesh-­‐out’  these  sequences.  Similarly,  Vogler’s  eight  character  
archetypes  can  also  be  populated  into  the  screenplay  as  ‘template’  roles  and  
again  the  writer  can  choose  to  use  all  or  some  and  name  the  individual  roles  
as  one  or  more  characters,  as  appropriate  for  the  story  she  is  trying  to  tell.    

Onto  these  layers  is  added  the  ‘entity’  layer  comprising  actual  character  
roles  and  locations  and  individual  scenes  that  involve  these  automatically  
added  characters  and  sequences.  The  ‘snippet’  layer  populates  individual  
scenes  with  dialog  and  action  description  until  the  screenplay  is  fully  
realized  and  then  more  layers  are  added  during  production  to  create  the  
script  storyboard,  shot-­‐list  or  production  breakdown  reports  for  example.  In  
this  way  a  script  can  be  viewed  as  being  ‘developed’  subject  to  an  additive  or  
layering  process.  

Stewart  McKie   -­‐  205  -­‐   As  of  15/08/2014  

Figure  4.3  -­‐  Screenplay  as  Additive  Layers  

However,  all  this  talk  of    ‘automated  screenwriting’  is  likely  to  be  anathema  
to  those  for  whom  screenwriting  is  perceived  primarily  as  a  creative  process  
and  ‘manufacturing’  screenplays  seen  as  a  negative  direction  that  will  result  
in  formulaic  writing  that  delivers  repetitive  experiences  for  audiences,  as  
Tom  Gauld’s  cartoon  from  The  Guardian  (review  section  23/03/2013)  in  
figure  4.4  suggests:  

Figure  4.4  –  Gauld’s  Structural  Journey  

And  many  may  regard  this  whole  exercise  as  too  trivial  to  care  about.  After  
all,  the  subtractive  use  of  entity  analysis  hardly  compares  to  the  richly  
creative  adaptation  process  presented  in  Adaptation  (2002,  Spike  Jonze)  and  
if  the  additive  screenplay  framework  is  based  on  a  structure  that  has  been  
characterized  as  a  kind  of    ‘infantile  bromide’  (Langford  2011:256)  it’s  
hardly  likely  to  be  selected  as  a  start  point  for  a  new  screenplay.  But  the  
reality  is  that  the  subtractive  and  additive  manufacturing  techniques  of  3D  
printing  can  be  applied  to  screenplay  metadata  generation  and  perhaps  over  
time,  ‘auto-­‐populating’  a  screenplay  with  more  or  less  metadata  could  
become  the  normal  way  to  start  the  ‘design’  of  a  screenplay  in  a  
Screenwriting  2.0  application.    

Stewart  McKie   -­‐  206  -­‐   As  of  15/08/2014  

Screenplay  as  Blueprint  

Silver-­‐Lasky  (2004:53)  characterizes  screenplays  as  a  ‘set  of  instructions’  

others  as  a  ‘blueprint’  (see  discussion  in  Price  2010:44-­‐47).  Ganz  thinks  
‘blueprint’  is  an  inadequate  metaphor  (2011:1431)  (but  does  see  a  role  for  
the  screenplay  as  a  ‘program  for  a  sequence  of  interactions  between  the  
reader  and  the  text’  (2011:  131))  as  does  Millard  (2010:14):  

…it  seems  a  less  than  ideal  metaphor  for  the  screenplay.  The  
development  of  the  screen  idea  inevitably  involves  collaboration,  and  
therefore  to  concentrate  solely  on  the  screenplay  as  a  source  for  the  
film-­‐to-­‐be  seems  unnecessarily  restrictive.  

Price  (2010:45)  is  also  not  a  fan  of  the  blueprint  metaphor  saying,  ‘In  its  
literal  sense,  a  blueprint  is  a  projection  of  a  design  for  a  material  object’  and  
that  ‘the  insidious  connotations  of  the  literal  meaning  [of  screenplay  as  
blueprint]  have  proved  persistently  damaging.’  

But  how  long  can  that  continue  to  be  a  realistic  attitude  given  the  increasing  
sophistication  of  technology  that  can  already  leverage  screenplay  texts  as  
design  blueprint  ‘inputs’  from  which  visual  ‘output  products’  are  
automatically  generated?  

Storyboarding  and  pre-­‐visualization  (‘previz’)  software  such  as  Storyboard  

Quick  and  FrameForge  3D  already  use  screenplay  data  as  the  ‘design’  input  
for  generating  2D  storyboards  or  basic  3D  animations  from  the  textual  
screenplay  content.  The  screenplay  file  is  imported  and  a  storyboard  
template  or  foundation  pre-­‐viz  automatically  produced  from  an  
interpretation  of  the  screenplay  data  and  metadata.  For  example  
FrameForge  (see  http://www.frameforge3d.com/Products/Core-­‐Version/  -­‐  
accessed  09  July  2013)  claim  that:  ‘FrameForge  Previz  Studio  can  
automatically  generate  blank  sets  for  every  unique  location,  and  you  can  
associate  your  stored  shots  with  text  in  the  script  on  a  line-­‐by-­‐line  basis…’  

FrameForge  then  lets  users  additively  enhance  this  pre-­‐viz,  for  example  by  
adding  background  images  and  3D  objects  or  people  from  its  digital  asset  

Stewart  McKie   -­‐  207  -­‐   As  of  15/08/2014  

library.  Storyboards  and  previz  animations  perform  a  function  analogous  to  
a  product  prototype  –  lying  somewhere  between  the  product  plan  and  the  
finished  object  (i.e.  the  filmed  movie).    

4.2  Screenplay  Models  

To  date,  the  discourse  around  screenplay  as  design  has  been  unduly  focused  
on  structure,  which  is  only  one  element  of  design.  For  example,  the  sitemap  
structure  of  a  website  is  integral  to  its  design  but  perhaps  more  important  to  
the  audience  (rather  than  the  website’s  information  architect)  is  the  user  
interface  or  ‘look  and  feel’  of  the  site  including  the  colours,  icons  and  
typefaces  used.      

The  screenwriting  structural  debate  has  focused  on  popular  ‘paradigms’  

such  as  Field’s  3  Act  structure  or  Vogler’s  Hero’s  Journey  and  whether  or  not  
to  proactively  use  acts  and  sequences  and  storylines  as  script  building  
blocks.  But,  in  practice,  all  these  so-­‐called  screenwriting  ‘paradigms’  are  
little  more  than  ways  of  grouping  scenes  together  into  dramatic  units,  
emphasizing  specific  plot  points  and  character  throughlines:  they  are  not  
comprehensive  designs  but,  as  suggested  above,  outline  ‘metadata’  
frameworks,  analogous  to  paint-­‐by-­‐numbers  for  artwork  generation  or  the  
steel  girders  of  a  tower  block  construction  to  which  everything  else  is  
connected.  

There  are  other  story  models,  at  varying  levels  of  granularity,  which  could  
be  used  to  drive  screenplay  design.  For  example  Soulier  and  Caussanel’s  
(2002)  model  of  the  narrative  and  its  narrative  ‘atoms’:  Situation,  
complication,  resolution  and  result.  Or  Chris  Huntley’s  Dramatica,  a  highly  
detailed  theory  of  story  (http://dramatica.com/theory/articles/Dram-­‐
differences.htm  -­‐  accessed  25  February  2013).  In  Huntley’s  article  How  and  
Why  Dramatica  is  Different  from  Six  Other  Story  Paradigms  
(http://dramatica.com/articles/how-­‐and-­‐why-­‐dramatica-­‐is-­‐different-­‐from-­‐
six-­‐other-­‐story-­‐paradigms  -­‐  accessed  25  February  2013)  he  also  outlines  
‘story  paradigms’  from  Huage,  McKee,  Seger  and  Truby.  All  of  which  share  

Stewart  McKie   -­‐  208  -­‐   As  of  15/08/2014  

similar  features  including  scenes  grouped  into  sequences,  sequences  
grouped  into  acts,  character  throughlines  and  specifically  identified  turning  
or  plot  points.  Yorke  (2013:256)  has  also  provided  a  useful  summary  of  12  
structural  paradigms  (not  including  his  own  version,  the  ‘3D  Roadmap  of  
Change’  (2013:64))  as  shown  in  figure  4.5  below:  

Figure  4.5  –  Yorke’s  Paradigm  Summary  

As  Yorke’s  summary  shows,  the  3  Act  structure  has  been  adapted  into  4  (3  
act  with  a  two-­‐part  second  act),  5  (Shakespeare’s  favourite)  and  even  8  act  
structures,  with  lots  of  different  interpretations  of  the  role  of  each  act  in  the  
narrative.  Yorke  also  thinks  that  some  of  those  who  decry  the  influence  of  3  
act  structures  –  he  quotes  Guillermo  Del  Toro,  David  Hare  and  Kaufman  as  
examples  (2013:xv)  –  ‘protest  to  much’  and  ‘however  much  they  hate  
it…they  can’t  help  but  follow  a  blueprint  they  profess  to  distrust’.  

Stewart  McKie   -­‐  209  -­‐   As  of  15/08/2014  

Another  contribution  to  screenplay  as  design  could  be  provided  by  the  
automated  application  of  specific  genre  conventions  to  the  screenwriting  
process  to  ensure  that  a  screenplay  ‘feels  like’  other  familiar,  released  
screenplays  considered  to  be  in  the  same  genre.  Conformance  to,  or  
subversion  of,  genre  conventions  could  be  considered  to  be  a  design  
‘objective’  that  a  screenwriter  could  espouse  when  writing  a  script.  But  then  
that  also  might  lead  to  sterile  writing  if  you  agree  with  David  Hare  that  ‘All  
great  work  is  now  outside  genre’,  as  quoted  in  Yorke  (2013:  xv).  

To  discuss  screenplay  as  design  we  must  also  consider  two  perspectives  in  
particular:  design  patterns,  as  they  apply  to  screenplays,  and  digital  
screenplay  content  manufacturing  i.e.  screenplay  content  generated  
automatically  from  a  set  of  design  patterns.  The  intention  here  is  not  to  
suggest  that  screenplay  designs  can  entirely  replace  human  creativity  but  
that  they  may  be  able  to  automate  some  of  the  more  mechanical  aspects  of  
screenwriting.    

4.3  Screenplay  Pattern  Language  

Christopher  Alexander  proposed  design  patterns  in  the  domain  of  

architecture  in  his  book,  A  Pattern  Language:  Towns,  Buildings,  Construction  
(1977),  in  which  he  states  that:    

Each  pattern  describes  a  problem  which  occurs  over  and  over  again  in  
our  environment,  and  then  describes  the  core  of  the  solution  to  that  
problem,  in  such  a  way  that  you  can  use  this  solution  a  million  times  
over,  without  ever  doing  it  the  same  way  twice.

(http://www.patternlanguage.com/leveltwo/caframe.htm?/leveltwo/../bio
s/designpatterns.htm,  accessed  January  15  2013).  

The  concept  was  subsequently  adopted  for  use  in  software  engineering  in  
Design  Patterns:  Elements  of  Reusable  Object-­‐Oriented  Software  (1994)  by  the  
so-­‐called  ‘gang-­‐of-­‐four’  –  Gamma,  Helm,  Johnson  &  Vlissides.  An  organization  
of  design  patterns  specific  to  a  particular  domain  is  referred  to  as  a  pattern  
language.  

Stewart  McKie   -­‐  210  -­‐   As  of  15/08/2014  

A  pattern  can  be  described  using  a  set  of  attributes,  such  as  those  outlined  
for  software  patterns  by  Maioriello  (2002)  
(http://www.developer.com/design/article.php/1474561/What-­‐Are-­‐
Design-­‐Patterns-­‐and-­‐Do-­‐I-­‐Need-­‐Them.htm  -­‐  accessed  03  January  2013).  

Pattern  Name   Describes  the  essence  of  the  pattern  in  a  short,  but  
expressive,  name  

Intent   Describes  what  the  pattern  does  

Also  Known   List  any  synonyms  for  the  pattern  

As  

Motivation   Provides  an  example  of  a  problem  and  how  the  pattern  
solves  that  problem  

Applicability   Lists  the  situations  where  the  pattern  is  applicable  

Structure   Set  of  diagrams  of  the  classes  and  objects  that  depict  the  
pattern  

Participants   Describes  the  classes  and  objects  that  participate  in  the  
design  pattern  and  their  responsibilities  

Collaborations   Describes  how  the  participants  collaborate  to  carry  out  

their  responsibilities  

Consequences   Describes  the  forces  that  exist  with  the  pattern  and  the  
benefits,  trade-­‐offs,  and  the  variable  that  is  isolated  by  

Stewart  McKie   -­‐  211  -­‐   As  of  15/08/2014  

 the  pattern  

Table  4.1  –  Pattern  Attributes  

To  see  how  this  could  be  applied  to  screenplays,  let’s  take  the  3  Act  structure  
and  attempt  to  describe  it  in  terms  of  pattern  attributes:  

Pattern  Name   3  Act  Structure  (Field,  1979)  

Intent   A  means  of  dividing  up  screenplay  

content  into  a  balanced  set  of  scenes  
that  reflect  Aristotle’s  notion  of  a  
beginning,  middle  and  end.    

Also  Known  As   The  ‘paradigm’  

Motivation   To  enable  an  audience  to  

understand  the  story  world  they  are  
in,  to  follow  the  complications  and  
conflicts  of  the  story  narrative  and  
its  characters,  and  to  be  satisfied  by  
a  resolution  to  the  story  narrative.      

Applicability   Conventional  linear  narrative  

Structure   Groups  of  scenes  to  reflect  a  one-­‐

quarter,  two  quarter,  one  quarter  
type  content  balance  in  the  3  acts  
with  an  ‘inciting  incident’  scene  
somewhere  around  the  middle  of  
the  first  act  and  a  ‘mid  point’  scene  

Stewart  McKie   -­‐  212  -­‐   As  of  15/08/2014  

 somewhere  around  the  middle  of  
the  second  act  and  a  ‘climax’  scene  
somewhere  around  the  middle  of  
the  third  act.  

Participants   All  the  main  entities  of  a  screenplay,  

namely:  scenes,  locations  and  
characters.  

Collaborations   The  interaction  of  characters  in  both  

the  action  and  dialog  elements  of  the  
screenplay  scene  content.  

Consequences   Complication  of  action  and  conflict  

of  characters.  The  benefit  of  a  
‘balanced’  script  from  a  timing  
perspective  for  the  audience.  A  trade  
off  of  structural  predictability  vs.  
structural  novelty.  Field’s  three  key  
scenes  as  the  variables  isolated  by  
the  pattern.  

Table  4.2  –  3-­‐Act  Structure  Pattern  

On  a  similar  basis,  how  about  the  notion  of  a  ‘protagonist’  pattern?  

Pattern  Name   Protagonist    

Intent   A  create  a  focal  character  in  the  

screenplay  whom  audiences  can  
relate  to  or  root  for.    

Stewart  McKie   -­‐  213  -­‐   As  of  15/08/2014  

Also  Known  As   The  ‘Hero’  

Motivation   To  facilitate  the  ability  of  an  

audience  to  focus  on  following  the  
narrative  path  of  a  single  character  
through  the  movie  from  beginning  
to  end.      

Applicability   All  character-­‐driven  screenplays  

that  do  not  feature  an  ensemble  
cast.  

Structure   A  thruline  that  reflects  the  

importance  of  the  character  to  the  
narrative  of  the  story.  

Participants   Normally  a  single  character  or  two  

characters  acting  as  a  team  who  in  
effect  function  as  a  single  
protagonist.  

Collaborations   The  interaction  of  the  protagonist  

with  other  characters  in  the  
screenplay,  especially  the  
Antagonist.  

Consequences   Complication  of  narrative  events  

and  character’s  composition  and  
conflict  of  protagonist  vs.  
antagonist.  The  benefit  of  a  focal  

Stewart  McKie   -­‐  214  -­‐   As  of  15/08/2014  

 character  for  the  audience.  A  trade  
off  of  expectation  of  success  vs.  
expectation  of  failure.  Character  
veracity  as  the  variable  isolated  by  
the  pattern.  

Table  4.3  –  Protagonist  Pattern  

Unlike  many  other  design  activities  –  manufacturing  an  aircraft  or  building  a  
house  for  example  –  screenwriting  does  not  typically  involve  forward-­‐
engineering  or  using  a  design  pattern  to  drive  the  manufacturing  of  
something.  But  there  are  some  candidates  for  design  patterns  that  could  
make  up  the  pattern  language  of  a  screenplay  to  drive  the  manufacture  of  a  
screenplay  in  some  automated  fashion.  

A  key  pattern  is  that  of  the  scene  and  the  scene  groupings  and  characters  
that  are  instantiated  to  deliver  structural  paradigms  such  as  Field’s  3-­‐Acts  or  
Vogler’s  Hero’s  Journey.  These  paradigms  can  be  reused,  as  patterns,  to  
manufacture  the  basis  of  a  screenplay  in  Alexander’s  words  -­‐  without  ever  
doing  it  the  same  way  twice  –  where  ‘doing  it’  means  populating  the  
screenplay  metadata  framework  generated  from  the  pattern  with  scene  
content  data  -­‐  thus  preserving  the  essential  creativity  of  screenwriting.  

If  a  writer  sets  out  to  write  a  ‘Hero’s  Journey’  screenplay,  he  knows  that  
potential  elements  of  this  pattern  include  a  set  of  12  sequences  of  scenes  or  
‘stages’  in  Vogler’s  terms  (Vogler,  1996:14)  and  a  prospective  cast  of  8  
character  archetypes  (1996:31)  plus  the  ‘hero’  (protagonist)  and  antagonist  
characters.  So  if  I  pick  this  pattern  as  a  ‘design’  for  my  screenplay,  I  should  at  
least  expect  that  screenwriting  software  applications  provide  the  means  to  
use  a  template  to  generate  an  empty  screenplay  framework  already  
provisioned  with  12  sequences  and  10  identifiable  character  types.  These  
sequences  in  turn  must  then  be  provisioned  with  scenes  by  the  screenwriter  
in  order  to  enable  the  characters  to  interact.  Like  the  concrete  foundation  
and  4  corner  posts  of  a  building,  this  screenplay  framework,  generated  from  
Stewart  McKie   -­‐  215  -­‐   As  of  15/08/2014  
a  pattern,  just  provides  a  base  to  work  on,  the  actual  ‘content’  of  the  building  
could  end  up  functioning  as  pretty  much  anything  from  an  art  gallery  to  a  
power  station.    

To  my  knowledge,  no  screenwriting  software  does  this  simple  job  of  
generating  a  screenplay  framework  based  on  a  known  pattern  like  the  
Hero’s  Journey  (other  than  perhaps  Chris  Huntley’s  Dramatica  –  designed  to  
specifically  support  his  story  pattern).  Presumably  the  argument  is  that  
nobody  actually  writes  screenplays  this  way  or  possibly  many  screenwriters  
agree  with  Langford’s  rather  extreme  opinion  of  the  three-­‐act  structure  as  
‘tyranny’  and  Vogler’s  Hero’s  Journey  as  ‘infantile  bromide’  in  Analysing  the  
Screenplay  (2011:256).    

Baboulene  argues,  correctly  in  my  opinion,  that  ‘Structure  is  a  consequence  
of  our  words,  not  a  template  for  our  words’  and  that  ‘Structure  is  an  
outcome  of  the  creative  process.’  (2010:  32).    In  his  opinion,  structure  is  not  
useful  to  write  a  screenplay  –  as  a  design  pattern  -­‐  but  as  an  analytic  tool  to  
help  rewrite  it.  As  he  says  (2010:  33):  ‘This  is  the  main  value  of  structure:  
fault  finding  and  optimization  after  the  creative  outpouring  is  complete  and  
we  are  in  rewrite  mode.’  

Alien  Screenwriter  Dan  O’Bannon  also  offers  the  perspective  of  structure  as  
an  ‘abstraction’  from  the  text  (2013:  22-­‐23).  He  considers  story  structure  to  
be  ‘an  invisible  construct  that  defines  the  relationships  between  parts  of  the  
story’  and  that  ‘The  only  way  to  detect  a  story’s  structure  is  for  a  
knowledgeable  person  to  examine  the  story  and  infer  the  structure  from  the  
story’s  visible  parts.  ‘  And  because  structure  helps  to  show  the  screenwriter  
what’s  missing  and  where,  O’Bannon  regards  it  positively  -­‐  as  an  
‘empowering’  tool  for  the  screenwriter.  

Yorke  quotes  McGovern’s  similar  viewpoint  (2013:  77-­‐78):    ‘You  write  a  

script  twice.  The  first  time  you  pour  out  all  your  passion,  anger,  energy  and  
frustration.  Then  you  go  back  and  write  it  with  your  head.  ‘  

Stewart  McKie   -­‐  216  -­‐   As  of  15/08/2014  

Writing,  or  rather  rewriting  ‘with  your  head’,  is  where  structure  comes  into  
play  as  a  remedial  rather  initiating  process.  Cron  (2102:76-­‐77)  is  also  not  a  
big  fan  of  story  structures  when  used  to  mandate  that  ‘certain  external  
events  must  happen  at  certain  specific  points  in  the  story’.  She  considers  this  
to  cause  stories  to  be  written  from  the  ‘outside-­‐in’  because  they  cause  
writers  to  ‘craft  plots  in  which  these  events  occur  rather  than  crafting  
protagonists  whose  internal  progress  depends  on  said  events  occurring.’  

Baboulene,  O’Bannon,  McGovern  and  Cron’s  perspectives  imply  that  these  

kinds  of  structural  paradigms  or  design  patterns  are  most  useful  to  ‘reverse-­‐
engineer’  out  of  screenplays  for  analysis  and  rewriting  purposes  rather  than  
‘forward-­‐engineer’  into  screenplays  for  manufacturing  or  initial  writing  
purposes.  You  do  not  ‘write  to’  a  structure  but  ‘rewrite  into’  a  structure.  
However,  this  argument  puts  screenwriting  out  of  sync  with  most  other  
kinds  of  manufacturing  as  few  people  in  the  real  commercial  world  –  and  
writing  movies  occurs  in  a  real,  commercial  world  -­‐  design  output  products  
or  services  with  no  specifically  structured  output  product  or  service  in  mind.    

But  what  these  perspectives  on  screenplay  structure  do  emphasize  is  that  
software  could  be  useful  as  a  means  to  facilitate  the  uncovering  of  structure  
implicit  in  the  design.  This  surfacing  of  structure  from  the  text,  via  diagnostic  
tools  provided  by  screenwriting  software,  could  provide  a  useful  rewriting  
tool  for  the  writer  and  a  quick  way  for  script  readers  and  other  ‘evaluators’  
to  easily  ‘get’  the  structure  of  a  script.  Perhaps  this  kind  of  analysis  could  
generate  a  simple  infographic  visualization  like  this  one  that  attempts  to  
summarize  the  story  of  Alien  using  just  a  handful  of  icons:  

Stewart  McKie   -­‐  217  -­‐   As  of  15/08/2014  

  

Figure  4.6  –  Alien  as  visualized  by  Milesi  &  Matteo  (2012:9)  

But  it  is  probably  unlikely  to  expect  that  screenwriting  software  will  be  able  
to  automatically  generate  the  rich  infographic  in  figure  4.7  that  charts  
selected  episodes  of  the  Inspector  Montelbano  TV  series  from  1994-­‐2012  
(see  http://cargocollective.com/federicafragapane/LA-­‐LETTURA-­‐
CORRIERE-­‐DELLA-­‐SERA  -­‐  accessed  09  July  2013)  anytime  soon:  

Stewart  McKie   -­‐  218  -­‐   As  of  15/08/2014  

Figure  4.7  -­‐  Inspector  Montalbano  Infographic  

Another  key  pattern,  already  ‘encapsulated’  in  the  pattern  discussed  above,  
is  that  of  the  Protagonist  and  Antagonist.  Most  screenplays  have  one  of  each  
and  since  this  is  a  likely  pattern,  again  it  should  be  easy  for  software  to  
prepopulate  the  script  with  a  protagonist  and  antagonist  characters.  By  
doing  this,  the  software  helps  to  encourage  the  screenwriter  to  focus  early  
on  who  these  characters  are,  their  motivation  and  backstory  and  the  crucial  
interactions  of  these  two  principal  character  types.  

Some  screenwriting  packages,  like  StoryO  from  Jungle  Software  Software,  

use  a  series  of  questions  to  help  screenwriters  to  outline  and  flesh-­‐out  their  
story,  its  events  and  characters.  The  questioning  paradigm  is  used  to  design  
the  framework  of  the  screenplay  and  ensure  that  the  work  is  structurally  
‘thought-­‐through’  before  any  actual  creative  writing  takes  place.  

So  what  we  can  already  foresee  happening  is  that  by  thinking  in  terms  of  
design  patterns  as  a  means  to  manufacture  screenplays,  we  can  propose  
how  the  screenwriting  software  might  help  to  automate  and  execute  the  
implementation  of  these  patterns  in  the  application.    

4.4  Reverse-­‐Engineering  Screenplay  Design  

Reverse-­‐engineering  screenwriting  strategies  and  tactics  from  the  released  

movie  is  a  popular  form  of  academic  and  movie  critic  analysis.  But  it’s  
important  to  remember  that  many  versions  of  screenplays  available  for  
movies  that  are  found  on  the  Internet  are  not  in  fact  the  actual  version  of  the  
screenplay  that  is  reflected  in  the  released  movie.  This  version  is  only  
available  when  a  transcript  of  the  movie  has  been  produced,  since  even  a  
‘shooting’  or  ‘continuity’  script  document  is  unlikely  to  reflect  the  final  result  
of  the  film  editing  process.      

Recently,  tools  have  been  released  that  deliver  analytics  directly  from  digital  
video.  For  example,  digital  analysis  of  storylines,  say  within  multi-­‐episode  
TV  series,  has  been  taken  to  the  next  step  by  the  French  researchers  

Stewart  McKie   -­‐  219  -­‐   As  of  15/08/2014  

Ercolessi,  Sénac  and  Bredin  and  their  web  service  StoryVisualizer  or  StoViz  
(http://www.irit.fr/recherches/SAMOVA/ERCOLESSI/StoViz/#    -­‐  accessed  
January  15    2013).  StoViz  analyzes  a  digital  video  feed  to  identify  and  isolate  
storylines  from  the  content  by  ‘de-­‐interlacing  based  on  scene  clustering’.  In  
essence  the  software  attempts  to  semantically  group  related  scenes  into  
storylines  using  a  number  of  clustering  algorithms.  Therefore  from  a  single  
feed  of  a  movie’s  digital  file,  proposed  storylines  can  be  ‘pulled  out’  and  
presented  on  their  own  timeline  in  order  to  better  understand  the  various  
storylines  encapsulated  within  a  single  video  narrative.  

This  frame-­‐by-­‐frame  video  analysis  approach  is  analogous  to  a  Magnetic  

Resonance  Imaging (MRI)  scan,  with  the  visual  frame  content  equating  to  
MRI  scan’s  visual  ‘slice’  content.  Here  a  whole  movie  scan  (cf.  a  ‘whole  body’  
scan),  is  used  to  discover  storylines  based  on  frame  content  in  much  the  
same  way  as  a  doctor  looks  for  signs  of  disease  or  other  anomalies  in  an  MRI  
scan  slice.  The  storylines  are  plotted  onto  a  timeline  in  much  the  same  way  
as  the  narrative  of  diseased  tissue  could  be  visualized  from  a  series  of  slices  
from  an  MRI  scan.  

The  screenshot  from  StoViz  (figure  4.8)  shows  how  the  selected  digital  feed  -­‐  
episode  2  from  US  TV  series  Malcolm  in  the  Middle  -­‐  can  be  de-­‐interlaced  into  
3  storylines,  each  of  which  can  be  played  individually  by  the  viewer  user  if  
required.  

Stewart  McKie   -­‐  220  -­‐   As  of  15/08/2014  

Figure  4.8  -­‐  StoViz  Storyline  Visualizer  

Attempting  the  same  reverse  engineering  of  storylines  from  a  screenplay  

text,  as  opposed  to  a  digital  movie  file,  isn’t  necessarily  that  much  easier.  The  
easiest  way  to  identify  storylines  is  obviously  for  the  writer  to  flag  scenes  in  
the  screenplay  that  belong  to  a  specific  pre-­‐defined  storyline  and  then  
provide  a  simple  visualization  of  the  scenes  in  that  storyline  and  a  means  to  
‘drilldown’  to  the  scene  content  at  the  click  of  a  mouse.  But  what  if  the  writer  
wants  the  software  to  identify  storylines  that  ‘emerge’  from  the  text  as  
opposed  to  those  already  defined  within  it?  

Then,  variants  of  some  the  techniques  used  by  StoViz,  such  as  speaker  
diarization  and  automatic  speech  recognition  come  into  play.    At  a  simplistic  
level  this  could  mean  that  a  potential  storyline  can  be  partially  determined  
by  commonality  between  scenes  of:  

• The  characters  in  a  scene  

• The  fact  that  characters  speak  to  one  another  or  take  part  in  action  
together  
• The  scene’s  location  
• The  scene  time  of  day  and  internal/external  classification    
• Props  used  in  a  scene  
• The  way  characters  speak  (e.g.  their  vocabulary  or  syntax)  
 
For  example,  the  scenes  that  make  up  the  storyline  in  Alien  of  the  clash  of  
wills  between  Ripley  and  Ash  could  be  largely  determined  by  the  software  
identifying:  
 
• Scenes  that  involve  both  Ripley  and  Ash  with  a  preference  for  scenes  
that  only  involve  these  two  characters  
• Scenes  that  involve  dialog  snippets  between  Ripley  and  Ash  
• Scenes  that  take  place  in  Ash’s  ‘domain’  i.e.  the  sick  bay  

As  with  all  analytics,  even  this  kind  of  simplistic  storyline  analysis  can  be  

Stewart  McKie   -­‐  221  -­‐   As  of  15/08/2014  

used  both  to  confirm  that  writer-­‐intended  storylines  emerge  from  the  
screenplay  text  and  to  identify  storylines  that  emerge  from  the  text  that  the  
writer  may  have  been  unaware  of  and  wish  either  to  minimize  or  develop  
further  in  order  to  change  the  overall  design  of  the  screenplay.  

Russell  Chun’s  Story  Visualizer  is  another  tool  that  reverse  engineers  
analytics  from  digital  movie  files.  (see  
http://www.russellchun.com/storystructure/storyvisualizer.html,  accessed  
15  January  2013)  According  to  Chun,  what  he  does  is:  

 “measure  the  level  of  excitement  or  drama  over  time  by  combining  two  
indices.  One  index  tracked  audio  levels,  assuming  that  dramatic  moments  are  
accompanies  by  louder  sounds  (explosions,  shouting,  musical  crescendos).  The  
second  index  tracked  changes  in  color,  assuming  that  dramatic  moments  are  
also  marked  by  rapid  visual  changes  on  screen  (subject  or  camera  motion,  
quick  edits).  The  combined  “drama”  index  is  plotted  to  show  the  movie’s  unique  
fingerprint”.    

The  screenshot  in  figure  4.9  shows  a  resulting  plot  from  the  analysis  of  the  
movie  footage  from  a  gladiatorial  fight  scene  in  Gladiator  (2000).  

In  fact,  this  kind  of  analysis  is  more  difficult  to  do  from  the  screenplay  text  
than  from  the  digital  movie  file  as  many  screenplays  will  simply  not  include  
enough  identifiable  metadata  relating  to  audio  levels  or  color  changes  to  
enable  this  kind  of  text  analytics  to  succeed.  

Stewart  McKie   -­‐  222  -­‐   As  of  15/08/2014  

  

Figure  4.9  –  Chun’s  Story  Visualizer  

In  fact  reverse-­‐engineering  many  of  the  de-­‐facto  standard  strategies  and  

tactics  of  screenwriting  from  a  screenplay  text  is  not  that  easy  to  do.  For  
example,  the  ‘public’  demo  script  on  my  software  deliverable  Scenepad  is  an  
iteration  of  Dan  O’Bannon’s  screenplay  for  Alien  (1979).  Alien  exemplifies  
many  of  the  standard  strategies  and  tactics  of  screenwriting  such  as:  

• an  inciting  incident  (the  distress  call  that  diverts  their  set  course)  
• a  turning  point  (when  the  Alien  emerges  from  Kane’s  chest)  
• raised  stakes  (the  emergence  of  Ash  as  an  antagonist  to  Ridley)  
• conflict  (Alien  vs.  humans,  upstairs  and  downstairs  crew)  
• complications  (Ash’s  secret  mission,  Alien’s  acid  blood)  
• climax  (time-­‐sensitive  ‘escape’  of  Ridley  to  the  evacuation  shuttle)  
• resolution  (Ridley  safe  in  shuttle  and  Alien  apparently  destroyed)  
• genre  conformance  (horror  -­‐  monster  loose  in  an  enclosed  space)  

So  how  can  these  kinds  of  standard  screenwriting  tactics  be  reverse  
engineered  from  the  screenplay  text?  As  an  example,  let’s  review  Alien’s  
‘turning  point’  from  this  perspective.  

Stewart  McKie   -­‐  223  -­‐   As  of  15/08/2014  

4.5  Event  Indicators  

One  structural  analytic  focus  might  be  to  try  to  locate  a  turning  point  in  the  
screenplay  by  identifying  what  changes  that  is  of  significance  –  what  
linguistic  marker  can  be  defined  as  an  ‘event  indicator’  in  the  text?    

In  the  past,  manual  textual  analytics  techniques,  such  as  concordances,  were  
used  to  analyze  Shakespeare  plays,  today  technology  can  help.  For  example,  
technology  has  made  navigating  texts  by  reference  to  a  specific  term  easier,  
such  as  the  mentions  of  ‘blood’  in  Macbeth  (see  http://www.shakespeare-­‐
navigators.com/macbeth/Blood.html,  accessed  4  March  2013).  Statistics  
based  on  words  used  by  Shakespeare  are  easy  to  find  using  sources  such  as  
Open  Source  Shakespeare.org  
(http://www.opensourceshakespeare.org/stats/,  accessed  4  March  2013.  
Folger  Shakespeare  Library  director  Michael  Witmore  used  DocuScope’s  
(http://www.cmu.edu/hss/english/research/docuscope.html)  rhetorical  
analysis  data  mining  technique  to  analyze  vocabulary  and  syntax  in  
Shakespeare’s  First  Folio  corpus  (see  
http://www.fastcompany.com/1800987/data-­‐minings-­‐thing-­‐shakespeare-­‐
takes-­‐center-­‐stage-­‐digital-­‐age,  accessed  4  March  2013).  But  these  analyses  
do  little  to  identify  event  notifiers.  

A  wide  variety  of  sentiment  analysis  tools  attempt  to  analyze  text  to  
determine  its  ‘polarity’  e.g.  positive,  negative  or  neutral  or  to  find  indicators  
of  an  emotional  state  e.g.  ‘happy’  or  ‘sad’.    Apart  from  helping  to  define  the  
general  ‘demeanor’  of  a  character’s  role,  for  example  a  positive  or  negative  
character  or  influence,  or  a  happy  or  sad  scene,  if  the  sentiment  analysis  
reveals  a  pattern  of  negativity  or  positivity  after  a  specific  scene  then  this  
itself  may  function  as  an  event  indicator,  indicative  of  an  inciting  incident,  
turning  point  or  twist  or  as  indicators  of  a  conflict  ‘rollercoaster’  within  the  
script  text.    

In  linguistic  studies,  one  kind  of  event  indicator  is  a  ‘discourse  marker’  
usually  discussed  in  the  context  of  discourse  analysis  (see  Schiffrin,  Deborah.  

Stewart  McKie   -­‐  224  -­‐   As  of  15/08/2014  

Discourse  Markers.  Cambridge  University  press,  1987).      

Schilder’s  analysis  of  temporal  discourse  markers  such  as  after,  before  or  
while  (see  http://www.aclweb.org/anthology-­‐new/W/W98/W98-­‐0310.pdf,  
accessed  29  March  2013)  indicates  that  these  marker  words  alone  can  be  
used  as  event  indicators.  According  to  Schilder,  analysis  of  the  clauses  that  
follow  these  words  may  be  classified  to  indicate  a  state,  activity,  
accomplishment  or  achievement.  The  problem  with  applying  this  to  
screenplays  e.g.  Alien  is  that  what  follows  the  marker  word  in  a  script  is  
usually  not  explicitly  stated  in  an  explicit  syntactic  unit  e.g.  ‘After  the  alien  
popped  out  of  his  chest  we  all  ran  away’  so  any  ‘after  effect’  from  the  event  is  
only  implicit  in  the  dialog  or  action  text  that  follows.  Or  the  marker  word  
simply  is  not  there  to  find,  say  to  mark  the  event  signifying  the  transition,  for  
example,  from  pre  to  post  Alien-­‐loose-­‐on-­‐the-­‐ship  worlds.    

Another  kind  of  discourse  marker  may  be  the  lexical  cues  to  dialog  acts  
discussed  by  Jurafsky,  Shriberg,  Fox  and  Curl  (see  
http://acl.ldc.upenn.edu/W/W98/W98-­‐0319.pdf,  accessed  29  March  29  
2013).  Although  Alien  actually  has  a  character  who  largely  speaks  in  lexical  
cues  (Brett)  these  kind  of  cues  add  little  value  as  event  markers  in  a  
screenplay.  However  they  could  be  used  to  identify  character  alliances  
within  a  screenplay.  For  example,  Brett’s  repeated  use  of  the  word  ‘Right’  in  
connection  with  dialog  involving  Parker  acts  as  a  token  representing  
agreement  and  is  indicative  of  the  fact  that  there  is  a  (positive)  relationship  
between  these  two  characters  (the  below  deck  engineers)  in  the  script.  So  
isolating  words  like  ‘Right’  and  identifying  the  current  speaker  and  the  
previous/next  speaker  could  be  a  way  to  propose  types  of  relationships  
between  characters.    

In  this  specific  ‘turning  point’  case  the  event  marker  that  really  identifies  a  
change  has  taken  place  is  that  a  new  character  is  introduced  –  the  Alien.  And  
here,  the  release  of  the  Alien  onto  the  ship  has  profound  significance  for  the  
rest  of  the  script.  We  might  be  able  to  infer  this  from  a  character  timeline  
because  from  this  scene  onwards  a  new  pattern  is  introduced  into  the  script,  

Stewart  McKie   -­‐  225  -­‐   As  of  15/08/2014  

namely  that  the  characters  that  were  active  in  the  script  prior  to  this  new  
character  introduction  steadily  become  inactive  (as  they  are  killed  off).  So  
one  event  indicator  of  a  turning  point  could  be  the  introduction  or  loss  of  a  
character  (or  characters)  that  leads  to  a  change  in  event  patterns  going  
forward.    

A  similar  obvious  event  marker  might  be  a  significant  change  of  primary  
location.  Consider  the  famous  cut  to  the  Vietnam  paddy-­‐field  in  Michael  
Cimino’s  The  Deerhunter  (1978)  switching  us  from  the  comfortable  buddy-­‐
buddy  life  at  home  in  the  Pennsylvania  steeltown  to  a  landscape  of  death  
and  destruction  now  surveyed  by  Michael’s  recently  acquired  thousand-­‐yard  
stare.  Vietnam  has  changed  everything  and  the  focus  changes  to  action  
located  in  the    ‘special  world’  of  Vietnam  (or  resulting  from  or  influenced  by  
it)  rather  the  ‘ordinary  world’  of  the  supermarket  and  bars  inhabited  by  
those  left  behind.  Location  change  is  often  most  obvious  as  a  marker  in  the  
transition  of  the  hero  of  the  Hero’s  Journey  from  the  ordinary  world  to  a  
new  unfamiliar  world  (and  back)  and  the  turning  point  that  may  be  
embodied  in  the  ‘call  to  adventure’  or  ‘crossing  the  first  threshold’  may  be  
identifiable  simply  from  a  major  location  change  that  follows  it.      

Other  kinds  of  event  indicators  that  could  be  used  to  identify  a  turning  point  
or  conflict  and  complications  include  a  single  unusual  event  (within  the  
context  of  the  screenplay  as  a  whole)  or  a  pattern  of  repetitive  events  –  for  
example  sex  and/or  violence.  Alien  also  exemplifies  a  pattern  of  repetitive  
events  after  the  turning  point,  namely  the  one-­‐by-­‐one  killings  of  the  crew  by  
the  Alien,  that  reflect  complications  and  conflict  that  also  lead  to  the  crisis,  
i.e.  Ripley  is  the  only  one  left  to  fight  off  the  Alien  and  prevent  it  getting  back  
to  earth.    In  Michael  Winner’s  Death  Wish  (1974),  a  single  and  
uncharacteristic  (in  terms  of  the  script  as  a  whole)  sexual  event  –  the  rape    -­‐  
is  the  turning  point  that  then  sets  off  the  series  of  violent,  vigilante  events  
that  comprise  the  complications  and  conflict  in  the  rest  of  the  movie  (a  
pattern  that  Winner  had  successfully  used  before  in  Chato’s  Land  (1972)).      

Stewart  McKie   -­‐  226  -­‐   As  of  15/08/2014  

Summary  

Screenplay  design,  as  reflected  in  the  primary  debate  around  structural  
paradigms,  is  in  its  infancy.  The  design  patterns  of  screenwriting  are  not  yet  
well  defined  but  some  aspects  of  a  screenplay’s  design  can  already  be  
reverse-­‐  engineered  from  the  finished  movie  by  digital  audio/video  analysis  
tools.  Although  screenplays  could  be  constructed  both  ‘subtractively’  and    
‘additively’  it  is  unlikely  they  will  ever  be  created  in  a  similar  fashion  to  a  
physical  3D  printed  object.    In  fact  design  may  be  something  that  it’s  best  to  
infer  from  a  screenplay  rather  than  work  to  as  a  creative  task.  

Identifying  aspects  of  screenplay  design,  through  event  notifiers  in  a  text,  
could  be  a  way  to  identify  where  important  design  elements  such  as  turning  
points  occur  that  may  help  a  screenwriter  to  improve  their  use  of  these  
essential  screenwriting  elements.  

Again,  datafication  of  the  screenplay  content  is  helpful  or  necessary  in  order  
to  facilitate  the  extrapolation  of  structure  from  content  or  to  help  identify  
event  notifiers  within  the  text.      

Stewart  McKie   -­‐  227  -­‐   As  of  15/08/2014  

5.  The  Road  to  Scenepad  

For  many  decades  the  technology  of  screenwriting  comprised  a  single  tool:  
the  typewriter.  The  courier  twelve  font  of  modern  day  screenplays  is  a  
legacy  of  the  default  fixed  size  font  of  pre-­‐golfball  typewriters;  that  all  
changed  in  the  1980’s  with  the  advent  of  the  personal  computer  (PC).  In  the  
last  decade,  new  form  factors  –  the  mobile  phone  and  tablet  –  have  become  
pervasive  and  stimulated  the  delivery  of  a  number  of  new  screenwriting  
applications  adapted  to  suit  the  form  and  function  of  these  new  devices.      

The  era  of  pervasive  personal  and  business  computing  is  generally  regarded  
as  beginning  with  the  commercial  release  of  the  IBM  PC  (or  IBM  5150)  in  the  
USA  in  1981.  The  PC’s  Microsoft  Disk  Operating  System  (MS-­‐DOS)  and  later  
MS  Windows  were  the  operating  system  platforms  that  triggered  the  
development  of  a  rapidly  expanding  universe  of  personal  and  business  
software  applications.  And  it  was  over  the  next  decade  that  the  first  
screenwriting  applications  appeared.  

Write  Brothers  Inc.  claims  that  it  released  the  first  ‘screenplay  formatting’  
program,  called  Scriptor,  in  1982.  However,  one  of  the  first  applications  to  
gain  real  market  traction  was  Final  Draft  (FD)  launched  a  decade  later  in  
1992,  which  now  claims  to  be  ‘the  world’s  best  and  best-­‐selling  scriptwriting  
program.  The  industry  standard  for  films,  television  shows  and  stage  plays.’  
Whether  this  is  true  or  not,  FD  is  certainly  a  de-­‐facto  standard  in  that  most  
other  screenwriting  applications  can  either  import  and/or  export  files  in  
Final  Draft’s  .fdx  xml-­‐based  file  format.  

In  fact  there  is  no  need  to  buy  a  specialist  screenplay  formatting  program  at  
all  since  most  standard  word  processing  applications  can  be  used  to  write  
and  format  screenplays  using  the  document  template  formatting  
functionality  that  comes  as  standard  with  leading  applications  such  as  
Microsoft  (MS)  Word.  Indeed  the  BBC’s  own  Script  Smart  is  a  set  of  custom  
templates  that  ‘add-­‐in’  to  MS  Word  to  facilitate  script  writing.  These  

Stewart  McKie   -­‐  228  -­‐   As  of  15/08/2014  

templates  make  it  easy  to  write,  format  and  print/export  a  screenplay  
without  the  need  to  learn  new  software  –  but  they  do  little  else.  

Early  screenwriting  applications  are  characterized  by  a  focus  on  script  

formatting,  feature-­‐film  movie  writing  and  a  single-­‐user,  writer  centric  
paradigm.  Over  time  more  specialized  screenwriting  applications  
supplemented  this  generic  functionality.  For  example  Movie  Outline,  
StoryView  and  StoryO  that  focus  on  planning  screenplays  and  story  outlining  
or  Character  Writer  and  Persona  focused  on  developing  an  in-­‐depth  
character  profiles  or  a    ‘character  spine’  for  a  script’s  cast  of  characters.  

Some  applications,  like  Chris  Huntley’s  Dramatica  Pro  and  Dramatica  Story  
Expert  are  specifically  designed  to  support  a  single  story  development  
paradigm;  in  this  case  Dramatica,  a  substantive  methodology  for  writing  
screenplays  based  on  a  four-­‐act  structure.  So  far  no  similar  programs  have  
been  specifically  developed  purely  to  support  a  single  popular  story-­‐writing  
paradigm,  such  as  Christopher  Vogler’s  Hero’s  Journey  for  example.  In  fact  
few  screenwriting  programs  even  contain  specific  built-­‐in  support  for  what  
is  widely  considered  the  most  influential  screenplay  structural  paradigm  of  
all  -­‐  Syd  Field’s  three-­‐act  structure.    

As  the  screenwriting  software  market  became  more  competitive  from  the  

end  of  the  1990’s  onward,  generic  screenwriting  packages  offered  variants  
or  specialized  functionality  to  service  needs  beyond  purely  writing,  
formatting  and  outputting  the  movie  screenplay.  In  2001  Final  Draft  
launched  Final  Draft  AV,  claimed  to  be  ‘the  first  software  application  
specifically  designed  for  writers  of  documentaries,  reality  TV,  and  
commercials  written  in  the  two-­‐column  script  format.’  Mariner  Montage  
provided  supplementary  facilities  for  marketing  your  script  (see  McKie,  
2006a)  and  Tim  Sheehan’s  Sophocles  (now  defunct)  delivered  a  number  of  
useful  innovations  for  analyzing  and  visualizing  script  content  such  as  
charting  and  character  relationship  networks  (see  McKie,  2006b).  

Scripts  also  became  input  for  other  non-­‐writing  applications  designed  for  

Stewart  McKie   -­‐  229  -­‐   As  of  15/08/2014  

movie  storyboarding,  pre-­‐visualization  or  production  management.  For  
example,  scripts  can  be  directly  imported  into  pre-­‐viz  and  storyboarding  
software  such  as  FrameForge  and  StoryBoard  Quick/Artist  or  into  Gorilla,  
used  for  production  budgeting,  scheduling  and  storyboarding  or  easySCOTT  
for  digital  set  management.  The  screenwriting  application  ecosystem  grew  
to  acknowledge  the  fact  that  the  script  has  a  lifecycle  that  extends  beyond  
the  writing  or  pre-­‐production  phase  and  into  the  production  phase  of  movie-­‐
making.    In  2012,  John  August  released  Highland  a  script  conversion  or  file  
interchange  tool  that  can  import/export  3  key  formats:  Final  Draft  .fdx  files,  
Adobe  PDFs  and  the  Fountain  markdown  format  (also  released  by  August’s  
team).  

The  Internet,  and  now  the  widespread  use  of  mobile  platforms  such  as  the  
Apple  iPhone  and  iPad,  also  triggered  a  new  wave  of  innovation  in  the  
screenwriting  market.  In  September  2012  there  were  at  least  ten  
screenwriting  applications  available  on  the  Apple  iPhone/iPad  App  Store  for  
writing  or  viewing  screenplays.  And  new  ‘Open  Source’  and  online  
screenwriting  applications  such  as  Celtx,  Trelby  and  Logline  are  also  proving  
popular,  partly  because  a  ‘freemium’  model  ensures  that  they  can  be  
delivered  for  free  initially,  only  requiring  a  paid  subscription  if  users  require  
more  specialist  features.  

Celtx  styles  itself  as  a  media  pre-­‐production  application.  It  uses  a  project  
paradigm  to  manage  the  production  of  a  wide  variety  of  output  products  
including  film  screenplays,  theatre  scripts  and  novels.  Like  its  predecessor  
Sophocles,  Celtx  also  includes  production  functionality  to  support  
scheduling  and  other  production  related  reporting.  Scripts  can  be  stored  
online  and  projects  shared  with  other  users  to  foster  collaborative  working  
via  the  Internet.  Celtx  allows  various  digital  assets  to  be  attached  to  the  
screenplay  to  enrich  the  content.  

I  use  the  term  ‘screenwriting  2.0’  to  refer  to  application  functionality  and  
ways  of  working  that  reflect  some  different  approaches  to  screenwriting  
than  those  generally  available  in  contemporary  screenwriting  applications  

Stewart  McKie   -­‐  230  -­‐   As  of  15/08/2014  

or  used  in  practice.    In  the  design  of  my  software  deliverable  Scenepad,  I  
tried  to  incorporate  functionality  and  practices  that  I  considered  to  be  
missing  or  inadequately  supported  in  most  current  screenwriting  
applications  especially  screenplay  analytics  and  screenplay  as  social  
network.  

It  is  the  following  brief  survey  of  screenwriting  technology  that  informs  my  
experimental  online  screenwriting  application  prototypes,  developed  during  
my  research,  and  my  final  application  deliverable,  submitted  with  this  thesis,  
called  Scenepad.    

5.1  Screenplay  Analytics  Today  

Screenplay  analytics  is  the  most  obvious  functionality  missing  from  most  of  
today’s  screenwriting  applications,  so  this  section  reviews  some  analyses  
and  visualizations  already  provided  by  the  current  generation  of  story-­‐
planning  and  screenwriting  software.  Note  that  these  are  generally  not  
applications  that  store  their  screenplay  content  as  data  in  a  relational  
database  (as  Scenepad  does)  but  store  the  content  in  some  kind  of  file  
format  as  a  document.    

Format  

The  most  common  screenplay  visualization  is  the  presentation  of  a  correctly  
formatted  script  for  viewing  on-­‐screen.  This  is  not  an  analysis  of  the  script  
content  as  such,  but  is  a  visualization  of  the  content  according  to  prevailing  
industry  conventions.  Often  this  view  is  itself  based  on  the  writer  selecting  a  
specific  format  template  to  work  with  when  they  begin  their  new  script.  The  
script  may  be  presented  in  an  “uncluttered”  full-­‐screen  view  or  as  part  of  a  
workspace  where  it  is  surrounded  by  other  panels  that  display  content  from  
within  the  script  (e.g.  a  role  list)  or  content  linked  to  the  script  (e.g.  images).    

In  figure  5.1  below,  the  script  (on  the  right)  is  next  to  two  navigational  
panels  on  the  left.  The  upper-­‐left  panel  offers  a  menu  of  functional  options,  
and  the  lower-­‐left  panel  displays  a  list  of  scenes  in  the  script  that  provides  

Stewart  McKie   -­‐  231  -­‐   As  of  15/08/2014  

both  a  scenic  breakdown  of  the  script  content  and  a  navigation  tool  to  allow  
the  user  to  click  on  a  scene  heading  and  “jump-­‐to”  the  scene  selected  in  the  
body  of  the  script  on  the  right.  This  kind  of  multi  panel  presentation  is  
commonly  used  in  current  screenwriting  applications.  

Figure  5.1  -­‐  Working  with  a  Formatted  Script  in  Celtx  

Hierarchy  

Another  common  visualization  is  of  the  script  hierarchy  (act  and/or  
sequence  and/or  scene)  in  the  form  of  an  interactive,  collapse-­‐and-­‐expand  
browser  tree,  which  is  used  to  navigate  the  script  as  a  whole  and  to  jump-­‐to  
specific  script  content,  such  as  a  scene  (see  figure  5.2).  By  expanding  each  
“node”  or  branch  in  the  tree  by  double-­‐clicking  nodes  with  their  mouse,  the  
user  can  display  acts,  sequences  and  scenes.  Usually  when  a  scene  is  selected  
(clicked  with  the  mouse)  in  the  tree,  the  scene  content  is  displayed  in  a  

Stewart  McKie   -­‐  232  -­‐   As  of  15/08/2014  

linked  panel.  These  content  navigation  trees  can  also  be  used  to  browse  the  
script  content  by  entities  other  than  scene  headings,  such  as  character  
names  (the  expansion  listing  scenes  in  which  the  specified  character  
appears)  or  locations  (the  expansion  listing  scenes  using  that  location).    

Figure  5.2  -­‐  Navigating  a  Script  Hierarchy  in  Sophocles  

Index  Cards  

Most  leading  screenwriting  packages  also  provide  index  card  visualization  

(see  figure  5.3).  Screenwriters  have  traditionally  used  physical  index  cards  
(e.g.  in  3"  x  5"  format)  or  more  recently  post-­‐it  notes  as  a  means  to  
breakdown  and  visualize  the  screenplay  narrative.  Index  cards  are  not  just  a  
way  of  visualizing  a  finished  script  draft  in  outline  but  also  a  way  to  plan  
screenplays,  especially  their  structure  and  narrative  flow,  before  any  
substantive  scene  writing  begins.  

For  example,  a  card  or  post-­‐it  may  represent  a  story-­‐beat  within  a  scene  or  a  
scene  itself  and  may  be  color-­‐coded  to  reflect  the  fact  that  it  relates  to  a  
specific  character-­‐arc  or  plotline  for  example.  Traditionally,  cards  are  pinned  
to  a  corkboard  or  notes  stuck  on  a  wall/whiteboard  so  that  the  whole  
screenplay  can  be  seen  at  once  and  individual  cards/notes  easily  moved  
around  to  manually  re-­‐organize  the  flow  of  the  script  if  required.  

Stewart  McKie   -­‐  233  -­‐   As  of  15/08/2014  

Screenwriting  software  can  present  script  scenes  in  the  same  way  on  a  
virtual  corkboard  and  make  use  of  the  mouse  to  provide  interactive  drag-­‐
and-­‐drop  card  reorganization  of  the  scene  cards.  

Figure  5.3  -­‐  Managing  Index  Cards  in  SuperNotecard  

Charts  

Some  other  kinds  of  screenplay  visualizations  use  business-­‐style  charts  to  
visualize  script  information.  These  include  simple  pie  charts  showing  the  
proportion  of  DAY  vs.  NIGHT  and  INT  vs.  EXT  scenes  (see  figure  5.4)  or  bar  
charts  of  scenes  represented  by  number  of  words  or  the  relative  proportion  
of  action  vs.  dialog  used  in  a  scene.    

Stewart  McKie   -­‐  234  -­‐   As  of  15/08/2014  

  

Figure  5.4  -­‐  Script  Analysis  Charts  in  Sophocles  

On  the  basis  of  these  charts,  even  without  reading  the  script,  one  can  tell  that  
most  of  the  movie  Casablanca  takes  place  inside  and  at  night,  which  makes  
sense  when  you  know  that  “Rick’s  Place”  is  a  nightclub.  This  charting  may  
seem  very  simplistic,  but  even  these  visualizations  may  tell  one  something,  
or  help  one  to  frame  questions,  about  a  script,  for  example:  

– More  exterior  night  scenes  could  mean  more  production  difficulty  

and  expense  vs.  more  interior  and  day  scenes.  Will  this  be  an  
expensive  movie  to  make?  

– The  lack  of  night  scenes  in  say  a  horror  movie  might  indicate  that  the  
script  is  deviating  from  genre  conventions.  Will  this  movie  meet  
audience  expectations?  

Stewart  McKie   -­‐  235  -­‐   As  of  15/08/2014  

Social  Network  

Another  visualization  provided  by  Sophocles  is  a  form  of  social  network  
showing  character  relationships  and  the  size  of  each  character  node  
(represented  in  figure  5.5  as  a  circle)  indicating  the  relative  “size”  of  the  
speaking  part  of  an  individual  character.  Again,  without  reading  the  
Casablanca  script,  one  can  tell  that  the  key  parts  are  those  of  Rick,  Ilsa,  
Laszlo  and  Renault  and  that  many  relationship  lines  lead  to  Rick  –  so  he  is  
probably  the  protagonist  (or  antagonist).  

If  metadata  was  available  to  further  identify  the  protagonist/antagonist  and  

their  “allies”,  then  this  network  visualization  could  be  further  improved  by  
the  use  of  color.  Say  green  to  represent  the  protagonist  and  allies  and  red  to  
represent  the  antagonist  and  allies.    

In  this  case  the  script-­‐level  social  network  is  a  useful  overview  of  all  the  
characters  in  the  script  but  needs  to  be  complimented  by  character-­‐specific  
networks  generated  around  a  single  selected  character,  that  use  the  size  of  
the  “satellite”  character  circles  to  show  how  important  the  character  is  in  
relation  to  the  “centre  node”  character.  In  the  case  of  Casablanca  this  could  
be  a  useful  way  of  visualizing  Renault’s  apparently  dual  loyalties:  To  his  
friend  Rick  and  to  his  Nazi  “boss”  Strasser.  

Stewart  McKie   -­‐  236  -­‐   As  of  15/08/2014  

  

Figure  5.5  -­‐  The  Social  Network  of  Casablanca  in  Sophocles  

Work  by  Agarwal,  Balasubramanian,  Zheng  and  Dash  and  by  Gil,  Kuenzel  
and  Suen  to  parse  screenplays  in  order  to  extract  social  networks  from  
movie  scripts  provides  much  more  sophisticated  techniques  than  that  used  
by  my  ScriptFAQ  prototype  that  automatically  generates  a  social  network  for  
a  specific  character  in  a  script  by  using  the  vis.js  library  (see  Appendix  C).  

Structure  

Visualizations  of  structure  are  hard  to  find  in  the  current  generation  of  
screenwriting  software  but  Plotbuilder  (www.plotbuilder.com)  makes  
explicit  use  of  what  the  application  calls  an  ‘excitement  graph’  as  the  basis  
for  constructing  your  script  by  enabling  the  writer  to  place  scenes  onto  
specific  plot  points  based  on  a  graphic  representation  of  a  script  structure  
(see  figure  5.6)  that  clearly  reflects  Field’s  three  act  structure  ‘paradigm’.  

Stewart  McKie   -­‐  237  -­‐   As  of  15/08/2014  

  

Figure  5.6  –  Plotbuilder  Excitement  Graph  

Timelines  and  Throughlines  

Visualizations  found  in  specialist  story  planning  or  outlining  software  such  
as  StoryView  (now  known  as  Outline  4D),  typically  focus  on  the  structure  of  a  
screenplay  (e.g.  the  acts,  sequences  and  scenes)  against  a  movie  timeline  
(above  the  structure  in  figure  5.7)  and  optionally  include  one  or  more  
“tracklines”  (which  would  be  below  the  structure  in  figure  5.7)  used,  for  
example,  to  display  a  character  throughline  or  a  plot  throughline.  This  
representation  of  acts,  sequences  and  scenes  as  larger  or  smaller  content  
blocks  depending  on  their  relative  duration  in  the  movie  is  subject  to  a  US  
patent.  

Stewart  McKie   -­‐  238  -­‐   As  of  15/08/2014  

  

Figure  5.7  -­‐  The  StoryView  Structural  View  of  Pulp  Fiction  

In  fact  StoryView  allows  the  whole  script  to  be  printed  in  the  form  of  a  wall  
poster  (see  figure  2.26)  to  visualize  the  entire  structure,  complete  with  a  
timeline  and  selected/all  tracklines.  

Figure  5.8  -­‐  The  StoryView  Script  Wall  Poster  

This  combination  of  timeline,  structure  and  throughline  lends  itself  well  to  a  
particularly  versatile  implementation  of  screenplay  analysis  and  
visualization,  as  illustrated  in  figure  5.8.  Here  the  visualization  uses  a  
framework  that  consists  of  structure  (acts,  sequences,  scenes)  and  timeline  
“below  the  line”  and  selected  throughlines  “above  the  line”.  So  against  a  

Stewart  McKie   -­‐  239  -­‐   As  of  15/08/2014  

constant  below  the  line  visual,  users  could  select  a  specific  type  or  set  of  
throughlines  to  display  above  the  line  –  for  example  plotlines  or  character  
arcs.  So  this  one  visualization-­‐framing  paradigm  can  generate  a  variety  of  
useful  timeline  and  structure  based  throughline  visualizations  from  the  
script.  

Figure  5.9  -­‐  A  possible  timeline,  structure  &  throughline  visualization    

Visualizing  the  screenplay  structure  against  an  estimated  movie  timeline  

(e.g.  based  on  1  minute  per  page  of  script)  helps  the  writer  to  understand  if  
the  script  content  is  well  balanced  over  the  duration  of  the  movie  and  
whether  act  breaks  occur  at  industry-­‐acceptable  points  along  the  movie’s  
timeline  to  satisfy  script  reader  and/or  audience  expectations.  Visualizing  
the  structure  against  a  trackline  helps  the  writer  to  understand  if  character  
or  plot  throughlines  contain  significant  gaps  that  might  somehow  negatively  
impact  the  flow  of  the  narrative  or  development  of  the  character.  

Stewart  McKie   -­‐  240  -­‐   As  of  15/08/2014  

Storyboarding  

Visualizing  the  script  as  a  storyboard  was  discussed  above  as  a  form  of  pre-­‐
visualization  of  the  movie  that  might  result  from  the  script.  Usually  this  is  
done  in  a  specialist  storyboarding  package  but  this  function  is  provided  in  
the  Celtx  screenwriting  application  (see  figure  5.10).  Note  that  that  this  
storyboarding  capability  depends  on  the  writer  (or  other  authorized  script  
collaborator  such  as  a  storyboard  artist)  manually  adding  the  storyboard  
images  to  the  script;  the  storyboard  is  not  automatically  generated  in  some  
way  by  the  package  itself.  

Figure  5.10  -­‐  Script  Storyboard  in  Celtx  

In  2013,  Amazon  (see  http://studios.amazon.com/storyteller  -­‐  accessed  11  

June  2013)  provided  the  ability  for  screenplays  submitted  to  its  Amazon  
Studios  to  have  storyboards  generated  from  and  linked  to  the  script  content  
in  a  dual  panel  visualization  comprising  script  text  on  the  left  and  
storyboard  on  the  right  (see  figure  5.11).  

Stewart  McKie   -­‐  241  -­‐   As  of  15/08/2014  

As  Jason  Parham  explains  (see  
http://www.complex.com/tech/2013/06/amazon-­‐storyteller-­‐turns-­‐scripts-­‐
into-­‐storyboards  -­‐  accessed  09  July  2013):  

So  how  does  it  work?  Storyteller  first  scans  a  movie  script  uploaded  to  
Amazon  Studios,  after  which  it  identifies  every  scene,  location  and  
character,  and  then  “casts  them  from  a  library  of  thousands  of  
characters,  props  and  backgrounds."  But  if  you're  not  totally  satisfied  
with  Storyteller's  board,  then  you  can  manually  recast,  change  
locations  and  upload  your  own  images.  

What  is  perhaps  of  more  interest  here  is  that  every  script  uploaded  to  
Amazon  Storyteller  is  being  ‘datafied’  and  that  data  refined  by  user  
intervention  (i.e.  to  develop  the  storyboard  from  the  initial  auto-­‐generated  
version).  So  Amazon  is  beginning  the  process  of  creating  a  ‘big  data’  
screenplay  repository  from  which  they  may  be  able  to  derive  interesting  
insights  later  on.  

Figure  5.11  –  Amazon  Storyteller  Storyboard  

What’s  certain  is  that  new  generations  of  ‘screenwriting  2.0’  applications    -­‐  

Stewart  McKie   -­‐  242  -­‐   As  of  15/08/2014  

following  the  lead  of  past  applications  like  Sophocles  and  my  own  Scenepad  –  
are  also  likely  to  opt  to  datafy  their  screenplay  content  storage  to  provide  
more  screenplay  analytic  capabilities.    

5.2  Social  Screenwriting  Today  

Other  functionality  generally  omitted  or  poorly  supported  by  traditional  

single-­‐author,  format-­‐focused  screenwriting  technology  is  the  social  aspect  
of  screenwriting,  which  manifests  itself  primarily  in  collaborative  
capabilities.  For  example,  in  Final  Draft,  there  is  a  function  called  
‘Collabowriter’,  which  enables  you  to  initiate  an  online  session  to  view  and  
edit  a  script  and  chat  via  instant  messaging  while  you  do  this.  

CollaboWriter  allows  you  to  collaborate  on  and  discuss  a  script  with  
other  Final  Draft  users  anywhere  in  the  world  via  the  Internet.  One  
person  initiates  the  session  (the  Host).  The  Host  or  another  person  can  
control  the  script  (the  Controller)  while  others  view  changes  as  they  
are  made.  CollaboWriter  also  contains  a  chat  window  so  ideas  and  
critiques  can  be  shared  instantly.  

(Final  Draft  User  Guide  Version  8:  314)  

This  ‘session-­‐based’  sharing  was  an  early  attempt  by  FD  to  provide  some  
kind  of  collaborative  script  sharing  that  leveraged  the  Internet.  But  FD  
simply  doesn’t  have  a  concept  of  a  ‘shared’  script  that  others  can  contribute  
to  driven  by  role-­‐based  permissions.  

Celtx  has  a  concept  of  sharing  a  script  ‘project’.  You  can  link  users  to  your  
project  and  give  them  permissions  to  access  your  project  e.g.  view  only  or  
change  rights.  Projects  can  also  be  ‘checked  out’  –  so  that  no  changes  can  be  
made  by  other  users  while  the  project  is  in  a  checked  out  state.  This  is,  in  
effect,  applying  basic  document  management  principles  to  a  screenplay  
script.  The  issue  with  this  approach  is  one  of  granularity  –  it  is  the  project  as  
a  whole  that  is  being  shared  rather  than  specific  entities  within  it  or  ways  of  
working  with  it  (see  Borst,  pp.223-­‐225).  But  for  many  scriptwriters  this  may  
be    ‘good  enough’  sharing.  

Stewart  McKie   -­‐  243  -­‐   As  of  15/08/2014  

Most  screenwriting  tools  do  not  recognize  the  need  either  for  a  script  idea  
entity  or  a  script  work  group  to  ideate  it  (cf.  MacDonald,  2004).  Script  
treatments,  synopses  and  pitches  may  all  be  regarded  as  potential  
deliverables  from  the  ideation  stage  of  the  screenplay  development  lifecycle  
(see  figure  5.12)  in  that  an  actual  script  need  not  have  been  written  in  order  
to  deliver  any  of  them.  To  support  ideation,  screenwriting  software  needs  
the  concept  of  a  script  idea,  an  ideation  workgroup  and  a  way  of  enriching  
ideas  to  support  the  ideation  process.  

And  because  many  screenwriting  tools  are  not  multi  user  there  is  no  support  
for  the  functional  roles  that  are  needed  once  you  open  up  a  screenplay  to  
collaboration  within  a  social  network.  If  screenplays  are  defined  as  
collaborative  artefacts  then  a  minimum  of  2  roles  are  needed:  View  and  
Writer/Partner.  The  former  only  allows  a  user  to  view  script  content  and  
add  value  to  it  (in  the  form  of  linked  content)  but  not  to  mess  with  the  core  
screenplay  content  (i.e.  add/change/delete  it).  The  latter  may  allow  a  user  to  
add/change/delete  specific  or  all  script  content.  Obviously  the  additional  
availability  of  add/change/delete  rights  for  the  Writer/Partner  role  could  
permit  more  granular  access  to  the  screenplay  content.      

5.3  Screenwriting  2.0  Prototypes  

The  practical  element  of  my  research  involved  the  creation  of  publicly  
accessible,  online  applications  as  prototypes  for  new  screenwriting  tools.  I  
began  my  PhD  as  a  non-­‐programmer  so  initially  a  third  party  coded  these  
sites,  as  acknowledged  elsewhere.  However  in  all  cases  the  applications  
were  designed,  data-­‐modeled  in  the  MySQL  database,  tested  and  deployed  
by  me.  By  the  end  of  my  PhD,  I  became  an  ‘amateur’  PHP  programmer  myself  
so  my  final  prototype  –  Scenepad  -­‐  contains  a  significant  amount  of  my  own  
coding  albeit  with  some  ‘hard’  coding  (e.g.  the  script  import  function)  done  
by  someone  else.    

I  briefly  discuss  and  give  examples  of  each  prototype  below.  In  each  case  I  
define  the  concept  and  aim  of  the  prototype,  the  technology  used  and  my  

Stewart  McKie   -­‐  244  -­‐   As  of  15/08/2014  

‘takeaways’  from  each  prototype  used  to  inform  my  final  deliverable,  
Scenepad.    

In  this  section  I  also  refer  to  two  ‘benchmark’  commercial  screenwriting  

applications:  Final  Draft  and  Celtx.  These  have  been  chosen  for  specific  
reasons.  Final  Draft  is  the  de-­‐facto  market  leading  screenwriting  application,  
and  (as  of  2013)  remains  a  traditional  software  package  in  that  it  must  be  
downloaded  and  installed  on  your  local  PC/MAC  and  costs  significantly  
more  than  the  free  or  downloadable  ‘apps’  so  prevalent  today.  Celtx  is  the  
most  sophisticated  of  the  newer  open  source  screenwriting  applications,  
offers  a  ‘freemium’  sales  model  –  free  to  start,  pay  for  premium  features  -­‐  
and  runs  both  offline  and  online.  My  references  to  these  applications  will  be  
to  specific  versions  namely  Final  Draft  8  and  Celtx  2.9.1.  

The  practical  deliverables  discussed  below  are:  

Year   Application   About  

2006   Scriptcloud   An  online  application  for  generating  a  frequent  

words  text  cloud  from  the  text  of  an  uploaded  text  
file  of  a  screenplay.  

2007   Scriptgeist   An  online  application  for  generating  various  text  

clouds,  charts  and  lists  from  the  text  of  an  
uploaded  text  file  of  a  screenplay.  

2008   Contentcloud   An  upgrade  of  Scriptcloud  that  allows  comparison  

of  two  text  clouds  from  different  texts.  

2010   Sceneclass   A  Wordpress  plugin  for  managing  a  group-­‐based  

script/scene  writing  class  within  a  Wordpress  

Stewart  McKie   -­‐  245  -­‐   As  of  15/08/2014  

 blog.  

2011   Scenewrite   An  online  screenwriting  application  for  use  in  the  

pre-­‐production  and  production  phases  of  a  script’s  
lifecycle.  

2012   Scenepad   An  online  screenwriting  application  for  use  in  the  

pre-­‐production  and  production  phases  of  a  script’s  
lifecycle.  

2013   Scenepad3   A  variant  of  Scenepad  with  a  ‘responsive’  UI  that  

should  work  more  effectively  on  tablet  devices.  

2014   ScriptFAQ   A  script  datafication  program  focused  on  

delivering  screenplay  analytics  based  on  a  
‘frequently  asked  questions’  paradigm.  

Table  5.1  –  List  of  Prototypes  

5.4  Scriptcloud  [2006]  

My  first  prototype  was  Scriptcloud.  Scriptcloud  launched  in  March  2006  on  
scriptcloud.com.  Hannah  King  coded  the  application  in  PHP  using  my  design  
and  data  model,  managed  in  the  MySQL  database.  As  figure  5.12  shows,  at  
retirement  in  July  2008  the  site  had  276  active  users  who  had  uploaded  343  
scripts  as  the  source  for  text  clouds.  

The  idea  behind  scriptcloud  was  to  allow  screenwriters  to  create  a  word  
frequency  cloud  from  the  text  of  their  screenplay,  in  a  similar  way  to  
creating  a  ‘tag’  cloud  from  the  tags  linked  to  blog  posts.  The  aim  was  to  
understand  if  this  simple  analysis  and  visualization  had  any  analytical  value  
in  terms  of  helping  the  writer  to  understand  their  screenplay  better.    

Stewart  McKie   -­‐  246  -­‐   As  of  15/08/2014  

The  screenplay  is  not  datafied,  instead  an  uploaded  text  file  is  parsed  and  
the  scriptcloud  generated  from  the  words  parsed  from  the  text  file.  The  
database  was  used  to  store  metadata  about  the  script  (e.g.  title,  genre)  and  
the  resulting  cloud  image  (fore  easier  retrieval  later)  not  to  store  the  words  
in  the  script.    

The  site  is  currently  located  at:  http://scriptcloud.tripos.biz  

Scriptcloud  requires  users  to  register  to  upload  a  text  file  of  an  English-­‐
language  screenplay  into  their  personal  script  ‘library’.  On  upload,  a  word  
cloud  is  generated  that  is  stored  in  the  database  and  can  be  downloaded  to  a  
PDF  or  simply  copied  to  paste  somewhere  else.  The  user  can  then  access  
their  clouds  from  their  library,  on-­‐demand.  I  added  hundreds  of  popular  
movie  scripts  for  reference  purposes  and  made  these  accessible  to  all  
registered  users.  

Stewart  McKie   -­‐  247  -­‐   As  of  15/08/2014  

Figure  5.12  –  Scriptcloud  Statistics  July  6,  2008  

Keyword  Text  Clouds  

Scriptcloud  leveraged  a  visualization  paradigm  that  had  recently  become  

popular  on  the  Internet:  The  tag  cloud.  A  tag  cloud  is  used  to  visualize  the  
textual  metadata  ‘tags’  added  to  a  piece  of  content  such  as  a  web  page  or  a  
blog-­‐posting  for  example.  The  purpose  of  the  cloud  is  to  give  an  indication  of  
the  popularity  of  a  specific  tag  in  terms  of  the  body  of  content  the  tag  has  
been  used  to  refer  to  and  to  help  categorize  content  for  searching  and  
navigation  purposes.  Tag  clouds  are  often  used  to  highlight  changing  
patterns  of  “buzz”  among  large  and  dynamic  data  sets,  such  as  active  news  
blogs.  Typically,  tagged  data  sets  could  comprise  daily  blog  postings,  new  
web  pages  or  images  or  videos  added  to  a  site  such  as  Flickr  or  YouTube.  

Figure  5.13  –  Scriptcloud.com  Home  Page  

Conventionally,  a  cloud  displays  more  popular  tags  in  a  bigger  font  and/or  in  
a  specific  colour.  The  cloud  itself  is  usually  limited  to  displaying  a  top-­‐n  
number  of  terms  (e.g.  the  top  50  tags  from  a  specific  content  set)  and  tags  
representing  a  single  language.  Tag  clouds  can  also  be  used  as  navigational  
Stewart  McKie   -­‐  248  -­‐   As  of  15/08/2014  
devices.  If  a  tag  word  is  “live”  (i.e.  represents  a  link  to  a  web  URL)  the  user  
can  navigate  to  a  web  site  or  page  or  blog  post  or  specific  instance  of  a  term  
within  the  content  set,  simply  by  clicking  the  tag.  

Although  a  script  as  a  whole  could  be  tagged  with  specific  metadata,  such  as  
the  genre  of  the  script  for  example,  and  a  tag  cloud  produced  from  a  set  of  
scripts  tagged  in  this  way  my  intention  with  Scriptcloud  was  to  show  how  
cloud  visualization  can  be  used  in  a  different  way  –  potentially  to  surface  
themes  on  the  script.  

A  ‘Scriptcloud’  uses  the  textual  content  of  a  script,  rather  than  any  user-­‐
added  metadata  tags,  to  produce  what  is  more  accurately  termed  a  
“keyword  cloud”  of  the  script  content  itself.  The  default  scriptcloud  
generated  from  a  user-­‐uploaded  script  (see  figure  5.14)  shows  the  64  most  
popular  terms  used  in  the  script  (after  an  exclusion  or  stop  list  is  applied  to  
filter  out  common  “noise”  words  e.g.  “a”,  “the”,  “and”  etc.).    

The  user  also  has  the  capability  to  produce  clouds  from  their  script  that  
exclude  additional  user-­‐specified  terms  (e.g.  profanities)  or  include  only  
specific  terms  e.g.  character  names  in  the  script.  What  a  scriptcloud  
represents  is  a  very  limited,  but  potentially  interesting,  linguistic  
‘fingerprint’  of  a  screenplay.  Users  have  reported  that  it  can  help  them  to  
identify  or  recognize  keywords  or  themes  in  their  script.  Their  attention  has  
also  been  drawn  to  overused  nouns  or  verbs  in  the  script,  prompting  them  
to  be  more  inventive  in  their  choice  of  vocabulary.    

Stewart  McKie   -­‐  249  -­‐   As  of  15/08/2014  

  

Figure  5.14  -­‐  Scriptcloud  of  My  Own  Carpini  Script  

Although  the  scriptcloud  produced  from  individual  scripts  is  interesting,  

what  is  more  interesting  is  what  clouds  presented  from  the  entire  corpus  of  
scripts  could  illuminate.  For  example,  the  range  of  keywords  presented  in  a  
scriptcloud  already  excludes  a  large  list  of  noise  words.  But  it  turns  out  that  
most  scriptclouds  show  that  the  most  popular  keywords  in  a  script  are  banal  
action  terms  such  as  “sits”  or  “walks”  or  colloquial  dialog  terms  such  as  
“gonna”  and  profanities  such  as  “shit”.  These  terms  are  seldom  keywords  
and  often  tell  us  little  or  nothing  about  the  thematic  content  of  the  script.  

Producing  a  scriptcloud  from  a  corpus  of  scripts  is  likely  to  produce  a  cloud  
full  of  relatively  banal  terms  that  can  then  be  iteratively  added  to  the  general  
exclusion  list  so  that  subsequent  scriptclouds  of  the  same  corpus  are  likely  
to  be  that  much  more  representative  of  genuine  keywords  in  the  corpus.  
Once  the  corpus  includes  a  large  enough  sample  of  scripts  within  a  specific  
genre  then  a  genre-­‐specific  scriptcloud  can  be  produced  that  also  facilitates  
the  creation  and  use  of  a  genre-­‐specific  exclusion  list.  Then  when  a  script  is  
uploaded  and  linked  to  specific  genre  metadata,  this  genre-­‐specific  exclusion  
list  could  be  used  as  the  default  stoplist  rather  than  any  generic  exclusion  
list.  

As  noted  above,  each  script  uploaded  to  the  site  is  also  tagged  with  a  genre  
classification.  If  a  significant  enough  content  set  were  established  on  a  site  
like  Scriptcloud,  new  kinds  of  scriptclouds  can  be  produced  that  indicate  the  
most  popular  genres  and  produce  a  fingerprint  of  the  most  popular  terms  
across  all  scripts  tagged  to  a  specific  genre.  This  could  be  useful  commercial  

Stewart  McKie   -­‐  250  -­‐   As  of  15/08/2014  

information  for  a  screenwriter:  On  the  one  hand  to  indicate  current  popular  
genres  and  on  the  other  to  enable  the  writer  to  compare  their  individual  
script’s  scriptcloud  to  the  aggregated  scriptcloud  produced  from  many  
scripts  sharing  the  same  genre  as  their  own  script.  

In  practice,  Scriptcloud  generated  minimal  useful  feedback  from  the  

hundreds  of  users  who  uploaded  scripts  to  the  site.  Scriptcloud,  like  most  
sites  that  leverage  user-­‐generated  content  (e.g.  Facebook  and  Twitter),  
depends  on  user  engagement  to  succeed  and  deliver  real  value.  The  more  
content  that  is  uploaded,  i.e.  scripts  in  this  case,  the  more  that  can  be  done  to  
leverage  that  content.  The  majority  of  users  uploaded  a  script,  got  their  
scriptcloud  and  never  came  back.    

In  any  case,  the  usefulness  of  a  Scriptcloud  as  a  script  analysis  tool  is  limited  
since  it  depends  largely  on  the  composition  of  the  exclusion  or  stoplist.  Too  
many  words  in  the  stoplist  and  keywords  that  might  be  indicative  of  theme  
could  be  lost,  having  been  excluded  from  the  cloud.  To  few  words  in  the  
stoplist  and  keywords  might  be  crowded  out  by  noise  words.  So  some  
algorithm  would  be  needed  to  optimize  the  stop  list  to  reduce  the  impact  of  
either  scenario  on  the  cloud  generation.    

Also  it  is  likely  to  be  coincidental  that  a  scriptcloud  does  in  fact  provide  
some  thematic  clues  to  a  script’s  content.  For  example,  the  scriptcloud  
generated  from  a  version  of  the  scripts  of  Coppola’s  The  Godfather  (1972)  
surfaces  the  word  ‘family’,  which  many  would  consider  an  important  
thematic  element  of  the  script.  However,  it  would  be  hard  to  tell  from  the  
surfacing  of  this  keyword  that  The  Godfather  is  not  about  the  humdrum  life  
of  some  regular  ‘mom  and  pop’  family  but  the  violent  saga  of  a  Mafia  clan.  

Stewart  McKie   -­‐  251  -­‐   As  of  15/08/2014  

  

Figure  5.15  -­‐  The  Godfather  Scriptcloud  

Refining  the  stoplist  to  exclude  more  of  the  most  frequent  words  from  the  
initial  scriptcloud  is  a  way  to  further  isolate  ‘suspected’  keywords.  In  the  
case  of  The  Godfather,  it  also  generates  a  useful  result  that  is  almost  text-­‐
book  ‘apple-­‐pie’  in  its  Americaness:  America,  family  and  love.  Yet  arguably  
the  ‘genco’,  ‘sicily’  and  ‘war’  terms  give  some  clue  that  this  script  is  not  likely  
to  be  about  some  Norman  Rockwell  family.    

Figure  5.16  -­‐  The  Godfather  Keywords  

But  what  if  a  theme  is  not  easily  identifiable  from  an  actual  word  in  the  
script?  For  example,  one  might  propose  that  one  theme  of  Friedkin’s  The  
French  Connection  (1971)  is  ‘persistence’  -­‐  as  in  Popeye  Doyle’s  persistence  
in  getting  his  man,  aka  ‘the  frog’.  However  the  term  ‘persistence’  does  not  
appear  in  the  scriptcloud  and  neither  does  anything  that  clearly  relates  to  
this  term.  What  does  surface  though  are  a  number  of  active  verbs  that  
suggest  this  script  might  be  some  kind  of  action  movie.  

Stewart  McKie   -­‐  252  -­‐   As  of  15/08/2014  

  

Figure  5.17  -­‐  The  French  Connection  Scriptcloud  

When  you’ve  seen  the  movie,  then  the  surfacing  of  words  like  ‘heroin’  and  
‘junk’,  ‘chase’  and  ‘pursuit’  and  ‘subway’  for  example,  are  significant  but  not  
especially  illuminating  if  you  haven’t  seen  the  movie.  We  get  no  clue  that  the  
movie  is  set  in  New  York  city  and  is  a  battle  between  a  couple  of  dogged  
NYPD  vice  squad  cops  and  a  French  drug  lord.    

Scriptclouds  are  a  simple,  interesting  visualization  of  the  screenplay  content  

that  might  just  surface,  through  the  words  used,  something  interesting  that  
confirms  or  denies  a  screenwriters  intention.  It  would  be  easy  enough  to  
include  a  scriptcloud  as  part  of  any  screenplay  package  that  is  provided  to  
readers  since  it  does  provide  some  clues  as  to  what  the  script  might  be  
about.  On  that  basis  there  seemed  no  reason  why  they  should  not  be  
included  in  Scenepad.    

5.5  Contentcloud  [2008]  

To  further  develop  the  potential  of  scriptclouds,  the  next  version  of  
Scriptcloud  was  Contentcloud,  a  site  that  focused  on  another  viable  use  for  
scriptclouds:  for  text  comparison  purposes.  For  example  to  compare  one  
version  of  a  script  with  another    -­‐  either  drafts  of  the  same  script  or  sequels  
–  or  one  ‘act’  of  a  script  with  another  within  the  same  script.  To  facilitate  this  
new  usage,  an  update  to  Scriptcloud  was  produced  to  allow  for  side-­‐by-­‐side  
comparison  of  2  sets  of  content.  The  scope  of  Scriptcloud  was  also  expanded  
to  allow  for  the  upload  of  any  type  of  text  file,  not  just  a  screenplay.  For  
example,  this  enhancement  enabled  the  text  of  a  screenplay  to  be  compared  
to  that  of  the  novel  it  was  adapted  from.  

Stewart  McKie   -­‐  253  -­‐   As  of  15/08/2014  

This  site  was  called  Contentclouds.com  and  was  again  coded  in  PHP  and  
MySQL  by  Hannah  King  based  on  my  design  and  data  model.  

The  site  is  currently  located  at:  http://contentcloud.tripos.biz  

Figure  5.18  -­‐  Contentclouds  Home  Page  

In  the  comparison  of  the  screenplays  of  The  Godfather  (1972)  and  The  
Godfather  II  (1974)  in  figure  5.19,  it  is  clear  that  the  ‘theme’  of  family  still  
runs  through  the  text.  The  Contentcloud  comparison  also  includes  words  
shared  and  not  shared  between  the  texts.  At  first  glance  these  clouds  may  
appear  to  be  of  the  same  script,  with  the  consistent  repetition  of  many  of  the  
same  nouns  and  verbs.  But  there  are  clues  that  these  scripts  are  different  
such  as  references  to  the  Lake  Tahoe  ‘boathouse’,  Las  Vegas  ‘gambling’  and  
the  Cuban  hospital  ‘nurse’  from  The  Godfather  II.  Given  enough  drafts  of  a  
single  script  it  might  even  be  possible,  by  comparing  scriptclouds,  to  trace  a  
real  change  of  theme  between  early  and  later  drafts.  This  in  turn  could  show  
how  the  writer’s  perspective  changed  as  she  wrote  or  how  other  pressures  –  
genre-­‐conformance,  budgetary  etc.  –  necessitated  changes  in  the  script.    

Stewart  McKie   -­‐  254  -­‐   As  of  15/08/2014  

  

Figure  5.19  -­‐  Comparing  Godfather  Scripts  

It’s  also  interesting  to  trace  scriptcloud  keywords  through  the  ‘act’  structure  
of  a  screenplay  or  theatre  script.  In  the  comparison  of  the  scriptclouds  of  
Shakespeare’s  Macbeth  Acts  1  and  2  below,  we  can  detect  some  kind  of  
‘turning  point’  that  has  changed  the  atmosphere  of  the  play  from  Act  1’s  
‘hail’,  ‘honour’,’love’  to  Act  2’s  ‘bloody’,  ’daggers’  and  ‘horror’.  There  is  also  
consistency  of  language  in  the  ‘thee’,  ‘thou’  and  ‘thy’  -­‐  words  that  would  have  
been  in  an  Elizabethan  stoplist  but  were  not  in  mine  as  they  do  not  reflect  
today’s  noise  words.  Further  indication  of  the  sensitivity  required  in  

Stewart  McKie   -­‐  255  -­‐   As  of  15/08/2014  

creating  an  appropriate  stop  list  for  the  texts  to  be  processed.    

Figure  5.20  -­‐  Comparing  Acts  in  Shakespeare’s  Macbeth  

Contentclouds  do  a  useful  job  of  basic    ‘delta’  analysis  between  two  texts  to  
highlight  similarities  and  differences  based  on  the  word  content  only.  A  
potential  development  of  Contentcloud  would  be  to  create  a  scrolling  
‘timeline’  view  of  drafts  where  only  most  frequent    ‘delta’  words  between  
each  draft  are  highlighted  as  this  might  help  to  show  how  new  themes  
emerge  or  theme  consistency  is  maintained  through  the  drafts.  

5.6  ScriptGeist  [2007]  

Thanks  to  funding  from  LCACE,  Scriptgeist.com  was  created  to  deliver  a  
better  platform  for  both  more,  and  more  sophisticated  analyses  and  
visualizations  of  screenplay  content.    The  aim  was  to  create  my  first  
screenplay  datafication  prototype  as,  unlike  both  Scriptcloud  and  
Contentcloud,  Scriptgeist  accepts  a  text  file  upload,  parses  the  content  and  
applies  rules  to  ‘shred’  it  into  a  set  of  relational  entity  data  stored  in  tables  in  
a  MySQL  database.  The  main  database  entities  created  from  the  script  
content  are  scenes,  characters  and  locations.    Scriptgeist  is  the  direct  

Stewart  McKie   -­‐  256  -­‐   As  of  15/08/2014  

antecedent  of  Scenepad  and  much  of  what  I  learned  from  this  prototype  is  
replicated,  more  or  less,  in  Scenepad.  

David  North  coded  the  application  in  Ruby  on  Rails  based  on  my  design  and  
data  model.  The  site  is  no  longer  accessible.  

Figure  5.21  -­‐  Scriptgeist  Home  Page  

Despite  offering  significantly  more  functionality  than  Scriptcloud,  the  site  

statistics  for  Scriptgeist  in  figure  5.22  show  that  it  only  attracted  some  89  
users  who  uploaded  71  scripts.  Scriptgeist  also  contained  a  user  engagement  
tool  called  Geistmeister  to  let  users  review  random  dialog  retrieved  from  a  
script  to  guess  the  script  it  came  from.  Correct  guesses  got  points  and  the  
top  5  leaderboard  is  displayed  on  the  home  page  (bottom  right).  Ironically,  
the  Geistmeister  challenge  proved  more  popular  than  the  main  function  of  
the  site  itself.  Perhaps  I  had  an  ‘Angry  Birds’  on  my  hands  and  didn’t  realize  
it.  

Scriptgeist  built  on  Scriptcloud  by  decomposing  the  script  content  into  
‘snippets’  of  action  or  dialog  and  then  generating  a  series  of  simple  analyses  
and  visualizations,  in  addition  to  more  scriptclouds,  from  this  content  that  

Stewart  McKie   -­‐  257  -­‐   As  of  15/08/2014  

provide  a  useful,  if  not  insightful,  foundation  for  further  script  content  
analyses.  

Figure  5.22  -­‐  Scriptgeist  Statistics  

Scriptgeist  enables  a  user  to  generate  three  keyword  clouds  based  on  all  the  
script  text  (as  in  Scriptcloud),  or  clouds  based  on  dialog  or  action  text  only.  

Stewart  McKie   -­‐  258  -­‐   As  of  15/08/2014  

Figure  5.23  -­‐  The  Godfather  Dialog  Cloud  

Scriptgeist  generates  a  character  throughline  chart  to  show  which  scenes  the  
top  10  characters  (by  dialog  quantity)  appear  in  throughout  the  script.  

Figure  5.24  -­‐  The  Godfather  Character  Throughline  

Scriptgeist  generates  pie  charts  to  show  the  scenes  in  a  script  categorized  by  
INT/EXT  and  DAY/NIGHT.  

Stewart  McKie   -­‐  259  -­‐   As  of  15/08/2014  

Figure  5.25  -­‐  The  Godfather  INT/EXT  Scenes  

A  list  of  scenes,  characters  and  locations  are  generated  from  the  uploaded  
script  text.  The  character  list  shows  the  number  of  words  and  dialog  
snippets  linked  to  a  character  to  help  analyze  comparative  part-­‐size.  

Figure  5.26  -­‐  The  Godfather  Characters  and  Part  Size  

Clicking  on  the  dialog  snippets  number  lists  all  dialog  text  spoken  by  a  
character  to  help  analyze  voice  consistency.  

Stewart  McKie   -­‐  260  -­‐   As  of  15/08/2014  

  

Figure  5.27  -­‐  The  Godfather  Character  Dialog  

Scriptgeist  delivered  a  wide  range  of  screenplay  analytics  functionality  and  

only  lacked  the  ability  to  write  and  manage  a  script  and  to  share  a  script  for  
social  networking  purposes.  At  the  time,  to  my  knowledge,  it  was  the  only  
application  of  its  kind  that  was  publicly  available.  

5.7  Sceneclass  [2010]  

Given  the  explosion  of  the  blogging  paradigm  and  pervasive  use  of  social  
networks  by  2010,  it  seemed  important  to  explore  how  to  socialize  the  scene  
writing  process.  I  decided  to  do  this  by  leveraging  the  popular,  open  source  
blogging  platform  Wordpress  as  the  basis  for  my  next  prototype  called  
Sceneclass.com.  The  aim  of  sceneclass  was  to  focus  on  scenewriting  as  a  
group  activity  where  the  scene  content  functioned  like  a  blog  post.  

The  application  was  coded  by  Nick  Ohrn  and  deployed  as  a  Wordpress  
plugin.  Apart  from  John  August’s  Scrippets,  a  text  formatting  plugin  for  
generating  screenplay-­‐formatted  text  from  plain  text  input,  this  was  the  first  
group  screenwriting  plugin  ever  produced  for  Wordpress.  The  site  is  
currently  located  at:  http://sceneclass.tripos.biz.    

Stewart  McKie   -­‐  261  -­‐   As  of  15/08/2014  

Note  that  some  of  the  screenshots  below  obscure  the  names  of  the  
participants  for  privacy  reasons.  

Figure  5.28  -­‐  Sceneclass  Home  Page  

Sceneclass  was  tested  in  July  2010  by  a  group  of  12  students  at  a  
scenewriting  class  tutored  by  Adam  Ganz  at  Royal  Holloway  as  part  of  a  
Masters  level  course  in  screenwriting  run  by  Sue  Clayton.  The  task  was  to  
write  a  ‘cold  opening’  for  the  hit  US  comedy  series  30  Rock.  In  a  period  of  
about  3  hours  the  students  not  only  got  up  to  speed  on  Sceneclass,  but  also  
wrote  a  scene  and  contributed  to  the  88  comments  posted  to  the  scenes  via  
Sceneclass.  Feedback  from  students  on  the  sceneclass  paradigm  was  positive  
and  they  had  no  difficulty  using  the  prototype  following  a  brief  ‘getting  
started’  presentation  on  the  day.  

The  Sceneclass  paradigm  starts  with  a  Wordpress  user  group  called  a  ‘Class’  
with  two  user  roles  -­‐  a  ‘Tutor’  and  a  ‘Student’  –  each  with  a  set  of  rights.  For  
example,  only  a  tutor  can  create  a  class.    

Stewart  McKie   -­‐  262  -­‐   As  of  15/08/2014  

  

Figure  5.29  –  Sceneclass  Class  Entity  

The  tutor  or  the  students  create  a  ‘Script’  as  a  container  for  ‘written  by  the  
students.  Students  could  create  their  own  script  and  scenes  to  share  with  the  
group  or  the  tutor  could  create  a  single  ‘group  script’  into  which  students  in  
the  group  add  their  own  scenes.    

Figure  5.30    -­‐  Sceneclass  Scene  Entities    

Scenes  are  written  just  like  a  regular  Wordpress  blog  post  [explain]  but  with  
certain  conventions  e.g.  character  names  are  in  CAPS  so  that  the  scene  text  
can  be  formatted  to  screenplay  conventions.  Once  a  scene  is  ‘posted’  to  the  
script  it  can  be  viewed  correctly  formatted.  The  scenes  are  flagged  as  
‘private’  or  ‘shared’  to  share  with  the  rest  of  the  group  so  that  each  scene  
may  be  commented  on  by  the  rest  of  the  class  in  much  the  same  way  as  any  
regular  blog  post  in  Wordpress  (or  any  other  blog).  

Stewart  McKie   -­‐  263  -­‐   As  of  15/08/2014  

  

Figure  5.31  –  Sceneclass  Scene  Comments  

Properly  formatted  scenes  can  be  viewed  by  anyone  in  the  class  group  and  
commented  on  to  provide  a  social  scenewriting  experience,  as  shown  below.    

The  blogging  paradigm  is  quite  suitable  for  scene  writing  in  that  each  blog  
post  functions  as  a  scene  and  a  collection  of  posts  represent  a  script.  With  a  
little  background  work  in  CSS,  scenes  can  be  viewed  correctly  formatted  and  
commented  on,  shared  and  rated/’liked’  just  like  a  regular  blog  post.  
Rewriting  is  easy,  you  just  edit  your  post,  and  scenes  can  be  tagged  to  aid  
with  searching  and  the  wealth  of  Wordpress  plugins  includes  word  cloud  
generators  that  could  create  a  scriptcloud  from  a  series  ‘scene  posts’  either  
by  a  single  writer  or  a  class  of  students.  

Stewart  McKie   -­‐  264  -­‐   As  of  15/08/2014  

  

Figure  5.32  –  Sceneclass  Student  Responses    

Sceneclass  showed  that  students  quickly  ‘got’  how  to  use  the  scenewriting  
and  commenting  paradigm  and  in  the  post-­‐session  feedback,  they  said  they  
got  value  from  it.  A  handful  of  other  Universities  and  colleges  expressed  an  
interest  in  using  Sceneclass  and  some  tried  to  use  it  but  it  failed  to  gain  any  
real  traction  with  these  prospective  users.  Despite  this  poor  take-­‐up,  
commenting  is  an  integral  part  of  Scenepad  and  is  informed  by  my  
experience  of  developing  and  using  Sceneclass.  

Stewart  McKie   -­‐  265  -­‐   As  of  15/08/2014  

Summary  

A  number  of  online  deliverables  were  made  publicly  available  to  prototype  
new  or  different  ways  of  managing  and  socializing,  analyzing  and  visualizing  
screenplay  content.    All  of  these  prototypes  were  used,  more  or  less,  by  a  
range  of  users  including  practitioners,  academics,  school,  undergraduate  and  
postgraduate  students.  Although  all  the  prototypes  included  ways  to  contact  
me  or  provide  feedback  and  in  some  cases  formal  feedback  was  requested,  
actual  feedback  was  minimal  and  so  their  research  value  limited.      

These  days,  the  availability  of  over  a  dozen  free  online  screenwriting  
applications  competing  for  user  mindshare  means  that  many  users  (myself  
included)  tend  to  sign  up,  try  the  app  and  then  never  return.  This  was  the  
case  with  my  prototypes  –  for  example  in  Scriptcloud  most  users  who  signed  
up  then  uploaded  a  script.  But  once  they  had  generated  a  scriptcloud  from  
their  script  they  never  came  back.  Scriptgeist  users  came  back  mainly  to  play  
the  Geistmeister  game  and  prospective  Sceneclass  users  could  not  put  
enough  time  and  effort  into  getting  Sceneclass  running  in  their  institutions  
despite  good  intentions.  

So  on  the  basis  of  the  user  response  to  these  prototypes  and  the  lack  of  
substantive  feedback,  it  is  difficult  to  draw  any  valid  conclusions  about  the  
usefulness  or  not  of  the  functionality  provided.    

Stewart  McKie   -­‐  266  -­‐   As  of  15/08/2014  

6  Scenepad  

Scenepad  is  my  final  ‘assessment’  deliverable.  It  is  informed  by  the  
prototypes  that  preceded  it,  as  discussed  above.    Scenepad  is  my  attempt  to  
distil  some  of  the  best  of  these  prototypes  into  a  single  application  that  can  
be  used  to  create  and  manage,  present  and  share,  analyze  and  visualize,  
screenplay  content  in  new  or  different  ways.  It  is  my  practical  contribution  
to  the  screenwriting  2.0  debate  and  I  hope  that  much  of  what  Scenepad  
delivers  will  become  standard  functionality  in  the  next  generation  of  
commercial  screenwriting  applications.  

Scenepad  was  coded  (at  various  times)  in  PHP  &  Javascript  by  myself,  Scott  
Darby,  Matt  Kenyon  and  Hannah  King  based  on  my  design  specifications  and  
my  data  model  in  MySQL.  Scenepad  is  owned  by  Tripos  Publishing  Ltd.,  the  
company  that  funded  the  occasional  costs  of  the  coding  effort.  

The  first  iteration  of  Scenepad  (‘scenepad1’)  went  into  a  limited  public  ‘beta’  
phase  in  January  2012  at  this  location:  http://scenepad.tripos.biz  

Scenepad2  iteration  can  be  found  at:  http://scenepad2.tripos.biz  

Scenepad3  iteration  can  be  found  at:  http://scenepad3.tripos.biz  

Sceneapad4,  my  final  submission  iteration,  can  be  found  at:  

http://scenepad4.tripos.biz  

A  full  list  of  contributors  to  Scenepad4  and  components  used  in  the  
application  can  be  found  on  the  site’s  Credits  page  at:  
http://scenepad4.tripos.biz/credits.php.  

The  screenshots  provided  in  this  section  reflect  all  4  iterations  of  Scenepad,  
each  of  which  had  a  slightly  different  user  interface.  The  second  to  fourth  
iterations  of  Scenepad  use  the  popular  Twitter  Bootstrap  framework  for  the  
user  interface  so  look  slightly  different  from  the  initial  version.  The  third  and  
fourth  iterations  use  examples  of  a  ‘responsive’  user  interface,  better  for  use  
Stewart  McKie   -­‐  267  -­‐   As  of  15/08/2014  
on  an  iPad  as  it  resizes  automatically.  These  four  iterations  are  referred  to  as  
Scenepad1,  2,  3  and  4  below.  

The  Scenepad1  beta  was  tried  by  a  small  group  ‘beta  test’  users  of  which  
only  a  couple  of  user  actually  engaged  usefully  with  the  prototype  by  
providing  a  few  comments  in  the  application  and  some  email  feedback.  I  
wish  to  thank  Carmen  Sofia  Brenes  of  the  Universidad  de  los  Andes  for  her  
help  in  this  beta  test.  As  a  work  in  progress  subject  to  multiple  user  interface  
and  functionality  changes,  it  was  perhaps  unrealistic  to  expect  busy  
academics  to  engage  with  testing  a  screenwriting  application  like  Scenepad.    

Figure  6.1  -­‐  Original  Scenepad1  Beta  Home  Page  

6.1  Scenepad  –  Screenplay  as  Data  

All  Scenepad  content  is  stored  as  data  and  metadata  in  tables  in  a  MySQL  
relational  database  management  system  (RDBMS).  So,  for  example,  the  PDF  
files  that  can  be  downloaded  to  print  out  script  or  scene  content  are  
generated  on-­‐the-­‐fly  by  querying  data  and  metadata  in  the  database  tables  
and  do  not  pre-­‐exist  as  files  stored  in  the  file  system.  Any  screenplay  textual  

Stewart  McKie   -­‐  268  -­‐   As  of  15/08/2014  

data  (e.g.  dialog  or  action  snippet  text)  is  not  stored  as  a  text  file  but  as  data  
in  a  column  in  a  table.  

Scenepad  clearly  has  a  ‘database’  rather  than  ‘document’  design  and  delivery  
paradigm.  This  will  not  suit  many  screenwriters,  maybe  most  screenwriters.  
But  this  approach  has  enabled  me  to  produce  an  application  that  is  fully  
featured  and  runs  entirely  online  for  relatively  low  cost.  I  chose  this  
approach  as  the  basis  for  Scenepad  largely  because  I  understand  SQL  
databases  better  than  the  XML  file  format,  which  would  have  been  the  
obvious  alternative  choice  as  a  way  to  datafy  the  screenplay  content.    

In  MySQL,  the  Scenepad  design  is  anchored  by  three  main  tables:  script,  
scene  and  snippet  and  these  main  entities  are  supported  by  role  and  location  
tables.  A  ‘snippet’  is  a  block  of  action  description,  dialog,  a  caption  or  a  
transition.  Snippets  belong  to  scenes  and  scenes  belong  to  scripts.  A  script  
must  contain  one  or  more  scenes  and  a  scene  must  contain  one  or  more  
snippets.  From  a  production  perspective,  a  snippet  could  be  linked  to  one  or  
more  shots.  The  current  version  of  Scenepad  does  not  support  shots  as  an  
entity  concept  or  linking  shots  to  snippets/scenes  as  indicated  by  the  dashed  
arrow  &  box,  but  this  would  be  relatively  easy  to  add  (although  low-­‐cost  
applications  like  ShotList  (see  http://solubleapps.com/shotlist/)  already  do  
a  very  good  job  of  this  on  iPhone  devices).  

Scenes  take  place  at  locations  and  scene  dialog  snippets  are  spoken  by  a  role.  
The  actual  content  of  a  scene  is  stored  in  two  tables:  the  snippet  table  
(representing  the  current  ‘datafied’  version  of  the  scene  content)  and  the  
version  table,  which  stores  the  text  of  current  and  all  previous  versions  that  
were  saved  to  the  database.    

Stewart  McKie   -­‐  269  -­‐   As  of  15/08/2014  

  

Figure  6.2  –  Core  Scenepad  Data  Entities  (Tables)  

The  datafication  of  the  script  in  this  way  means  that  any  kind  of  presentation  
of  the  script  content,  from  an  individual  snippet  to  the  whole  script,  is  in  fact  
based  on  a  database  query.  Scenepad  does  not  retrieve  a  file  containing  the  
script  content  but  assembles  the  script  content  on  the  fly  using  a  query  
depending  on  what  you  want  to  see  and  how  you  want  to  see  it.  This  is  a  
different  way  of  interacting  with  a  screenplay  than  the  traditional  
document-­‐focused  presentation.  Scenepad  provides  a  PDF  version  of  the  
script  at  multiple  levels:  Script,  Role,  and  Location.  The  PDFs  include  all  the  
scenes  that  are  linked  to  the  whole  script,  role  or  location.  Sequence  and  
storyline  level  PDFs  are  also  available.  

Screenplays  stored  as  document  files  are  usually  very  easy  to  search  to  find  
specific  instances  of  a  term.  Scenepad  also  includes  3  search  pages  to  search  

Stewart  McKie   -­‐  270  -­‐   As  of  15/08/2014  

all  snippet  text  or  only  dialog  snippets  or  action  snippets.  

6.2  Scenepad  -­‐  Scenewriting  

Traditional  document-­‐centric  screenwriting  software  is  focused  on  writing  

scripts.  Scenepad  is  focused  on  writing  scenes.  In  Scenepad,  the  script  acts  
merely  as  a  container  for  scenes.    The  primary  entity  that  the  user  interacts  
with  to  actually  write  a  Scenepad  script,  is  the  scene.  This  means  that  
individual  scenes  can  be  shared,  rather  having  to  share  a  whole  script  and  an  
individual  scene  can  be  versioned  and  compared,  which  facilitates  
collaborative  working  and  could  help  with  the  activity  of  ‘polishing’  a  script  
that  often  focuses  on  specific  scenes  or  script  dialog  content  only.  

Given  Syd  Field’s  assertion  that  ‘The  Scene  is  the  single  most  important  
element  in  your  screenplay’  (2005:162),  screenwriter  David  Mamet’s  
definition  of  a  film  as  essentially  a  progression  of  scenes  (2007:85)  and  
Wells  (2011:101)  perspective  of  scenes  as  a  micro-­‐narrative  in  their  own  
right  (in  animation  scripts),  surprisingly  few  screenwriting  pundits  and  
academics  have  focused  on  scenewriting  as  a  discipline.  Any  significant  
reference  to  the  art  of  scenewriting  is  also  missing  from  works  focused  on  
screenplay  ‘coverage’  such  as  Sher’s  Reading  Screenplays  and  Garfinkel’s  
Screenplay  Story  Analysis.      

McKee  considers  scenes  to  be  comprised  of  a  series  of  ‘beats’  and  scenes  
themselves  to  form  the  beats  of  sequences:  ‘Ideally  every  scene  is  a  STORY  
EVENT’  (1998:35-­‐38).  Scofield  defines  a  scene  as  having  four  basic  
elements:  event  and  emotion,  function,  structure  and  a  pulse.  (2007:14).  
Aronson  emphasizes  the  importance  of  the  scene  that  includes  the  first  
turning  point  as  the  one  striking  and  unpredictable  scene  (2010:101)  and  
what  the  audience  needs  to  know  about  characters  in  a  scene  (ibid:93).  
Batty  and  Waldeback  include  a  useful  section  on  scenes  as  an  aspect  of  
structure  and  narrative,  emphasizing  the  need  for  scenes  to  have  their  own  
arc,  the  importance  of  topping  and  tailing  a  scene,  and  paying  attention  to  
scene  transitions  in  relation  to  their  preceding  and  succeeding  scenes,  

Stewart  McKie   -­‐  271  -­‐   As  of  15/08/2014  

commenting  (2008:39-­‐47):  

Much  scene  work  tends  to  be  intuitive  but  it  is  very  useful,  not  only  for  
writers  but  for  all  development  personnel,  to  analyze  and  approach  
them  with  a  committed  level  of  craft.    

Sternberg  (1997:65-­‐66)  analyzes  scenes  by  dividing  scene  content  into  

dialogue  text  and  scene  text,  which  roughly  correspond  to  Scenepad’s  dialog  
and  action  snippet  types  except  that  Sternberg  considers  that  certain  other  
‘instructions’  about  either  a  snippet  or  a  scene  e.g.  character  cues,  transition  
instructions  and  scene  headings  to  be  part  of  the  scene  text  rather  than  
metadata  that  refers  either  to  scene  or  snippet  text.  

Scene  Versioning  

Most  scriptwriting  applications  also  lack  the  ability  to  version  scenes  and  
compare  versions.  One  exception  is  Mariner’s  Storymill  application  that  does  
include  the  ability  to  compare  a  current  to  a  prior  version  (or  ‘snapshot’  as  
they  call  it)  and  to  replace  the  current  version  with  a  prior  version  or  delete  
versions.    

There  are  a  number  of  reasons  why  scene  versioning  can  help  scriptwriters,  
especially  with  the  rewriting  process:  

1.  To  track  the  development  of  a  scene  over  time.    

2.  To  easily  return  to  a  previous  version  of  a  scene  to  replace  the  current  
version.  
3.  To  compare  different  versions  of  the  same  scene  written  by  different  
people  e.g.  within  a  writing  team  or  a  screenwriting  class.  
4.  To  ‘cherrypick’  the  best  text  from  alternative  versions  and  aggregate  it  
into  a  current  version.      

Stewart  McKie   -­‐  272  -­‐   As  of  15/08/2014  

  

Figure  6.3  -­‐  Mariner  Storymill  Scene  Snapshots  

Scenepad1  allows  comparison  of  the  current  version  against  any  number  of  
previous  versions  and  highlights  the  differences  between  the  current  and  
‘reference’  version  –  red  for  deleted  words  and  green  for  inserted  words.  

Stewart  McKie   -­‐  273  -­‐   As  of  15/08/2014  

Figure  6.4  -­‐  Scenepad1  Scene  Versioning  

Scene  Grouping    

Scenepad3  &  4  include  the  capability  to  group  scenes  into  dramatic  units  or  
by  specific  types  of  scene.  Scenes  can  be  linked  to  either  a  storyline  or  a  
sequence  or  flagged  as  having  a  specific  content  e.g.  SFX,  VFX,  Sexual,  
Violence,  Crowds  or  stunts.  Flagging  scenes  in  this  way  enables  storyline  and  
sequence  thrulines  to  be  visualized  and  a  set  of  dials  used  to  quickly  
understand  the  how  profane,  violent  or  stunt  heavy,  for  example,  a  script  is  
written  to  be  (see  figure  6.5).    Scene  status  can  also  be  indicated  with  a  
‘traffic  light’  to  show  writing  status  e.g.  in-­‐play  (amber)  or  final  (green)  in  
the  scene  list  (see  figure  6.6).  

Stewart  McKie   -­‐  274  -­‐   As  of  15/08/2014  

Figure  6.5  –  Scenepad3  –  Scene  Content  Types  and  Analysis  

Figure  6.6  –  Scenepad4  Managing  Scene  Grouping  

Scene  Structure  Template  

Scenepad3  &  4  also  enable  writers  to  quickly  apply  the  basics  of  Vogler’s  
Hero’s  Journey  as  a  template  for  their  script  by  optionally  selecting  to  use  
the  template  when  the  initial  script  entity  is  created.  This  ‘starter’  template  
automatically  creates  the  12  core  sequences  and  8  character  archetypes  for  
the  writer  to  build  on.    

Stewart  McKie   -­‐  275  -­‐   As  of  15/08/2014  

  

Figure  6.7  -­‐  Scenepad3  Hero’s  Journey  Template  

Scene  Sequencing  and  Storylining  

Scene  sequencing,  grouping  scenes  into  dramatic  units  called  ‘sequences’,  is  
missing  from  both  Final  Draft  and  Celtx.  While  both  include  scene  navigators  
to  jump  around  a  script  and  focus  on  a  specific  scene,  with  Celtx  providing  a  
nice  drag-­‐and-­‐drop  feature  for  reordering  scenes  on  the  fly,  neither  formally  
recognize  either  the  3-­‐Act  structure  or  grouping  scenes  either  into  
sequences  or  plot/storylines.  Scenepad  3&4  provides  a  drag  and  drop  scene  
reordering  while  Scenepad1  provided  a  script  navigator  to  navigate  through  
scenes  by  means  of  Acts  and  Sequences  using  a  conventional  collapse  and  
expand  tree  (see  figure  6.8  below).  

Stewart  McKie   -­‐  276  -­‐   As  of  15/08/2014  

  

Figure  6.8  –  Scenepad1  Act  and  Sequence  Navigator  

Scenepad4  lets  writers  tag  scenes  as  ‘belonging’  to  a  sequence  or  a  storyline  
(see  figure  6.9).  This  enables  the  script  to  be  managed  by  sequence  or  
storyline  so  that,  for  example,  it  would  be  possible  for  a  specific  user  or  user  
role  to  manage  one  or  more  sequences  or  storylines  to  effect  a  division  of  
labour  within  a  writing  team.  Scenepad4  also  lets  you  view  scenes  or  print  a  
PDF  of  the  scenes  by  sequence  or  storyline  so  the  view/PDF  only  includes  
scenes  tagged  to  that  specific  sequence  or  storyline  to  help  review  and  
manage  these  structural  entities  more  easily.    

Figure  6.9  –  Scenepad4  Scene  Storyline  Manager  

Scene  Storyboard  

Scenepad4  includes  a  simple  storyboard  view  for  attaching  images  to  a  

Stewart  McKie   -­‐  277  -­‐   As  of  15/08/2014  

scene.    These  images  could  be  traditional  hand-­‐drawn  boards  or  images  
found  on  the  Internet  –  as  is  the  case  in  figure  6.10.  Given  that  Scenepad  is  
multi-­‐user  and  role-­‐based  it  would  also  be  possible  to  create  a  
‘storyboarder’  role  that  only  has  access  to  scene  text  and  the  storyboarding  
functions  so  this  task  in  the  script  development  process  can  be  easily  
‘farmed  out’  to  an  external  expert.    

Figure  6.10  –  Scenepad4  Scene  Storyboard  

6.3  Scenepad  as  Social  Network  

Scenepad  recognizes  that  it  may  be  advantageous  to  the  writer  or  simply  
necessary  to  share  a  script  with  others.  Scenepad  enables  the  script  owner  
to  decide  the  sharing  status  of  her  script.  The  script  status  ‘private’  means  
script  content  is  only  accessible  to  the  script  ‘owner’  user.    The  script  status  
‘shared’  means  script  content  is  may  be  accessible  to  another  user  or  other  
users  that  are  part  of  a  group  that  the  script  owner  creates  and  invites  users  
to  join.    Scenepad  groups  (see  figure  6.11)  could  reflect  a  screenwriting  class  
at  a  college,  a  writing  team  or  the  members  of  a  production  set  for  example.  

Stewart  McKie   -­‐  278  -­‐   As  of  15/08/2014  

The  script  status  ‘public’  means  the  script  content  can  be  viewed  (only)  by  
any  Scenepad  user.  This  is  useful,  for  example,  to  provide  ‘reference’  
screenplays  to  a  scriptwriting  class.  
 

Figure  6.11  –  Scenepad4  Groups  

Group  members  have  a  role  within  the  group  and  may  have  the  sharing  
rights  of  ‘Partner’  or  ‘Viewer’  –  the  former  enabling  more  content  addition  
and  editing  rights  than  the  latter  (see  figure  6.12).  

Stewart  McKie   -­‐  279  -­‐   As  of  15/08/2014  

Figure  6.12  -­‐  Scenepad4  Group  Users  

Individual  scripts  can  be  shared  with  a  specific  group  so  all  users  within  the  
group  can  engage  with  the  script.  

Screenplay  as  Content  Hub  

If  you  accept  that  a  script  has  a  creative  lifecycle  then  you  should  also  accept  
that  it  has  the  potential  to  have  a  ‘scriptstream’  –  a  variant  of  the  ‘lifestream’  
concept  conceived  by  Eric  T  Freeman  at  Yale  University  in  1997:    

A  lifestream  is  a  time-­‐ordered  stream  of  documents  that  functions  as  a  

diary  of  your  electronic  life;  every  document  you  create  and  every  
document  other  people  send  you  is  stored  in  your  lifestream.  The  tail  
of  your  stream  contains  documents  from  the  past  (starting  with  your  
electronic  birth  certificate).  Moving  away  from  the  tail  and  toward  the  
present,  your  stream  contains  more  recent  documents  -­‐-­‐-­‐  papers  in  
progress  or  new  electronic  mail;  other  documents  (pictures,  
correspondence,  bills,  movies,  voice  mail,  software)  are  stored  in  
between.  Moving  beyond  the  present  and  into  the  future,  the  stream  
contains  documents  you  will  need:  reminders,  calendar  items,  and  to-­‐
do  lists.  

Replace  ‘life’  with  ‘script’  and  ‘your’  with  ‘its’  and  you  have  the  concept  of  a  
‘scriptstream’  that  in  turn  helps  to  reframe  screenwriting  as  a  process,  and  
the  screenplay  as  a  content  hub.      

This  scriptstream  content  may  take  the  form  of  images,  location  videos,  
voice  clips  and  continuity  snaps.  All  of  which  serve  the  dual  purposes  of  both  
helping  to  realize  the  script  in  a  professional  manner  and  providing  a  record  
of  the  realization  process  itself  for  future  diagnosis  and  analysis  say  in  the  
editing  process  or  to  assist  with  filming  reshoots  or  pick  up  shots.  

The  scriptstream  provides  rich  content  that  annotates  the  script  itself  and  
can  significantly  enhance  its  meaning  and  immersion  of  workgroup  roles  in  
the  ‘world’  of  the  script.  There  are  at  least  three  ways  this  content  can  be  
added  to  the  script.    

1.  Writers  could  sit  at  a  desk  and  attach  content  found  on  the  Internet.    

Stewart  McKie   -­‐  280  -­‐   As  of  15/08/2014  

2.  Cast  and  crew  ‘on  set’  could  capture  content  as  it  is  produced.    
3.  Cast  and  crew  ‘on  location’  or  simply  ‘out  in  the  world’  can  capture  
content  e.g.  photos  that  they  think  will  enhance  the  realization  process.    

In  the  case  of  [2]  and  [3]  above,  this  depends  on  the  use  of  mobile  devices  as  
content  capture  devices  and  then  on  the  ability,  somehow,  to  link  content  
captured  ‘on  the  go’  to  the  relevant  aspects  of  the  script  –  for  example  to  link  
a  video  of  a  building  to  a  location  in  the  script  or  a  recording  of  a  birdsong  in  
a  forest  glade  to  the  soundtrack  of  a  scene.  Scenepad4  provides  a  content  
hub  at  script  level  (see  figure  6.13)  that  enables  users  to  view  almost  all  data  
associated  with  a  script  in  one  place  and  to  link  social  content  in  the  form  of  
comments,  documents,  images  and  web  links.  

Stewart  McKie   -­‐  281  -­‐   As  of  15/08/2014  

Figure  6.13    -­‐  Scenepad4  Script  Manager  Content  Hub  

Scenepad4  also  provides  a  content  hub  concept  at  scene,  role  and  location  
entity  levels.  We  have  already  seen  above  how  a  scene  can  have  a  storyboard  
attached  to  it.  Scenepad  also  assumes  that  both  roles  and  locations  may  also  
have  their  own  gallery  of  images  attached  to  them  to  act  as  further  
visualizations  of  the  ‘world’  of  the  script  (e.g.  role  gallery  in  figure  6.14).  
These  galleries  do  not  function  as  a  storyboard  in  that  the  images  are  not  
intended  to  be  sequenced  (i.e.  as  individual  ‘shots’  are  in  a  scene  storyboard)  
instead  they  function  as  a  place  to  gather  visual  research,  say  about  the  
character,  and  to  visually  remind  the  writer  of  the  character  they  are  writing  
for.  A  location  or  role  gallery  can  also  be  useful  to  propose  the  kind  of  
location  or  character  that  is  in  the  writer’s  mind.      
 

Figure  6.14    -­‐  Scenepad4  Role  Gallery  

Script  Mashups  

Once  the  screenplay  becomes  a  content  hub  then  the  need  for  ‘mashups’  
Stewart  McKie   -­‐  282  -­‐   As  of  15/08/2014  
becomes  more  likely.  Here,  the  term  mashup  is  used  to  reflect  the  ability  to  
use  application  programing  interfaces  (APIs)  to  link  a  screenwriting  
application  to  other  online  applications  in  order  to  combine  content  or  
provide  specialist  functionality  provided  by  the  third  party  application.    

Scenepad1  included  a  mashup  with  Dropbox  -­‐  a  popular  online  file  storage  
service  -­‐  that  facilitates  its  ability  to  support  the  Screenplay  as  content  hub  
paradigm.  Scenepad  allows  script,  scene,  role  and  location  entities  to  have  
any  number  of  attachments  attached  to  them.  The  simple  way  to  do  this  is  
simply  to  link  to  the  content  via  a  web  link.  But  by  linking  Scenepad  to  
Dropbox,  users  can  retain  control  over  their  own  attachment  content,  by  
physically  storing  it  in  their  Dropbox  account,  and  Scenepad  can  leave  the  
storage  and  management  of  this  content  to  a  service  that  is  optimized  for  
this  purpose.  Neither  Final  Draft  nor  Celtx  have  any  support  for  mashups  of  
this  kind.  Scenepad2  included  a  mashup  with  blogging  platform  Tumblr  to  
link  content  posted  to  Tumblr  with  a  Scenepad  script.  Scenepad3/4  enables  
users  to  view  photos  posted  by  email  to  kee.ps  and  to  view  Tweets  posted  to  
a  specific  Twitter  user  account  (which  could  be  the  name  of  a  script  for  
example).    

In  the  script  realization  stage  (i.e.  during  production),  writers,  cast  and  crew  
adopt  the  generic  role  of  ‘content-­‐providers’  to  the  script,  each  with  a  
specific  focus  in  terms  of  the  content  they  provide  and  the  script  entities  that  
they  link  it  to.    Now  the  script  can  become  the  ‘hub’  of  a  universe  of  content  
that  helps  to  realize  and  reflect  the  world  of  the  script  realization  process.  
And  surrounding  the  script  with  this  content  may  begin  before  any  actual  
filming  has  taken  place  as  part  of  the  pre-­‐production  activity.  Then,  as  a  
result  of  this  additional  linked  content,  a  script  is  realized  in  two  ways,  both  
as  an  intermediary  ‘rich-­‐content’  artefact  and  as  some  form  of  end  product  –  
the  audio/visual  experience  itself.      

Scene  Commenting  

Scenepad4,  like  the  Sceneclass  prototype  before  it,  recognizes  that  scenes  

Stewart  McKie   -­‐  283  -­‐   As  of  15/08/2014  

are  the  most  likely  entity  to  benefit  from  social  interaction  through  
commenting  (cf.  Sceneclass  above).  So  the  scene  navigator  includes  the  entry  
and  viewing  of  4  types  of  comment:  A  general  comment,  a  highlight,  a  
question  and  an  idea  or  suggestion  as  shown  in  figure  6.15  below.  The  scene  
text  is  displayed  in  the  left  hand  panel  and  the  commenting  panel  on  the  
right.  

Figure  6.15  –  Scenepad4  Scene  Commenting    

This  means  that  if  a  script/scene  is  shared  with  a  group  of  users,  say  a  
screenwriting  class,  multiple  users  can  easily  comment  on  the  same  scene  at  
the  same  time  as  part  of  a  scene  feedback  and  development  assignment  for  
example.  

Screenplay  Mobile  Access  and  Delivery  

Today’s  social  networkers  are  used  to  interacting  with  their  social  network  
from  mobile  devices,  especially  their  smartphones.  They  tweet  from  their  
phone  they  submit  pictures  taken  with  their  phone  camera,  they  rate  or  post  

Stewart  McKie   -­‐  284  -­‐   As  of  15/08/2014  

comments  on  content  they  view  on  their  phone,  from  their  phone.  

Scenepad4  recognizes  this  need  for  mobile  access  to  screenplay  data  and  the  
potential  utility  of  contributing  data  to  a  screenplay  from  a  mobile  device.  
This  begins  with  generating  a  Quick  Response  code  (QR  code)  for  each  script  
to  enable  content  to  be  accessed  from  mobile  devices  (see  figure  6.16).  The  
script  owner  can  then  copy  the  QR  code  from  Scenepad  and  paste  it  
wherever  it  might  be  utilized  from  e.g.  a  blog  or  Facebook  page.    A  
smartphone  user  with  a  QR  reader  app  installed  can  simply  snap  the  code  to  
access  the  screenplay  content  via  the  Scenepad  mobile  site.  

Figure  6.16  -­‐  Scenepad4  Script  QR  code  

The  Scenepad  mobile  site  provides  view  only  access  to  script  content  
(scenes,  roles,  locations)  and  to  a  range  of  Scenepad  analytics  including  
charts  and  word  clouds  (see  figure  6.17).    

Stewart  McKie   -­‐  285  -­‐   As  of  15/08/2014  

  

Figure  6.17  -­‐  Scenepad4  Mobile  site  home  page  

The  Scenepad  mobile  app  allows  users  to  ‘pull’  content  and  analytics  via  
their  mobile  device  but  does  not  yet  allow  users  to  push  content  directly  into  
Scenepad  via  their  mobiles  devices,  for  example  enabling  content  rating  
from  the  mobile  app  or  pushing  photos  taken  on  the  phone  camera  into  a  
Scenepad  storyboard.  This  is  perfectly  practical  but  would  require  a  more  
sophisticated  app  on  the  device  and  ideally  the  addition  of  an  application  
programming  interface  (API)  to  Scenepad.    

One  workaround  for  posting  content  to  Scenepad  is  to  allow  Scenepad  users  
to  post  content  to  other  online  applications  that  do  have  a  more  
sophisticated  smartphone  app  and  an  API  and  then  ‘pull’  the  data  from  these  
apps  into  Scenepad.    Please  note  that  as  this  functionality  depends  on  3rd  
party  APIs  that  frequently  change,  the  Scenepad  functions  may  not  continue  
to  work  as  expected,  or  at  all.  There  are  three  examples  of  API-­‐based  
interaction  with  3rd  party  sites  in  my  Scenepad  prototypes.  

In  Scenepad2  users  could  post  any  kind  of  data  to  a  Tumblr  blog,  link  it  

Stewart  McKie   -­‐  286  -­‐   As  of  15/08/2014  

somehow  to  specific  Scenepad  ‘categories’  (e.g.  a  role  name)  so  that  
Scenepad  could  connect  to  their  blog  via  the  Tumblr  API  and  display  all  the  
data  for  a  specific  category  assembled  onto  a  Scenepad  Tumblr  ‘wall’.  Figure  
6.18  shows  some  images  posted  to  the  Scenepad  Tumblr  blog  of  potential  
actors  for  the  casting  of  the  role  of  Maggie  in  one  of  my  own  scripts.  The  
images  were  posted  to  Tumblr  and  categorized  as  <scriptname>-­‐<role>-­‐
<rolename>  so  that  they  can  be  retrieved  from  Tumblr  and  displayed  in  the  
correct  Scenepad  wall  (in  this  case  the  wall  for  the  role  ‘Maggie’).  

Figure  6.18  -­‐  Scenepad2    -­‐  showing  Tumblr-­‐sourced  images  displayed  on  Scenepad  role  wall  

In  Scenepad  3  &  4  users  can  display  photos  posted  to  a  kee.ps  account  via  
Scenepad.  Kee.ps  in  an  online  service  like  Flickr  and  others  that  lets  you  take  
a  photo  on  your  smartphone  and  post  it  to  your  online  photo  portfolio  
where  they  can  be  stored  in  different  directories.  So  if  you  create  a  directory  
(e.g.  ‘Alien’  on  kee.ps  to  represent  your  script,  then  Scenepad  can  use  the  
kee.ps  API  to  retrieve  these  photos  and  display  them  as  shown  in  figure  6.19.  

Stewart  McKie   -­‐  287  -­‐   As  of  15/08/2014  

  

Figure  6.19  -­‐  Scenepad4  -­‐  showing  Images  sourced  from  Kee.ps  displayed  under  script  Keeps  
tab  

A  third  example  of  using  an  API  to  access  3rd  party  site  content  is  the  
displaying  of  tweets  linked  to  a  script  by  use  of  the  Twitter  API.  By  setting  up  
a  user  account  on  Twitter,  say  using  the  name  of  your  script,  multiple  users  
can  then  post  Tweets  to  this  account  and  these  tweets  displayed  in  Scenepad  
as  a  timeline  as  shown  in  figure  6.20.  

Figure  6.20  -­‐  Scenepad4    -­‐  showing  Tweets  sourced  from  Twitter  displayed  under  Script  Tweets  
tab  

Theoretically  any  content  posted  to  a  3rd  party  site,  with  an  API  that  
Scenepad  can  use,  could  be  retrieved  and  surfaced  in  Scenepad,  significantly  
extending  the  possibilities  of  the  screenplay  as  content  hub.  

Stewart  McKie   -­‐  288  -­‐   As  of  15/08/2014  

6.4  Scenepad  Analytics  

The  design  of  Scenepad  is  script  data-­‐centric  rather  than  script-­‐document  
centric.  The  content  of  the  script  is  contained  in  the  text  of  individual  data  
snippets,  which  are  in  turn  defined  contextually  by  their  snippet  and  scene  
metadata.  The  snippet  text  is  essentially  of  two  types:  dialog  and  action.  
Other  snippet  types,  such  as  ‘caption’  or  ‘transition’  function  as  metadata  
that  relate  more  to  scene  contextualization  and  the  filmic  flow  of  the  
content.  It’s  worth  remembering  that  an  important  purpose  of  content  
analytics  is  to  give  anyone  about  to  read  the  script  an  impression  of  the  
script  content  without  needing  to  read  it  first  and  to  help  writers  ‘see  the  
forest  from  the  trees’.    

So  in  what  ways  and  for  what  purposes  can  this  Scenepad  snippet  text  be  
analyzed?  

Entity  Grouping  

Scenes  are  an  obvious  entity  to  group  in  order  to  get  some  insight  into  
various  aspects  of  the  ‘balance’  of  a  script.  For  example,  scenes  can  be  
grouped  as  internal/external  and  day/night.  This  helps  give  an  immediate  
feel  for  script  balance  that  may  have  genre  implications  e.g.  more  night  than  
day  scenes  might  be  expected  in  a  horror,  or  may  indicate  budgetary  
considerations  e.g.  lots  of  external  night  scenes  may  be  more  difficult  and  
expensive  to  film.  Scenes  can  also  be  grouped  to  reflect  their  respective  
dialog/action  content.  This  helps  to  identify  a  potentially  ‘dialog-­‐heavy’  
script  and  those  scenes  that  standout  as  either  dialog  or  action-­‐centric.  

Stewart  McKie   -­‐  289  -­‐   As  of  15/08/2014  

  

Figure  6.21  -­‐  Scenepad1  Script  Dashboard  

Figure  6.22  -­‐  Scenepad1  Scene  Dashboard  

In  all  screenwriting  software,  the  scene  content  is  a  collection  of  snippets.  
Other  ways  to  group  snippets  are  by  role  or  location.  For  example,  in  
Scenepad  you  can  view  all  dialog  snippets  by  role  (role-­‐dialog)  and  all  action  
snippets  by  location  (location-­‐action).  The  purpose  of  this  simple  grouping  
analysis  is  to  help  the  writer(s),  the  talent  and  set/location  managers  focus  
on  a  specific  POV  of  the  script.  

Stewart  McKie   -­‐  290  -­‐   As  of  15/08/2014  

The  writer  is  helped  because  role-­‐dialog  grouping  facilitates  establishing  
consistency  of  tone  for  a  given  role.  A  quick  skim  of  a  role-­‐dialog  list  will  
help  to  identify  any  dialog  that  seems  ‘out  of  character’.  For  the  talent  
playing  the  role,  the  role-­‐dialog  list  isolates  all  their  dialog  across  the  whole  
script  so  it’s  easier  to  get  a  feel  for  their  complete  part,  from  a  dialog  
perspective.  The  location-­‐action  grouping  isolates  action  sequences  within  a  
specific  location,  action  that  may  span  multiple  scenes  using  the  same  
location,  so  that  production  managers  can  better  plan  their  assets  to  realize  
the  action.  

Figure  6.23  -­‐  Scenepad4  Role-­‐Dialog  

By  grouping  action/dialog  snippets  by  role,  writers  and  talent  can  get  a  
quick  idea  of  relative  part  size  and  by  identifying  a  role  as  protagonist  or  
antagonist  writers  can  get  an  idea  of  the  balance  of  these  two  key  roles  and  
when  they  enter  and  exit  the  script.    

Stewart  McKie   -­‐  291  -­‐   As  of  15/08/2014  

  

Figure  6.24  -­‐  Scenepad1  Top  10  Roles  

Figure  6.25  -­‐  Scenepad1  Protagonist  vs.  Antagonist  

Locations  can  be  grouped  by  their  use  in  the  script  and  by  the  number  of  
roles  that  are  involved  at  the  location.  This  provides  some  insight  into  the  
most  popular  and  ‘crowded’  locations  in  the  script.  

Stewart  McKie   -­‐  292  -­‐   As  of  15/08/2014  

  

Figure  6.26  -­‐  Scenepad1  Top  10  Locations  

Snippet  Text  Analysis  

The  grouping  analytics  discussed  above  is  not  focused  on  the  actual  text  and  
individual  words  contained  in  action/dialog  snippets,  whereas  snippet  text  
analysis  is.  Scenepad  provides  a  range  of  word  clouds  to  help  analyze  textual  
content.  The  word  clouds  are  generated  at  various  levels:  

Scriptcloud  –  word  clouds  of  all  text  in  the  script  

Dialog  Cloud  –  a  word  cloud  of  all  dialog  text  in  the  script  

Action  cloud  –  a  word  cloud  of  all  action  text  in  the  script  

Noun  cloud  –  a  word  cloud  containing  only  nouns  

Verb  cloud  –  a  word  cloud  containing  only  verbs  

Stewart  McKie   -­‐  293  -­‐   As  of  15/08/2014  

  

Figure  6.27  -­‐  Scenepad1  Script-­‐Level  Word  Clouds  

Like  those  on  the  original  Scriptcloud  site,  all  the  Scenepad  word  clouds  are  
subject  to  a  ‘stop-­‐list’.  And  like  Scriptcloud,  the  Scenepad  scriptclouds  give  
an  ‘impression’  of  the  vocabulary  of  the  script  by  displaying  ‘top  n’  words  by  
frequency  or  those  words  that  are  used  over  a  certain  limit  e.g.  all  verbs  
used  more  than  5  times.    

Unlike  Scriptcloud,  Scenepad  provides  forms  for  the  user  to  set  the  topN  
value  (e.g.  20  or  50)  and  frequency  limits  and  to  manage  (insert  and  delete)  
stoplist  words.  A  relatively  easy  addition  would  be  to  enable  a  script-­‐specific  
stoplist  –  for  example  to  handle  non-­‐English  scripts  since  the  default  stoplist  
includes  666  common  English  words  only.    

Unlike  Scriptcloud,  Scenepad  provides  some  ability  to  extract  both  Noun  and  
Verb  clouds  from  the  script  textual  content.  This  is  done  by  using  a  lexicon  of  
Stewart  McKie   -­‐  294  -­‐   As  of  15/08/2014  
tagged  parts  of  speech  based  on  the  Brown  Corpus,  compiled  in  the  1960s  by  
Henry  Kucera  and  W.  Nelson  Francis  at  Brown  University,  Providence,  
Rhode  Island.  The  code  used  to  parse  the  source  text  and  compare  it  to  the  
lexicon  is  based  on  the  tagger  code  originally  written  by  Eric  Brill  and  
subsequently  modified  by  Mark  Watson.    

Essentially  words  from  the  script  content  are  compared  to  the  part  of  speech  
classification  of  the  lexicon  word  tags  (e.g.  VB  to  VBZ  for  verbs  and  NN  to  NR  
for  nouns)  and  included  in  the  Noun  and  Verb  clouds  as  appropriate.  If  a  
word  is  tagged  with  both  noun  and  verb  tags  then  the  primary  tag  is  used  to  
define  which  cloud  the  word  appears  in.  For  example,  the  lexicon  entry  Aid  
NNP  NN  VB  is  classified  as  both  a  noun  and  a  verb  but  would  be  selected  
into  the  noun  cloud  rather  than  the  verb  cloud  whereas  Aim  VB  NN  NNP  
would  be  selected  into  the  verb  cloud  due  to  the  order  of  part  of  speech  
identifiers.  This  simplistic  approach  will  inevitably  result  in  some  words  
being  misclassified  so  the  word  clouds  are  unlikely  to  be  accurate  every  
time.  

The  addition  of  verb  and  noun  clouds  helps  to  also  give  an  impression  of  the  
vocabulary  of  action  and  dialog  snippets  from  the  perspective  of  two  key  
parts  of  speech  contained  in  a  screenplay.  Most  nouns  in  a  script  can  be  
directly  mapped  to  a  ‘prop’  used  in  a  production  set.  Most  verbs  in  a  script  
are  indicative  of  the  kind  of  action  that  the  script  embodies.    

A  potential  application  of  noun  and  verb  analytics  is  for  the  purposes  of  a  
marketing  firm  analyzing  a  script  for  product  placement  purposes  –  to  
identify  a  script  that  offers  a  context  that  presents  an  opportunity  for  
specific  product  placements.  Sternberg  (1997,  p.49)  gives  a  specific  example  
of  an  analysis  of  the  2nd  draft  of  the  script  of  Thelma  and  Louise  (1991)  by  
product  placement  agency  Prime  Time  Marketing  that  indicates  how  the  
availability  of  noun  and  verb  clouds  could  have  helped  the  agency  make  a  
decision  about  whether  and  where  to  use  product  placement  in  this  script.    

A  potential  application  of  verb  analytics  is  to  train  an  actor  in  a  particular  

Stewart  McKie   -­‐  295  -­‐   As  of  15/08/2014  

kind  of  action  so  they  can  perform  better  or  in  a  specific  ‘signature’  way.  A  
high  proportion  of  verbs  in  a  script  is  also  likely  to  indicate  an  action-­‐filled  
script,  which  in  turn  may  help  to  confirm  or  deny  the  genre  pretentions  of  
the  author.    

Thrulines  

Scenepad3  includes  thrulines  for  storylines,  keyroles  (antagonist  and  

protagonist)  and  nominated  characters.  The  thruline  simply  visualizes  
scenes  that  have  been  tagged  as  belonging  to  a  storyline  or  that  specific  roles  
or  characters  appear  in.      

Figure  6.28  -­‐  Scenepad3  -­‐  Alien  Keyrole  Thruline  

Figure  6.29  -­‐  Scenepad3  Alien  End  of  Script  Thruline  

The  set  of  thrulines  for  characters  in  Alien  shows  how,  towards  the  end  of  
the  script,  Ripley  is  the  only  crew  member  left,  which  either  indicates  total  
focus  on  a  single  character  towards  the  end  of  the  script  or  the  situation  in  
Alien  where  all  other  characters  are  dead.  In  figure  6.29  we  see  how  in  scene  
222  (in  this  version  of  the  script)  the  Alien  disposes  of  Dallas  (see  dot  in  row  

Stewart  McKie   -­‐  296  -­‐   As  of  15/08/2014  

5)  leaving  just  it  (row  2)  and  Ripley  (row8)  remaining.  Note  that  Ripley  is  
the  only  dot  left  at  the  end  of  the  script  (the  ship’s  cat  not  being  a  formal  
character).    

Sentiment  Analysis  

Scenepad3  includes  some  sentiment  analysis  capability  based  on  

functionality  provided  by  J  W  Hennessey’s  phpInsight  class  (see  
https://github.com/JWHennessey/phpInsight  -­‐  accessed  18  June  2013).  The  
sentiment  analysis  is  either  of  dialog  or  action  snippets  within  a  script  and  
provides  a  positive,  neutral  and  negative  rating  for  each  textual  snippet.  The  
analysis  is  based  on  ‘a  dictionary  of  words  that  are  categorised  as  positive,  
negative  or  neutral  and  a  naive  bayes  algorithm  to  calculate  sentiment.  To  
improve  accuracy  phpInsight  removes  'noise'  words.’  

As  with  the  stop  list  applied  to  word  clouds  this  dictionary  largely  
determines  the  polarity  of  the  text  and  so  the  efficacy  of  this  dictionary  is  
essential  to  the  accuracy  of  the  analysis.  

Figure  6.30  -­‐  Scenepad4  Dialog  Sentiment  Analysis  

Figure  6.30  shows  a  sentiment  analysis  of  every  snippet  in  the  Alien  script  

Stewart  McKie   -­‐  297  -­‐   As  of  15/08/2014  

(i.e.  its  positive,  neutral,  negative  score).  This  Scenepad  sentiment  analysis  
could  be  improved  in  future  by  applying  it  to  the  content  of  scenes  in  order  
to  chart  the  positive/negative/neutral  flow  of  the  script  scene-­‐by-­‐scene.  

6.5  Scenepad  for  Business  

Scenepad  recognizes  that  screenwriting  is  a  business  process  by  providing  

functionality  for  tracking  script  coverage  (see  figure  6.31)  and  submissions;  
for  posting  time  and  billing  to  a  script  and  for  managing  ‘to-­‐do’  tasks  related  
to  a  script.  This  enables  much  of  the  business  of  screenwriting  to  be  
managed  from  within  Scenepad  rather  than  in  a  separate  application  or  
spreadsheets.  Again,  as  mentioned  with  storyboarding  above,  a  ‘coverage’  
user  role  could  be  created  that  only  allows  viewing  of  scene  content  in  order  
to  allow  external  readers  to  easily  contribute  coverage  to  the  script.  

Figure  6.31  –  Scenepad4  Managing  Script  Coverage  

Scenepad  also  helps  to  manage  the  overall  ownership  of  a  script  that  may  
have  been  written  and  rewritten  by  multiple  parties.  Based  on  the  
ownership  of  a  snippet,  Scenepad  provides  a  dashboard  that  shows  how  
much  of  the  script  is  owned  by  the  top  5  users  who  have  worked  on  the  
script  –  in  terms  of  the  whole  script  content  and  the  action  or  dialog  content  
only,  as  shown  in  figure  6.32.  

Stewart  McKie   -­‐  298  -­‐   As  of  15/08/2014  

  

Figure  6.32  -­‐  Scenepad4  Script  Ownership  dashboard  

6.6  Screenplay  Production  

Scenepad  recognizes  that  screenwriting  and  screenplay  realization  is  a  long-­‐

duration,  social  process  that  includes  the  production  of  the  movie.  Most  
screenwriting  applications  do  not.  So  they  do  not  include  the  kind  of  
production  management  capabilities  found  in  packages  such  as  Gorilla  or  
Chimpanzee,  both  from  Jungle  Software.  

Scenepad  includes  a  complete  production  management  sub  module  for  

managing  shoot  days,  cast,  crew  and  props,  scene  storyboards  and  
production  budgets  for  example.  In  this  way  Scenepad  supports  multiple  
social  networks  by  using  the  datafied  screenplay  as  the  network  content  
anchor.    One  social  network  is  focused  on  the  writing/rewriting  of  the  
screenplay  content  and  another  is  focused  on  the  realization  of  the  
screenplay  content  and  an  audio  visual  work.  

Stewart  McKie   -­‐  299  -­‐   As  of  15/08/2014  

By  linking  to  and  leveraging  an  online  photo  archiving  site  (kee.ps.com),  the  
production  module  also  allows  practitioners  to  take  on-­‐set/on-­‐location  
photos  with  the  their  mobile  phones,  upload  them  to  kee.ps  and  then  access  
these  photos  through  Scenepad  as  a  record  of  the  shoot.  A  ‘shootstream’  if  
you  will.      

Figure  6.33  -­‐  Scenepad4  Script  Day  Content  Hub  

Summary  

The  various  Scenepad  prototypes  show  that  using  a  ‘datafied’  script  as  a  
foundation  enables  many  of  the  screenwriting  2.0  concepts  discussed  above  
to  be  applied  and  realized  in  screenwriting  software.  Scenepad4  provides  
significantly  more  screenplay  analytics  and  visualization  functions  than  are  
available  in  any  of  today’s  commercial  screenwriting  applications.  
Scenepad4  delivers  more  social  networking  capabilities  than  any  of  today’s  
screenwriting  applications.  Scenepad4  supports  more  screenplay  structural  
capabilities  that  most  of  today’s  screenwriting  applications.    

Scenepad4  is  a  fully  working  prototype  that  shows  the  potential  of  the  next  

Stewart  McKie   -­‐  300  -­‐   As  of  15/08/2014  

generation  of  commercial  screenwriting  2.0  applications.  It  contains  
imported  ‘reference’  scripts  e.g.  Alien  and  has  been  used,  by  me,  to  create  a  
full-­‐length  screenplay  script  from  scratch  (Reconciliation  –  referred  to  in  
some  of  the  figures  above).  Scenepad  is  only  an  initial  step  towards  
screenwriting  2.0  and  I  fully  understand  that  many  screenwriters  would  
neither  be  interested  in  nor  use  the  kind  of  functionality  that  Scenepad  
provides.  But  it  does  deliver  some  indication  of  ‘what  happens  next’  and  why  
the  datafication  of  screenplays  is  inevitable.  

Stewart  McKie   -­‐  301  -­‐   As  of  15/08/2014  

7.  What  Happens  Next?  

To  return  to  Mark  Norman’s  question  cited  at  the  beginning  of  this  thesis,  
the  datafication  of  screenplay  content  –  the  transition  from  a  document-­‐
centric  to  a  data-­‐centric  screenplay  -­‐  has  many  interesting  implications  for  
screenwriting  as  a  practice,  as  a  research  area  and  for  the  development  of  a  
new  generation  of  screenwriting  2.0  software  applications.    

But  first  it  will  require  moving  on  from  the  kind  of  thinking,  referred  to  by  
Edgar  (2009:6-­‐7)  in  How  Plays  Work,  that  persists  in  the  idea  that  creative  
works  are  not  subject  to  rules:  ‘writers  who’ve  read  any  twentieth-­‐century  
literary  theory  are  understandably  irked  by  the  arithmetical  reductionism  of  
so  much  thinking  in  this  field,  with  its  mechanical  lists,  symbols,  charts  and  
graphs.’    Edgar  reflects  that  rules  are  ‘a  sedimentation  of  all  of  the  plays  
(and,  to  an  extent,  all  of  the  stories)  that  we  have  ever  encountered’  and,  at  
some  level,  reflect  what  audiences  expect  to  see.    Here,  Edgar  is  referring  to  
play  scripts  and  a  theatre  audience  but  the  same  principles  apply  equally  
well  to  the  structure  debate  around  screenplays  and  a  cinema  audience.  

By  moving  beyond  this  kind  of  thinking,  designers/developers  of  new  

screenwriting  applications  may  reflect  approaching  the  screenplay  and  
screenwriting  from  different  ‘start  points’  than  that  of  traditional  screenplay  
application  designers.    

For  example,  it  is  much  more  likely  that  application  designers  will  approach  
screenwriting  from  a  social  networking  or  analytics  perspective  when  they  
come  to  design  the  next  generation  of  screenwriting  applications,  rather  
than  as  simply  a  way  to  write  and  print  out  a  ‘correctly’  formatted  script.    
The  influence  of  games  designers  may  mean  that  screenplay  applications  are  
approached  from  the  ‘start  point’  of  playing  the  script  and  managing  the  
branching  storylines  of  a  user-­‐controlled  narrative  for  example.  Or  the  
pervasive  use  of  mobile  devices  may  mean  that  screenplays  are  ‘reverse-­‐
engineered’  from  a  series  of  images  being  linked  together  using  simple  apps  

Stewart  McKie   -­‐  302  -­‐   As  of  15/08/2014  

akin  to  Disney’s  Story  (see  http://story.us)  where,  in  essence,  the  image  is  
used  as  the  stimulus  to  create  the  text  rather  than  the  other  way  around.  

Something  that  screenplay  datafication  will  also  help  to  prevent  is  the  
screenplay  as  pre-­‐formed  artefact  becoming  anymore  redundant  than  
improvisational  directors  have  made  it  already.  

Back  in  1974,  Metz  (1974:201-­‐204,  referring  to  Marcel  Martin)  discusses  
whether  a  “film-­‐makers  cinema”  was  replacing  a  “scriptwriter’s  cinema”.  
Referencing  the  work  of  French  director  Jean  Luc  Godard,  Metz  suggests  that  
Godard  ‘is  one  of  those  men  for  whom  inspiration  can  only  be  fired  during  
the  shooting’  and  that  it  is  during  the  shooting  that  the  scenario  is  ‘born’  as  a  
consequence  of  the  film.  In  his  section  ‘Ban  the  Script?’,  Lee  (2013:6-­‐10)  
refers  to  the  improvisational  preferences  of  British  director  Mike  Leigh  and  
Danish  director  Lars  Van  Trier’s  Dogme  movement    whose  ‘philosophy  was  
initially  to  get  totally  away  from  the  script.’    

This  might  suggest  that  the  future  of  the  screenplay  as  an  artefact  is  in  
doubt.  However,  this  is  unlikely  to  be  the  case  if  all  screenplays  are  datafied.  
In  fact,  in  this  case,  it  is  possible  that  it  is  filmmaking  as  solely  human-­‐
directed  process  that  is  in  doubt.  Pre-­‐viz  software  is  the  pathfinder  here.    If  
screenplays  are  datafied,  and  especially  if  they  are  datafied  using  a  
standardized  format,  then  it  is  only  a  matter  of  time  before  ever-­‐more  
sophisticated  algorithms  can  interpret  this  data  more  effectively  and  start  to  
generate  ‘first  cut’  movies  (initially  probably  only  animations),  directly  from  
the  script.    The  job  of  the  production  team  is  then  to  take  this  first  attempt  
and  develop  it  either  directly  (in  video)  or  indirectly  first  by  making  changes  
to  the  screenplay,  thus  changing  the  data,  and  re-­‐generating  the  movie  from  
the  script.  Somehow  this  does  not  feel  like  it  is  many  years  off.  

In  this  scenario  the  screenplay  is  not  something  you  can  ‘get  way  from’  since  
the  screenplay  data  is  the  movie  or  at  least  a  significant  part  of  it.  CAD/CAM  
becomes  SGF  or  screenplay  generated  filmmaking.  And  this  will  not  just  
apply  to  animations,  where  the  utility  and  practicality  of  this  approach  is  

Stewart  McKie   -­‐  303  -­‐   As  of  15/08/2014  

obvious.  It’s  not  hard  to  see  that  with  the  databases  of  a  Google  providing  
access  to  images  from  across  the  web,  real-­‐world  ‘streetviews’  and  satellite  
images  of  almost  everywhere  on  earth  or  a  YouTube  full  of  previously  
released  commercial  movies  and  clips,  and  millions  of  user-­‐generated  ‘home’  
movies  and  clips  there  is  already  a  vast  archive  of  digital  content  for  a  
screenwriting  application  to  search  and  retrieve  from.    

For  example,  take  a  single  action  snippet  from  a  ‘human’  movie  or  a  hybrid  
‘humanimation’,  combining  human  actors  with  CGI  background,  that  can  be  
translated  into  a  short  sequence  or  a  single  shot  like:    

 They  paused  and  watched  the  setting  sun  disappear  into  the  Ocean.  

or  

There  it  is!  A  1920  yellow  Rolls-­‐Royce.  

The  sunset  clip  or  the  clip  of  a  Rolls-­‐Royce  could  simply  retrieved  from  the  
online  database  and  inserted  into  the  movie’s  screenplay  database  as  a  link  
to  the  online  clip  that  is  in  turn  linked  to  the  ‘action’  snippet  data  record.  

Basically  this  only  requires  a  search-­‐and-­‐retrieve  algorithm  that  processes  

the  screenplay  data  and  proposes  clips  or  images  by  ‘reaching  out’  to  online  
databases  that  the  software  automatically  inserts  into  the  proto  movie’s  
timeline.  The  ‘filmmaker’  can  then  choose  from  these  ‘suggestions’  or  use  
them  as  input  to  inform  a  more  refined  search  of  their  own.    

This  promises  a  new  kind  of  automated  ‘bricolage’  to  reference  Wells’  article  
in  Analysing  the  Screenplay  (2011)  that  takes  the  idea  of  Ruecker  et  al.  of  
‘playing’  the  theatre  play  script  (see  
http://academia.edu/3155085/WATCH_MY_MOVES_FROM_DIGITAL_PLAYS
_TO_THE_DIGITAL_PLAYBOOK  -­‐  accessed  11  July  2013)  to  a  whole  new  
level.    

Aside  from  automated  movie  generation  from  a  script,  which  is  surely  the  

Stewart  McKie   -­‐  304  -­‐   As  of  15/08/2014  

most  important  new  development  that  screenplay  as  data  can  enable,  the  
future  of  screenwriting  technology  and  move  to  screenwriting  2.0  is  likely  to  
be  characterized  by  a  number  of  other  innovations  resulting  from  the  
improved  application  of  technology  to  the  screenplay  as  artefact  or  
screenwriting  as  a  process.    

Innovation  could  emanate  from  the  use  of  ‘cloud’  technology  both  to  store  
and  share  scripts  (already  widely  in  use  now  by  many  online  screenwriting  
applications  like  Scenepad)  and  application  programming  interfaces  or  APIs  
(surfaced  as  web  services)  for  deploying  and  accessing  applications,  
mashing  them  together  and  storing  and  accessing  content;  the  pervasiveness  
of  social  networking,  collaboration  and  crowdsourcing;  and  the  form  factor  
and  new  user  interface  (UI)  controls  -­‐  including  touch,  swipe  and  pinch  –  
offered  by  mobile  platforms  such  as  the  iPhone  and  iPad.  

The  new  tactile  UIs  are  not  just  changing  the  way  we  work  with  content  on  
handheld  devices.  Large,  wall-­‐based,  touch  screens  are  also  an  interesting  
possibility  for  placing  the  script,  and  specifically  scene  content  within  a  
script,  at  the  centre  of  a  display  of  content  linked  to  the  script/scene.  This  
kind  of  large  tactile  surface  was  glimpsed  in  Minority  Report  (2002)  and  
more  recently  in  Den  Som  Draeber  (2010)  and  is  implemented  in  the  real  
world  by  Hornall  Anderson  and  Nike  among  others.  These  ‘touch-­‐walls’  
present  a  content  ‘canvas’  that  could  enable  writing  teams  to  work  in  a  
visual,  collaborative  and  tactile  way  with  both  text  and  images  to  develop  
script  content  and  re-­‐fashion,  rather  than  just  rewrite,  content  relationships  
with  the  swipe  of  a  hand.    Dynamic,  node-­‐based  representations  of  content  
and  relationships,  such  as  those  facilitated  by  arbor.js  
(http://www.arborjs.org)  are  one  dynamic  way  to  present  data  on  this  kind  
of  touch-­‐wall.  

Screenwriting  2.0  assumes  that  a  screenplay  is  a  dynamic,  online  and  

interactive  audio-­‐visual-­‐textual  artefact  that  is  subject  to  continual  
enhancement  and  enrichment.  It  is  not  a  paper-­‐based  document  but  a  digital  
artefact  subject  to  an  ongoing  development  process  that  many  roles,  other  

Stewart  McKie   -­‐  305  -­‐   As  of  15/08/2014  

than  ‘writing’  roles  can  participate  in  and  contribute  to  as  part  of  the  
screenplay  development  project  team.  Clearly,  these  stakeholders  will  be  
able  to  interact  with  this  screenplay  artefact  in  many  different  ways  and  on  
or  from  many  different  devices  -­‐  something  that  will  inevitably  also  impact  
the  process  and  practice  of  screenwriting.  

By  framing  screenwriting  as  a  multi-­‐user  business  -­‐  rather  than  only  a  

single-­‐user  creative,  process  -­‐  future  screenwriting  applications  must  look  
and  feel  more  like  a  social  networking  application  rather  than  a  document  
writing  application.  By  recognizing  that  screenwriting  is  a  process,  writers  
may  be  encouraged  to  view  their  screenplay  as  a  social  artefact  and  expect  
that  their  content  can  benefit  from  exposure  to  different  stakeholders  in  the  
their  screenplay  community  of  interest.    

The  screenplay  itself  will  become  an  interactive  online  text  where  the  
anchoring  reality  of  the  scene  text  is  augmented  by  layers  of  additional  
linked  content  that  surrounds  and  enriches  this  text.  Of  course  the  printed  
script  will  not  go  away  anytime  soon  but  how  long  can  it  be  before  the  
digital  script  on  an  iPad  essentially  replaces  the  printed  document  on  set  
and  augmented  reality  screenplays  become  the  norm?    

Much  of  this  linked  content  will  come  directly  from  user-­‐generated  content  
captured  on  mobile  devices.  These  devices  will  act  as  the  primary  means  to  
capture  and  connect  content  to  the  screenplay  such  as  photos,  audio  and  
video  clips.  This  content  will  flow-­‐through  to  the  screenplay  via  an  
application  programming  interface  (API)  that  enables  any  external  
application  to  interface  to  it  –  both  to  put  data  into  the  screenplay  and  to  get  
data  from  it.    In  this  sense,  the  screenplay  becomes  the  analog  of  the  
Facebook  ‘wall’  that  a  community  of  interest  engages  with  to  enhance  and  
enrich  the  core  content.  

This  community  engagement  could  prove  especially  useful  in  the  production  
phase  of  the  screenplay  realization  process  for  activities  such  as  location  
management,  shot  planning  and  script  continuity.  The  mere  fact  that  it  will  

Stewart  McKie   -­‐  306  -­‐   As  of  15/08/2014  

become  easier  to  capture  and  connect  this  content  will  itself  mean  that  the  
augmented  screenplay  will  inevitably  become  a  reality  in  the  same  way  that  
a  Twitter  timeline  or  Flickr  portfolio  receives  a  constant  stream  of  fresh  
content  from  a  user  generation  that  is  used  to  capturing  all  kinds  of  content  
whenever,  wherever.  

There’s  no  reason  why  this  new  kind  of  datafied  screenplay  could  not  also  be  
managed  on  its  own  dedicated  device  –  a  ‘Screenplay  Kindle’  if  you  will.  By  
tying  the  hardware  to  a  known,  datafied  content  format  the  device  can  
provide  additional  capabilities  that  enhance  the  user  experience  when  you  
interact  with  the  content,  for  example:    

-­‐  VCR  Buttons  to  jump  through  the  script  by  scene  or  by  role  or  by  location  

-­‐  A  button  to  trigger  a  Siri-­‐like  voice  to  read  out  the  dialog  within  a  scene  

-­‐  A  button  to  mark  a  note  entry  in  the  text  and  type  in  the  note  content      

The  ability  to  build  a  device  dedicated  to  screenplay  content  would  be  a  lot  
easier  if  a  standard  screenplay  data  format  (e.g.  an  XML  schema)  was  agreed  
that  is  ‘open’  and  not    ‘owned’  by  a  specific  software  vendor  (e.g.  Final  Draft).  
Once  a  properly  documented  standard  is  adopted,  this  is  likely  to  encourage  
the  wider  software  development  community  to  engage  with  screenplay  
content  management  as  a  domain  and  help  to  drive  the  development  of  low-­‐
cost  add-­‐on  apps  to  further  enhance  the  screenwriting  experience.  

Datafication  of  screenplays  means  that  stakeholders  will  come  to  expect  a  
great  deal  more  from  their  screenwriting  application  in  the  way  of  analytics.  
Writers  will  expect  more  analysis  to  help  them  improve  the  quality  of  their  
script,  financial  stakeholders  will  expect  more  analysis  to  help  them  judge  
the  commercial  potential  of  a  script,  and  script  readers  and  analysts  will  
expect  more  data-­‐based  evidential  analytics  to  support  or  challenge  their  
intuition-­‐based  judgments  of  a  screenplay.  Some  of  the  visualizations  that  
the  how-­‐to  manuals  have  presented  as  diagrams  will  be  able  to  be  generated  
directly  from  the  screenplay  content  at  the  click  of  a  button.  

Stewart  McKie   -­‐  307  -­‐   As  of  15/08/2014  

Some  of  these  visualizations  should  become  more  accurate  and  reliable  over  
time.  For  example  as  further  research  determines  the  optimum  stop  list  for  a  
text  cloud  or  the  optimum  positive/neutral/negative  indicators  for  
sentiment  analysis  not  just  for  screenplays  generally  but  specifically  by  
screenplay  genre.  Indeed,  a  set  of  consistently  datafied  scripts  within  a  given  
genre  should  provide  a  sufficiently  interesting  ‘big  data’  corpus  to  drive  a  
whole  new  research  area  -­‐  screenplay  genre  analytics  -­‐  that  could  have  
significant  impact  on  the  definition  of  screenplay  genres  and  therefore  what  
constitutes  genre  conformity  or  subversion  in  a  given  screenplay  instance.  
Once  this  online  genre  database  is  available  individual  scripts  can  be  
submitted  for  analysis  and  comparison  to  genre  norms  and  some  kind  of  
‘genre  conformance’  report  produced  as  feedback  for  the  script  
writer/owner.  

A  library  of  screenplay  algorithms  could  be  developed  in  order  to  define  
common  analytical  tasks  such  as  how  to  generate  a  role  throughline  or  
identify  a  turning  point.  The  potential  of  event  identification  in  a  screenplay  
also  seems  a  rich  area  for  further  research.  If  events  can  be  recognized  and  
subsequent  patterns  visualized  there  is  real  potential  for  the  deep  structure  
of  a  screenplay  to  be  surfaced  evidentially  rather  than  intuitively.  This  could  
re-­‐energize  the  screenplay  structure  debate  in  that  this  kind  of  data-­‐driven  
emergent  structure  is  potentially  much  more  interesting  to  the  screenwriter  
as  questioning  tool  than  attempting  to  write  to  some  deliberately  imposed  n-­‐
act  structure.    

In  terms  of  screenwriting  2.0,  it  is  the  datafication  of  screenplays  that  will  
fundamentally  drive  ‘what  happens  next’  and  there  is  every  reason  to  be  
excited  about  what  that  ‘next’  could  be  in  terms  of  the  screenplay  as  artefact  
and  screenwriting  as  practice.    My  Scenepad  application  deliverable  merely  
scratches  the  surface  of  all  this  potential  but  does  deliver  an  indication  of  
how  the  datafication  of  the  screenplay  can,  I  believe,  positively  impact  
creating  and  managing,  presenting  and  sharing,  analyzing  and  visualizing  
textual  screenplay  content.  

Stewart  McKie   -­‐  308  -­‐   As  of  15/08/2014  

Bibliography  

Anderson,  Chris  (2012).  Makers:    The  New  Industrial  Revolution.  London:  

Random  House  Business  Books.  

Argentini,  Paul  (1998).  Elements  of  Style  for  Screenwriters.  Los  Angeles:  Lone  
Eagle  Publications.        

Arijon,  Daniel  (1976).  Grammar  of  the  Film  Language.  New  York:  Focal  Press.        

Armer,  Alan  A.  (1993).  Writing  the  Screenplay.  2nd  Ed.  Belmont,  CA:  Wadsworth  
Pub.  Co.        

Aronson,  Linda  (2001).  Screenwriting  Updated  New  (and  Conventional)  Ways  of  
Writing  for  the  Screen.  Los  Angeles:  Silman-­‐James  Press.    

Aronson,  Linda  (2010).  The  21st  Century  Screenplay.  Los  Angeles:  Silman-­‐James  
Press.        

Aristotle,  Poetics,  trans.  Malcolm  Heath,  London:  Penguin,  1996.  

Baboulene,  David  (2010).  The  Story  Book.  DreamEngine  Media.  

Batty,  Craig  and  Waldeback,  Zara  (2008),  Writing  for  the  Screen;  Creative  and  
Critical  Approaches.  Basingstoke:  Palgrave  Macmillan.  

Borst,  Terry  (2012).  Mastering  Celtx.  Boston:  Course  Technology.  

Bruhn  Jensen,  Klaus  (Ed.)(2002).  A  Handbook  of  Media  and  Communication  

Research:  Qualitative  and  Quantitative  Methodologies.  London:  Routledge.  

Card,  Stuart,  MacKinlay  Jock  and  Shneiderman,  Ben  (1999).  Readings  in  
Information  Visualization:  Using  Vision  to  Think.    London:  Academic  Press.  

Chatman,  Seymour  (1980).  Story  and  Discourse:  Narrative  Structure  in  Fiction  

Stewart  McKie   -­‐  309  -­‐   As  of  15/08/2014  

and  Film.  Ithaca,  New  York:  Cornell  University  Press.  

Cockburn,  Alistair  (2005).  Crystal  Clear:  A  Human-­‐Powered  Methodology  for  

Small  Teams.  Upper  Saddle  River  NJ:  Pearson  Education  Inc.  

Cole,  Hillis  R.,  and  Haag,  Judith  H.  (1989).  The  Complete  Guide  to  Standard  Script  
Formats  –  Part1:  The  Screenplay.  Los  Angeles:  CMC  Publishing.  

Cooper,  Dona  (1994).  Writing  Great  Screenplays  for  Film  and  TV.  New  York:  
Prentice  Hall.      

Corrigan,  Timothy  J.  (2007).  A  Short  Guide  to  Writing  About  Film.  New  York:  
Pearson  Longman.    

Cowgill,  Linda  J.  (1999).  Secrets  of  Screenplay  Structure  How  to  Recognize  and  
Emulate  the  Structural  Frameworks  of  Great  Films.  Los  Angeles:  Lone  Eagle  
Publications.        

Cowgill,  Linda  J.  (1997).  Writing  Short  Films  Structure  and  Content  for  
Screenwriters.  Los  Angeles:  Lone  Eagle  Publications.        

Cron,  Lisa  (2012).  Wired  for  Story.  New  York:  Ten  Speed  Press.  

Decker,  Dan  (1998).  Anatomy  of  a  Screenplay.  Chicago:  The  Screenwriters  

Group.  

De  Haan,  Erik  (2004).  The  Consulting  Process  as  Drama.  Learning  from  King  Lear.  
London:  Karnac.        

Donovan,  John  J.,  ‘Three-­‐tiered  Client-­‐Server  Systems  Gain  as  Enterprise-­‐Wide  

Solution’,  Interview,  Investor's  Business  Daily,  (October  1994).  

Edgar,  David  (2009).  How  Plays  Work.  2nd  Ed.  London:  Nick  Hern  Books.  

Edwards,  Rona  and  Skerbelis,  Monika  (2009).  I  Liked  It.  Didn’t  Love  It.  2nd  Ed.  
Beverly  Hills:  Edwards  Skerbelis  Entertainment.  

Stewart  McKie   -­‐  310  -­‐   As  of  15/08/2014  

Egri,  Lajos  (2004).  The  Art  of  Dramatic  Writing.  New  York:  Touchstone  (original  
edition,  1942).      

Eliashberg  Jehoshua,  Hui  Sam  K.,  Zhang  John  Z.  From  Story  Line  to  Box  Office:  A  
New  Approach  for  Green-­‐Lighting  Movie  Scripts  in  Management  Science  Vol.  53,  
No.  6,  June  2007,  pp.881-­‐893.  

Elberse,  Anita  (2013).  Blockbusters:  Why  Big  Hits  –  and  Big  Risks  –  are  the  Future  
of  the  Entertainment  Business.  New  York:  Henry  Holt  

Epstein,  Alex  (2002).  Crafty  Screenwriting:  Writing  Movies  That  Get  Made.    New  
York:  Henry  Holt.  

Epstein,  Edward  Jay  (2012).  The  Hollywood  Economist:  The  Hidden  Financial  
Reality  Behind  the  Movies.  New  York:  Melvyn  House  Publishing.      

Few,  Stephen  (2006).  Information  Dashboard  Design.  Sebastopol,  CA:  O’Reilly  

Media.  

Field,  Syd  (2005).  Screenplay:  The  Foundations  of  Screenwriting.  Rev.  Ed.  New  
York:  Delta  Trade  Paperbacks.        

Flinn,  Denny  Martin  (1999).  How  Not  to  Write  a  Screenplay.  New  York:Watson-­‐
Guptill  Publications.  

Forster,  E.  M.  (1927).  Aspects  of  the  Novel.  London:  Penguin  Books,  2005.  

Franzosi,  Roberto  (2010).  Quantitative  Narrative  Analysis.  London:  SAGE  

Publications.  

Frensham,  Ray  (2003).  Teach  Yourself  Screenwriting.  2nd  Ed.  London:  Hodder  
Headline  Ltd.    

Freytag,  Gustav  (1899).  Technique  of  the  Drama:  An  Exposition  of  Dramatic  
Composition  and  Art.  (trans.  Elias  J.  McEwan).  Amsterdam:  Fredonia  Books,  

Stewart  McKie   -­‐  311  -­‐   As  of  15/08/2014  

2005.  

Ganz,  Adam  (2011).  ‘Let  the  audience  add  up  two  plus  two.  They’ll  love  you  
forever’  in  Nelmes,  Jill  (ed.)  Analysing  the  Screenplay.  Oxford:  Routledge,  127-­‐
141.  

Garfinkel,  Asher  (2007).  Screenplay  Story  Analysis:  The  Art  and  Business.  New  
York:  Allworth  Press.    

Gaurav,  Vaish  (2013).  Getting  Started  with  NoSQL.  Birmingham:  Packt  

Publishing.  

Geroimenko,  Vladimir,  and  Chaomei  Chen  (2006).  Visualizing  the  Semantic  Web:  
XML-­‐Based  Internet  and  Information  Visualization.  2nd  Ed.  London:  Springer.    

Goldman,  William  (1996).  Adventures  in  the  Screen  Trade.  London:  Abacus.    

Gotham  Writers  Workshop  Faculty  (2006).  Writing  Movies.  Ed.  Alexander  Steele.  
New  York:  Bloomsbury.  

Grove,  Elliot  (2001).  Raindance  Writer's  Lab  Write  +  Sell  the  Hot  Screenplay.  
Boston:  Focal  Press.        

Hoad,  Phil  (2007).  Stars?  Check.  Script?  Don’t  need  one  yet…  The  Guardian  (Film  
Section)  13  July  2007.  

Horton,  Andrew  (1999).  Writing  the  Character-­‐Centered  Screenplay.  2nd  Ed.  

Berkeley:  University  of  California  Press.        

Indick,  William  (2004).  Psychology  for  Screenwriters:  Building  Conflict  in  Your  
Script.  Studio  City,  CA:  Michael  Wiese  Productions.        

James,  Henry  (1884).  ‘The  art  of  fiction’  in  Shapira,  Morris  (ed.),  Henry  James:  
Selected  Literary  Criticism.  London:  Penguin,  1963.  

Kaplan,  R.S.  and  Norton  D.P.  (1996).  Using  the  balance  scorecard  as  a  strategic  

Stewart  McKie   -­‐  312  -­‐   As  of  15/08/2014  

management  system  in  Harvard  Business  Review,  January/February,  1996.    

Karetnikova,  Inga  (1990).  How  Scripts  are  Made.  Carbondale  IL:  Southern  Illinois  
University  Press.  

Kawin,  Bruce  F  (1992).  How  Movies  Work.  Berkeley  and  Los  Angeles:  University  
of  California  Press.  

Kenning,  Jennifer  (2006).  Writing  a  Great  Movie:  Key  Tools  for  Successful  
Screenwriting.  New  York:  Continuum.        

Kitchen,  Jeff  (2006).  How  to  Be  Your  Own  Script  Doctor.  New  York:  Lone  Eagle  
Publications.        

Lacey,  Nick  (2000).  Narrative  and  Genre  Key  Concepts  in  Media  Studies.  
Basingstoke:  Palgrave.      

Langford,  Barry  (2011).  ‘Beyond  McKee’  in  Nelmes,  Jill  (ed.)  Analysing  the  
Screenplay.  Oxford:  Routledge,  251-­‐262.  

Lavandier,  Yves  (2004).  Writing  Drama.  Trans.  Bernard  Besserglik.  Paris:  Le  
Clown  &  L’Enfant.  

Lee,  Jason  (2013).  The  Psychology  of  Screenwriting:  Theory  and  Practice.  
London:  Bloomsbury  Publishing  plc.  

Lewin,  K  (1946).  Action  research  and  minority  problems.  Journal  of  Social  Issues,  
2(4),  34-­‐46.  

Liu,  Zhi-­‐Qiang.  Script  Visualization  (ScriptViz):  a  Smart  System.  Dissertation.  

Centre  for  Lucey,  Paul  (1996).  Story  Sense:  Writing  Story  and  Script  for  Feature  
Films  and  Television.  New  York:  McGraw-­‐Hill.        

Mackendrick,  Alexander,  and  Paul  Cronin  (2004).  On  Film-­‐Making:  An  

Introduction  to  the  Craft  of  the  Director.  London:  Faber  and  Faber.    

Stewart  McKie   -­‐  313  -­‐   As  of  15/08/2014  

MacDonald,  Ian  W.  (2004),  ‘Disentangling  the  Screen  Idea’  in  Journal  of  Media  
Practice,  5:2,  89–99.  

(2010)    ‘…So  it’s  not  surprising  I’m  neurotic’  The  Screenwriter  and  the  Screen  
Idea  Work  Group’  in  Journal  of  Screenwriting,  1:1,  45-­‐58.  

Mamet,  David  (2007).  Bambi  vs.  Godzilla.  London:  Simon  &  Schuster  UK  Ltd.  

Maras,  Steven  (2009).  Screenwriting:  History,  Theory  and  Practice.  London:  

Wallflower  Press.  

Matteo,  Civaschi  and  Milesi,  Gianmarco  (2012).  Life  in  Five  Seconds:  Over  200  
Stories  for  Those  With  No  Time  to  Waste.  London:  Quercus  Publishing  Ltd.  

Mayer-­‐Schönberger,  Viktor  and  Cukier,  Kenneth  (2013).  Big  Data.  London:  John  
Murray  (Publishers).  

McGinn,  Colin  (2005).  The  Power  of  Movies  How  Screen  and  Mind  Interact.  New  
York:  Pantheon  Books.    

McKee,  Robert  (1997).  Story  Substance,  Structure,  Style  and  the  Principles  of  
Screenwriting.  New  York:  Regan  Books.        

McKie,  Stewart  (2006a).  ‘Mariner  Montage:  Screenwriting  software  with  a  

marketing  edge’  in  ScriptWriter,  September  2006,  49-­‐51.  

McKie,  Stewart  (2006b).  ‘Sophocles’  in  ScriptWriter,  November  2006,  36-­‐39.  

Metz,  Christian  (1974).  Film  Language:  A  Semiotics  of  the  Cinema.  Chicago:  
University  of  Chicago  Press.  

Millard,  Kathryn  (2010).  ‘After  the  typewriter:  the  screenplay  in  a  digital  era’  in  
Journal  of  Screenwriting,  1:1,  11-­‐25.  

Millard,  Kathryn  (2011).  ‘The  screenplay  as  prototype’  in  Nelmes,  Jill  (Ed.)  
Analysing  the  Screenplay.  Oxford:  Routledge,  142-­‐157.  

Stewart  McKie   -­‐  314  -­‐   As  of  15/08/2014  

Murdock,  Maureen  (1990).  The  Heroine's  Journey.  Boston:  Shambala  
Publications.  

Norman,  Marc  (2008).    What  Happens  Next.  A  History  of  American  

Screenwriting.  London:  Aurum  Press  Ltd.    

O’Bannon,  Dan  with  Lohr,  Matt  R  (2013).  Dan  O’Bannon’s  Guide  to  Screenplay  
Structure.  Studio  City,  CA:  Michael  Wiese  Productions.      

Palmer,  Frederick  (1919).  Palmer  Plan  Handbook.  Los  Angeles:  Palmer  Photoplay  
Corporation.  

Parker,  Philip  (1999).  The  Art  and  Science  of  Screenwriting.  Exeter:  Intellect.      

Pepperman,  Richard  D  (2005).  Setting  Up  Your  Scenes:  The  Inner  Workings  of  
Great  Films.  Studio  City,  CA:  Michael  Wiese  Productions.        

Pfister,  Manfred  (1977).  The  Theory  and  Analysis  of  Drama.  Cambridge:  
Cambridge  University  Press,  1991.  

Pope,  Thomas  (1998).  Good  Scripts,  Bad  Scripts  Learning  the  Craft  of  
Screenwriting  Through  the  25  Best  and  Worst  Films  in  History.  New  York:  Three  
Rivers  Press.  

Price,  Steven  (2010).  The  Screenplay:  Authorship,  Theory  and  Criticism.  

Basingstoke,  Hants:  Palgrave  Macmillan.        

Propp,  V  (1968).  Morphology  of  the  Folktale.  2nd  Ed.  Austin,  TX:  University  of  
Texas  Press.        

Rimmon-­‐Kenan,  Shlomith  (2002).  Narrative  Fiction.  2nd  Ed.  London:  Routledge.  

Savin-­‐Baden,  Maggi  and  Howell  Major,  Claire  (2013).  Qualitative  Research:  The  
essential  guide  to  theory  and  practice.  Abingdon:  Routledge.  

Scofield,  Sandra  (2007).  The  Scene  Book:  A  Primer  for  the  Fiction  Writer.  

Stewart  McKie   -­‐  315  -­‐   As  of  15/08/2014  

Penguin:  London.  

Seger,  Linda  (2003).  Advanced  Screenwriting  Raising  Your  Script  to  the  Academy  
Award  Level.  Los  Angeles:  Silman-­‐James  Press.      

Sher,  Lucy  (2011).  Reading  Screenplays:  How  to  analyze  and  evaluate  film  
scripts.  Harpenden:  Kamera  Books.  

Siegel,  Eric  (2013).  Predictive  Analytics:  The  Power  to  Predict  Who  Will  Click,  Buy,  
Lie,  or  Die.  New  Jersey:  John  Wiley.  

Silver-­‐Lasky,  Pat  (2004).  Screenwriting  for  the  21st  Century.  London:  Batsford.    

Sims,  Chris  and  Johnson,  Hillary  Louise  (2012).  Scrum:  A  Breathtakingly  brief  and  
agile  introduction.  Dymaxicon.  

Snyder,  Blake  (2005).  Save  the  Cat!  The  Last  Book  on  Screenwriting  You'll  Ever  
Need.  Studio  City,  CA:  Michael  Wiese  Productions.    

Sternberg,  Claudia  (1997).  Written  for  the  Screen:  The  American  Motion-­‐Picture  
Screenplay  as  Text.  Tubingen:  Stauffenburg  Verlag.  

Suppa,  Ron  (2005).  Real  Screenwriting  Strategies  and  Stories  From  the  Trenches.  
Boston:  Thomson  Course  Technology  PTR.  

Takach,  Geo  (2006).  ‘Projecting  Knowledge  Management:  Screenwriting  As  A  

Tool  For  Knowledge-­‐Sharing  And  Action’  in  Journal  of  Knowledge  Management  
Practice,  Vol.  7,  No.  1,  March  2006.  

Tierno,  Michael  (2002).  Aristotle's  Poetics  for  Screenwriters  Storytelling  Secrets  

From  the  Greatest  Mind  in  Western  Civilization.  1st  Ed.  New  York:  Hyperion.        

Tufte,  Edward  R  (1990).  Envisioning  Information.  Cheshire,  Conn.:  Graphics  

Press.    

Vaish,  Guarav  (2013).  Getting  Started  with  NoSQL.  Birmingham:  Packt  

Stewart  McKie   -­‐  316  -­‐   As  of  15/08/2014  

Publishing.  

Vale,  Eugene  (1944).  The  Technique  of  Screenplay  Writing.    London:  Souvenir  
Press,  1973.  

Van  Sijll,  Jennifer  (2005).  Cinematic  Storytelling  the  100  Most  Powerful  Film  
Conventions  Every  Filmmaker  Must  Know.  Studio  City,  CA:  Michael  Wiese  
Productions.        

Vogler,  Christopher  (1998).  The  Writer's  Journey:  Mythic  Structure  for  Writers.  
2nd  Ed.  Studio  City,  CA:  Michael  Wiese  Productions.        

Voytilla,  Stuart  (1999).  Myth  and  the  Movies.  Studio  City,  CA:  Michael  Wiese  
Productions.  

Wells,  Paul  (2011).  ‘Boards,  beats,  binaries  and  bricolage’,  in  Nelmes,  Jill  (ed.)  
Analysing  the  Screenplay.  Oxford:  Routledge,  89-­‐105.  

Werbach,  Kevin  and  Hunter,  Dan  (2012).  For  the  Win:  How  Game  Thinking  Can  
Revolutionize  Your  Business.  Philadelphia:  Wharton  Digital  Press.  

Wright,  Kate  (2004).  Screenwriting  is  Storytelling:  Creating  an  a-­‐List  Screenplay  
That  Sells.  New  York:  A  Perigee  Book.    

Yanno,  Drew  (2006).  The  3rd  Act:  Writing  a  Great  Ending  to  Your  Screenplay.  
London:  Continuum.  

Yorke,  John  (2013).  Into  the  Woods:  A  Five-­‐Act  Journey  into  Story.  London:  
Pearson.  

Stewart  McKie   -­‐  317  -­‐   As  of  15/08/2014  

Screenwriting  and  Other  Software  References  

Amazon  Studios.  
 http://studios.amazon.com/storyteller    -­‐  accessed  11  June  2013.  

BigML  
https://bigml.com  -­‐  accessed  18  March  2014.  

Celtx  
http://www.celtx.com/  -­‐  accessed  23  January  2012.  

Character  Writer  
 http://typingchimp.com/characterwriter/index.html  -­‐  accessed  23  January  
2012.  

Dataseed  
 http://getdataseed.com  -­‐  accessed  18  March  2013.    

Dramatica  Pro  
http://www.dramatica.com/theory/articles/index.htm  -­‐  accessed  23  January  
2012.  

easySCOTT  
http://www.easyscott.de/en/easyscott.html  -­‐  accessed  17  June  2013.  

Final  Draft    
http://www.finaldraft.com/products/  -­‐  accessed  23  January  2012.  

Final  Draft  AV  

 http://www.finaldraft.com/company/about-­‐us.php  -­‐  accessed  23  January  2012.  

FrameForge  
http://www.frameforge3d.com/kb/entry/25/  -­‐  accessed  23  January  2012.  

Google  Sketchup  

Stewart  McKie   -­‐  318  -­‐   As  of  15/08/2014  

http://www.sketchup.com  -­‐  accessed  01  February  2013.  

Gorilla  
http://www.junglesoftware.com/home/  -­‐  accessed  23  January  2012.  

Highland/Fountain  
http://johnaugust.com/2013/highland-­‐ships  -­‐  accessed  29  March  2013.  

Logline  
https://loglineapp.com  -­‐  accessed  29  March  2013.  

Montage  
 https://www.marinersoftware.com/discussions/forum/montage/  -­‐  accessed  17  
June  2013.  

Movie  Outline  
http://www.movieoutline.com/  -­‐  accessed  23  January  2012.  

PC-­‐ACE  
http://sociology.emory.edu/faculty/rfranzosi/pc-­‐ace/  -­‐  accessed  30  December  
2013.  

Persona  
https://www.marinersoftware.com/products/persona/  -­‐  accessed  29  March  
2013.  

Plotbuilder  
http://www.plotbuilder.com  -­‐  accessed  17  June  2013.  

Script  Supervisor  
http://peterskarratt.com  -­‐  accessed  17  June  2013.  

ShotList  
 http://solubleapps.com/shotlist/  -­‐  accessed  18  March  2014.  

Stewart  McKie   -­‐  319  -­‐   As  of  15/08/2014  

Sophocles  
http://en.wikipedia.org/wiki/Sophocles_-­‐  accessed  software  -­‐  accessed  29  
March  2013.  

Splunk  
 http://splunk.com  -­‐  accessed  18  March  2014.  

StoryBoard  Artist  Studio  

http://www.powerproduction.com/artist_studio.html  -­‐  accessed  23  January  
2012.  

StoryO  
http://www.junglesoftware.com/storyo/what_is_storyo.php  -­‐  accessed  23  
January  2012.  

StoryView  
http://www.write-­‐bros.com/  -­‐  accessed  23  January  2012.  

Trelby  
http://www.trelby.org  -­‐  accessed  March  29  2013.  

Stewart  McKie   -­‐  320  -­‐   As  of  15/08/2014  

Web  References  

85th  Academy  Awards.  

http://www.brandwatch.com/2013/02/introducing-­‐the-­‐social-­‐oscars-­‐tracking-­‐
award-­‐buzz/  
http://www.brandwatch.com/2013/02/the-­‐power-­‐of-­‐crowds-­‐predicting-­‐the-­‐
oscar-­‐winners/  -­‐  accessed  25  February  2013.  

Agile  manifesto.  http://agilemanifesto.org/principles.html  -­‐  accessed  05  

February  2014.  

Alexander,  Christopher.  
http://www.patternlanguage.com/leveltwo/caframe.htm?/leveltwo/../bios/desi
gnpatterns.htm  -­‐  accessed  15  January  15.  

Aman,  Siama  and  Szpakowicz  Stan.  Identifying  Expressions  of  Emotion  in  Text  in  
V.  Matousek  and  P.  Mautner  (Eds.):  TSD  2007,  LNAI  4629,  pp.  196–205,  2007.  
http://www-­‐scf.usc.edu/~saman/pubs/2007-­‐TSD-­‐paper.pdf  -­‐  accessed  12  May  
2014.  

Agarwal  A.,  Balasubramanian  S.,  Zheng  J.,  Dash  S.  Parsing  Screenplays  for  
Extracting  Social  Networks  from  Movies.  Proceedings  of  the  3rd  Workshop  on  
Computational  Linguistics  for  Literature  (CLfL)  @  EACL  2014,  pages  50–58,  
Gothenburg,  Sweden,  April  27,  2014.  
http://aclweb.org/anthology//W/W14/W14-­‐0907.pdf  -­‐  accessed  12  May  2014.  

Ayzenshtat,  Mark,  Elena  Filatova,  Kristina  Holst,  and  Vladislav  Shchogolev.  

EMuse:  Rapid  Screenplay  Prototyping  Language.  Dissertation.  University  of  
Columbia.  
http://www1.cs.columbia.edu/~sedwards/classes/2003/w4115/eMuse.final.pdf    
-­‐  accessed  07  February  2007.    

Barber,  Ian  (20  November  2009).  Part  of  Speech  Tagging.    

http://phpir.com/part-­‐of-­‐speech-­‐tagging/  -­‐  accessed  02  April  2012.  

Stewart  McKie   -­‐  321  -­‐   As  of  15/08/2014  

Beaudouin-­‐Lafon  and  Mackay.  
https://www.lri.fr/~mackay/pdffiles/Prototype.chapter.pdf  -­‐  accessed  11  March  
2014.  
 
Bio  printing  
http://www.explainingthefuture.com/bioprinting.html  -­‐  accessed  03  January  
2013.    

Blackstock  A.  and  Spitz  M.  Classifying  Movie  Scripts  by  Genre  with  a  MEMM  
Using  NLP-­‐Based  Features.  
http://nlp.stanford.edu/courses/cs224n/2008/reports/06.pdf  -­‐  accessed  12  May  
2014.  

Boese,  Steve.  http://steveboese.squarespace.com/storage/sample-­‐

screenplay-­‐page.gif      -­‐  accessed  7  June  2013.  

Box  Office  Mojo  (Avatar).  (http://boxofficemojo.com/news/?id=2667  -­‐  accessed  

30  August  2012.  

boyd,  d.  m.,  and  Ellison,  N.  B.  (2007).  ‘Social  network  sites:  Definition,  history,  
and  scholarship’  in  Journal  of  Computer-­‐Mediated  Communication,  13(1),  article  
11.  http://jcmc.indiana.edu/vol13/issue1/boyd.ellison.html  -­‐  accessed  09  July  
2013.  

Brill,  Eric  (1992).  A  Simple  Rule-­‐Based  Part  of  Speech  Tagger.  

http://acl.ldc.upenn.edu/H/H92/H92-­‐1022.pdf  -­‐  accessed  02  April  2012.  

Brown  Corpus.    
http://en.wikipedia.org/wiki/Brown_Corpus#Part-­‐of-­‐speech_tags_used  -­‐  
accessed  02  April  2012.  

Chun,  Russell  (Story  Visualizer).  

http://www.russellchun.com/storystructure/storyvisualizer.html  -­‐  accessed  15  
January  2013.  

Stewart  McKie   -­‐  322  -­‐   As  of  15/08/2014  

Clark,  Jeff.  www.neoformix.com  -­‐  accessed  11  June  2013.  
http://www.neoformix.com/Projects/TranscriptAnalyzer/index.html  -­‐  accessed  
06  July  2008.  

Clayton,  Sue  (2006).  Development  Lab-­‐The  Importance  of  Good  Peer  Review  in  
the  Context  of  Screenwriting.  
 http://www.adm.heacademy.ac.uk/resources/case-­‐studies/development-­‐lab-­‐
the-­‐importance-­‐of-­‐good-­‐peer-­‐review-­‐practice-­‐in-­‐the-­‐context-­‐of-­‐screenwriting/  -­‐  
accessed  10  December  2012.  

Coats,  Emma.  http://slacktory.com/2012/07/pixar-­‐story-­‐rules-­‐illustrated-­‐by-­‐

icanlegothat/  -­‐  accessed  03  September  2012.  

Couchbase.  http://www.couchbase.com/docs/couchbase-­‐devguide-­‐
2.0/modeling-­‐documents.html  -­‐  accessed  10  June  2013.  

Creative  Skillset.  http://www.skillset.org/film/jobs/script/  -­‐  accessed  04  

September  2012.  

Dai,  James  (Story  Charting).  http://www.storycharts.ca/pages/casablanca/  -­‐  

accessed  11  June  2013.  

Davenport,  G,  and  Murtaugh,  M  (1997).  ‘Automatist  Storyteller  Systems  and  the  
Shifting  Sands  of  Story’  in  IBM  Systems  Journal  36  (1997):    446-­‐456.  
http://www.research.ibm.com/journal/sj/363/davenport.html  -­‐  accessed  05  
February  2007.      

DiNucci,  Darcy  (1999).  ‘Fragmented  Future’  in  Print  Magazine.  (April  1999):  32,  
221-­‐22.  http://darcyd.com/fragmented_future.pdf  -­‐  accessed  15  December  
2012.  

Disney’s  Story.  http://story.us  -­‐  accessed  12  February  2014.  

Donath,  Judith  S.  (1995).  ‘The  Illustrated  Conversation’  in  Multimedia  Tools  and  
Applications  1  

Stewart  McKie   -­‐  323  -­‐   As  of  15/08/2014  

http://smg.media.mit.edu/papers/Donath/IllustratedConversation.pdf  -­‐  
accessed  07  February  2007.    

Eliashberg,  Josh.  ‘Revenge  of  the  Nerds,’  Part  V:  Can  Computer  Models  Help  
Select  Better  Movie  Scripts?.  Knowledge@Wharton,  November  29  2006.  
http://knowledge.wharton.upenn.edu/article/revenge-­‐of-­‐the-­‐nerds-­‐part-­‐v-­‐can-­‐
computer-­‐models-­‐help-­‐select-­‐better-­‐movie-­‐scripts/  -­‐  accessed  17  December  
2006.  

Eliashberg  Jehoshua,  Hui  Sam  K.,  Zhang  John  Z.  Green-­‐lighting  Movie  Scripts:  
Revenue  Forecasting  and  Risk  Management.  
http://forum.johnson.cornell.edu/workshop/MARKETING/Hui.pdf  -­‐  accessed  18  
March  2014.  
 

Ennegram.  http://www.9types.com/epd/1.php  -­‐  accessed  02  July  2013.    

Epstein,  Alex  (Complications  Ensue  blog)  

http://complicationsensue.blogspot.co.uk/2006/11/polling-­‐numbers.html  -­‐  
accessed  14  March  2014.  

FarsiteForecast.com.  http://www.farsiteforecast.com/vanity-­‐fair-­‐polls-­‐and-­‐
parties/  -­‐  accessed  18  March  2013.  

Fischer,  Gerhard.  http://l3d.cs.colorado.edu/~gerhard/papers/ccmcd2001.pdf  -­‐  

accessed  05  February  2014.  

Fountain  (John  August).  http://www.fountain.io  -­‐  accessed  29  March,  2013.  

Freeman,  Eric  T.  http://www.cs.yale.edu/homes/freeman/lifestreams.html  -­‐  

accessed  18  March  2014.  

Giles  and  Anderson.  http://clearcongressproject.com/charviz/  -­‐  accessed  11  

June  2013.  

Gloor,  Peter.  http://www.vanityfair.com/online/oscars/2013/02/mathematical-­‐

formula-­‐oscar-­‐winners-­‐imdb  -­‐  accessed  08  March  2012.  

Stewart  McKie   -­‐  324  -­‐   As  of  15/08/2014  

Gold  Light.  http://www.davekellyentertainment.com/marketsolution.html  -­‐  
accessed  11  August  2007.  

Google  Ngram  viewer.  http://books.google.com/ngrams/  -­‐  accessed  26  April  

2013.  

Grimes,  Seth  (2011).  12  Things  the  Semantic  Web  Should  Know  about  Content  
Analytics.  http://semanticnavigation.opentext.com/wp-­‐
content/uploads/2011/06/12-­‐Things.pdf  -­‐  accessed  18  December  2013.  

Hamlet  social  network:  

http://services.alphaworks.ibm.com/manyeyes/view/SJjqGFsOtha6ymEZWajPF2
-­‐  accessed  23  September  2008.  

http://www.jibble.org/shakespeare/images/hamlet.xml-­‐00000384.png  -­‐  
accessed  06  July  2008.  

Hanser  E.,  McKevitt  P.,  Lunney  T.,  and  Condell  J.  Scenemaker:  Intelligent  
Multimodal  Visualisation  of  Natural  Language  Scripts,  in  R.  Setchi  et  al.  (Eds.):  
KES  2010,  Part  IV,  LNAI  6279,  pp.  430–439,  2010.  
http://www2.denizyuret.com/bib/hanser/hanser2010scenemaker/hanser2010.
pdf  -­‐  accessed  12  May  2014.  

Gil  S.,  Kuenzel  L.  ans  Suen  C.  Extraction  and  Analysis  of  Character  Interaction  
Networks  From  Plays  and  Movies.    

http://www.stanford.edu/~cysuen/projects/GilKuenzelSuen-­‐
CharacterInteractionNetworks.pdf  -­‐  accessed  12  May  2014.  

 Henderson,  Peter  and  De  Silva,  Nishadi  (2006).  A  Narrative  Approach  to  
Collaborative  Writing:  A  Business  Process  Model  in  8th  International  Conference  
on  Enterprise  Information  Systems  (ICEIS),  Paphos,  Cyprus,  23  -­‐  27  May  2006.  
http://eprints.soton.ac.uk/261806/1/Camera-­‐ready_paper_499.pdf    -­‐  accessed  
12  May  2014.  

Stewart  McKie   -­‐  325  -­‐   As  of  15/08/2014  

Hollywood  Reporter  (Spec  Scripts).  
http://fencingwiththefog.blogspot.ca/2006/01/screenplay-­‐sales-­‐statistics.html    
-­‐  accessed  18  March  2014.  

Hornall  Anderson  touch  walls.  

 http://www.hornallanderson.com/casestudy/better-­‐living-­‐through-­‐technology  
-­‐  accessed  02  April  2012.  

Huntley,  Chris.  http://dramatica.com/articles/how-­‐and-­‐why-­‐dramatica-­‐is-­‐

different-­‐from-­‐six-­‐other-­‐story-­‐paradigms  -­‐  accessed  25  February  2013.  

Jackson,  Peter.  http://gointothestory.blcklst.com/2011/04/peter-­‐jackson-­‐on-­‐

screenwriting-­‐process.html    -­‐  accessed  04  September  2012.  

Jewell,  Michael  O.,  Lawrence,  K.  Faith,  Tuffield,  Mischa  M.,  Prugel-­‐Bennett,  
Adam,  Millard,  David  E.,  Nixon,  Mark  S.,  schraefel,  m.c.  and  Shadbolt,  Nigel  R.  
(2005)  ‘OntoMedia:  An  Ontology  for  the  Representation  of  Heterogeneous  
Media’  in  Multimedia  Information  Retrieval  Workshop,  ACM  SIGIR.  
http://image.ntua.gr/swamm2006/resources/paper12.pdf  -­‐  accessed  07  
February  2007.        

Jinhong,  Shen,  Seiya  Miyazaki,  Terumasa  Aoki,  and  Hiroshi  Yasuda.  Filmmaking  
Production  System  with  Rule-­‐Based  Reasoning.  Dissertation.  The  University  of  
Tokyo,  2003.  http://sprg.massey.ac.nz/ivcnz/Proceedings/IVCNZ_66.pdf  -­‐  
accessed  05  February  2007.          

Jinhong,  Shen,  Seiya  Miyazaki,  Terumasa  Aoki,    and  Hiroshi  Yasuda.  Filmmaking  
Production  System  with  Rule-­‐Based  Reasoning.  Image  and  Vision  Computing  
New  Zealand,  University  of  Tokyo.  
http://sprg.massey.ac.nz/ivcnz/Proceedings/IVCNZ_66.pdf  -­‐  accessed  07  
February  2007.  

Jurafsky,  Shriberg,  Fox  and  Curl.  http://acl.ldc.upenn.edu/W/W98/W98-­‐

0319.pdf  -­‐  accessed  29  March  2013.  

Stewart  McKie   -­‐  326  -­‐   As  of  15/08/2014  

Kao,  Albert  B.  and  Couzin,  Iain  D.  (2014).  Decision  accuracy  in  complex  
environments  is  often  maximized  by  small  group  sizes.  Proceedings  of  the  Royal  
Society  B  –  accessed  23  April  2014.  
http://rspb.royalsocietypublishing.org/content/281/1784/20133305.abstract  -­‐  
accessed  28  April  2014.  

Kendall  and  Kendall.  Slide  8-­‐3  in  http://www.mgt.ncu.edu.tw/~ylchen/sasd-­‐

slide/chap08.ppt    -­‐  accessed  11  March  2014.  

Kleinbard,  Lily.  Adapting  'Life  of  Pi'  for  the  Big  Screen  Took  170  Script  Revisions.  
http://www.theatlantic.com/entertainment/archive/2012/11/adapting-­‐life-­‐of-­‐
pi-­‐for-­‐the-­‐big-­‐screen-­‐took-­‐170-­‐script-­‐revisions/265367/  -­‐  accessed  18  March  
2014.  

Korsakow  system.  http://korsakow.org  -­‐  accessed  18  March  2014.  

Maioriello,  James.  
http://www.developer.com/design/article.php/1474561/What-­‐Are-­‐Design-­‐
Patterns-­‐and-­‐Do-­‐I-­‐Need-­‐Them.htm  -­‐  accessed  03  January  2013.  

Mairesse,  F  and  Walker  M.  PERSONAGE:  Personality  Generation  for  Dialog  in  
Proceedings  of  the  45th  Annual  Meeting  of  the  Association  of  Computational  
Linguistics,  pages  496–503,  Prague,  Czech  Republic,  June  2007.  
http://acl.ldc.upenn.edu/P/P07/P07-­‐1063.pdf  -­‐  accessed  12  May  2014.  

McKee,  Robert.  http://artscienceblog.blogspot.co.uk/2012/04/should-­‐good-­‐

brand-­‐stroy-­‐chart-­‐like-­‐good.html  -­‐  accessed  11  June  2013.  

McNiff,  Jean.  http://www.jeanmcniff.com/ar-­‐booklet.asp  -­‐  accessed  11  March  

2014.  

Media  Technology  (RCMT)  and  School  of  Creative  Media,  2004.  

http://www.ieeesmc.org/announcements/Newsletter/Dec2004/SMCS-­‐E-­‐
NewLetter-­‐Liu2004.pdf  -­‐  accessed  05  February  2007.  

Stewart  McKie   -­‐  327  -­‐   As  of  15/08/2014  

MediaWiki  Screenplay  Extension  http://siege.org/projects/mwspe  -­‐  accessed  05  
April  2006.  

Montelbano  Infographic.  http://cargocollective.com/federicafragapane/LA-­‐

LETTURA-­‐CORRIERE-­‐DELLA-­‐SERA  -­‐  accessed  09  July  2013.  

Moretti,  Franco  (2011).  Network  Theory,  Plot  Analysis.  Literary  Lab,  Pamphlet  2,  
May  1,  2011.  http://litlab.stanford.edu/?page_id=255  -­‐  accessed  31  December  
2013.  

Moubry,  Sean.  http://screenplayXML.org  -­‐  accessed  10  June  2013.  

Murtagh,  Fionn,  Ganz,  Adam  and  McKie,  Stewart  (2008).  ‘The  Structure  of  
Narrative:  the  Case  of  Film  Scripts’  in  Pattern  Recognition,  42  (2),  302-­‐312,  2009.  
http://arxiv.org/abs/0805.3799  -­‐  accessed  24  May  2008.      

Murtaugh,  Michael.  The  Automatist  Storytelling  System:  Putting  the  Editor's  

Knowledge  in  Software.  Dissertation.  MIT  Media  Lab.,  1996.  
http://xenia.media.mit.edu/~murtaugh/thesis/    -­‐  accessed  05  February  2007.          

Myers,  Scott.  http://www.gointothestory.com/2009/01/spec-­‐script-­‐sales-­‐

analysis-­‐2008-­‐big.html  -­‐  accessed  27  July  2009.  

Narrativization  -­‐  http://www.narrativization.com  -­‐  accessed  June  17,  2014.  

Nature,  vol.  453,  p.  708,  4  June  2008.  

http://www.nature.com/news/2008/080604/full/453708a.html  -­‐  accessed  June  
17,  2014.  

Neumann,  Petra,  Stefan  Schlechtweg,  and  Sheelagh  Carpendale  (2005).  

‘ArcTrees:  Visualizing  Relations  in  Hierarchical  Data’.  EUROGRAPHICS  -­‐  IEEE  
VGTC  Symposium  on  Visualization,  2005,  IEEE.  
http://pages.cpsc.ucalgary.ca/~pneumann/wiki/papers/Neumann_2005_ATV.p
df    -­‐  accessed  7  February  2007.  

Stewart  McKie   -­‐  328  -­‐   As  of  15/08/2014  

New  York  Times:    

Candidates  Debate.  
(http://www.nytimes.com/interactive/2007/12/15/us/politics/DEBATE.html  -­‐  
accessed  on  06  July2008.  

Democratic  Debate.  
http://www.nytimes.com/interactive/2008/04/16/us/politics/20080416_DEBAT
E_GRAPHIC.html#  -­‐  accessed  06  July  2008.  

Storylines.  http://www.nytimes.com/2005/04/24/magazine/24TV.html  -­‐  

accessed  06  July  2008.  

Ng,  Kal,  Marc  Aurel  Schnabel,  and  Thomas  Kvan.  Architectural  Animation  
becomes  Alive.  Communicating  Space(s)  [24th  eCAADe  Conference  Proceedings  
/  ISBN  0-­‐9541183-­‐5-­‐9]  Volos  (Greece)  6-­‐9  September  2006,  598-­‐603.  
http://wwwpeople.arch.usyd.edu.au/~marcaurel/publications/ecaadekal.pdf    -­‐  
accessed  07  February  2007.    

Nike  Touch  Wall  

http://www.flickr.com/photos/davidjkelley/sets/72157624420150769/  -­‐  
accessed  02  April  2012.      

Nitsche,  Michael,  and  Maureen  Thomas(2007).  ‘Stories  in  Space:  the  Concept  of  
the  Story  Map’  in  Visual  Storytelling  Second  International  Conference,  ICVS  2003,  
Toulouse,  France,  November  20-­‐21,  2003,  Proceedings  Series:  Lecture  Notes  in  
Computer  Science,  Vol.  2897  Balet,  Olivier;  Subsol,  Gérard;  Torguet,  Patrice  (Eds.)  

http://www.lcc.gatech.edu/~nitsche/download/Nitsche_ICVS_03.pdf  -­‐  accessed  

07  February  2007.  

The  Numbers.  http://www.the-­‐numbers.com  -­‐  accessed  27  July  2009.  

The  Object  Management  Group  (OMG).  http://www.omg.org  -­‐  accessed  12  

February  2014.  

Stewart  McKie   -­‐  329  -­‐   As  of  15/08/2014  

Optimizely  https://www.optimizely.com,  accessed  July  09,  2014.  

O’Reilly,  Tim  (2005).  What  is  Web  2.0:  Design  Patterns  and  Business  Models  for  
the  Next  Generation  of  Software.  O’Reilly  Media.  
http://oreilly.com/web2/archive/what-­‐is-­‐web-­‐20.html  -­‐  accessed  17  December  
2013.  

Paisa,  Sabi.  http://scriptsuper.wordpress.com/2013/04/09/directors-­‐

homework/about  -­‐  accessed  17  June  2013.  

Parham,  Jason.  http://www.complex.com/tech/2013/06/amazon-­‐storyteller-­‐

turns-­‐scripts-­‐into-­‐storyboards  -­‐  accessed  09  July  2013.  

Pate,  John.  A  LATEX2ε  class  for  Draft  Screenplay(2006).  

http://www.tug.org/texlive/Contents/live/texmf-­‐
dist/doc/latex/screenplay/screenplay.pdf  -­‐  accessed  23  January  2012.  

Protovis.  http://mbostock.github.io/protovis/ex/  -­‐  accessed  11  June  2013.  

Ramirez,  Alejandro.  Towards  a  Cinematically  Enhanced  Narrative  in  Proceedings  

of  the  2005  ACM  SIGCHI  International  Conference  on  Advances  in  Computer  
Entertainment  Technology,  2005,  ACM  Press.  
http://delivery.acm.org/10.1145/1180000/1178550/p365-­‐
ramirez.pdf?key1=1178550&key2=7686480711&coll=&dl=acm&CFID=15151515
&CFTOKEN=6184618  -­‐  accessed  07  February  2007.    

Reconstruction  Vol.8,  No.3  (2008).  

http://reconstruction.eserver.org/Issues/083/mckie.shtml  -­‐  accessed  June17,  
2014.      

Ruecker,  Stan.  
http://academia.edu/3155085/WATCH_MY_MOVES_FROM_DIGITAL_PLAYS_TO
_THE_DIGITAL_PLAYBOOK  -­‐  accessed  11  July  2013.  

Schmidt,  Benjamin  (Argo).  

Stewart  McKie   -­‐  330  -­‐   As  of  15/08/2014  

http://www.prochronism.com/2013/02/current-­‐events-­‐argo-­‐
anachronisms.html  -­‐  accessed  12  March  2013.  

Screenwriting  2.0  in  the  Classroom?  Teaching  the  Digital  Screenplay.  

http://framescinemajournal.com/article/screenwriting-­‐2-­‐0-­‐in-­‐the-­‐classroom/  -­‐  
accessed  17  Dec.  2013.  

ScriptReader.  http://scriptreaderonline.com  -­‐  accessed  03  September  2012.  

Sentiment  analysis:  

Go,  Bhayani  and  Huang.  

http://cs.stanford.edu/people/alecmgo/papers/TwitterDistantSupervision09.pd
f  -­‐  accessed  11  June  2013.  

Hennessey.  J.W.  (phpInsight  class).  

https://github.com/JWHennessey/phpInsight  -­‐  accessed  18  June  2013.  
 
Thet,  Na  and  Khoo.  
 http://jis.sagepub.com/content/36/6/823.full.pdf+html  -­‐  accessed  11  June  
2013.  
 
Yessenov  and  Misailovic.  
http://people.csail.mit.edu/kuat/courses/6.863/report.pdf  -­‐  accessed  11  June  
2013  

Scali,  Gabriele  and  Graham  Howard  (2004).  XML  Coding  of  Dramatic  Structure  
for  Visualization.  Dissertation.  System  Simulation  Ltd.,  2004.  
http://www.spacespa.it/studi_ricerche/pdf/XML_Coding.pdf    -­‐  accessed  01  
February  2007.  

Schilder  http://www.aclweb.org/anthology-­‐new/W/W98/W98-­‐0310.pdf  -­‐  

accessed  29  March  2013.  

Stewart  McKie   -­‐  331  -­‐   As  of  15/08/2014  

Segel,  Edward  and  Heer,  Jeffrey.  Narrative  Visualization:  Telling  Stories  with  
Data.  http://vis.stanford.edu/files/2010-­‐Narrative-­‐InfoVis.pdf  -­‐  accessed  18  
March  2014.  

Shakespeare:  

Macbeth.  http://www.shakespeare-­‐navigators.com/macbeth/Blood.html  -­‐  

accessed  04  March  2013.    

Open  Source  Shakespeare.org  (http://www.opensourceshakespeare.org/stats/  -­‐  

accessed  04  March  2013.    

DocuScope.  http://www.cmu.edu/hss/english/research/docuscope.html)  -­‐  

accessed  04  March  2013.  

Shakespeare’s  First  Folio  corpus  http://www.fastcompany.com/1800987/data-­‐

minings-­‐thing-­‐shakespeare-­‐takes-­‐center-­‐stage-­‐digital-­‐age  -­‐  accessed  04  March  
2013.  

Shooting  People  (online  community).  http://shootingpeople.org.  

SMAP  (2007)  http://dl.acm.org/citation.cfm?id=1339060  -­‐  accessed  June  17,  

2014.      

Soulier,  Eddie  and  Caussanel  (2002).  Narrative  tools  to  improve  Collaborative  
Sense-­‐Making.  AAAI  Technical  Report  WS-­‐02-­‐09.  
http://www.aaai.org/Papers/Workshops/2002/WS-­‐02-­‐09/WS02-­‐09-­‐002.pdf  -­‐  
accessed  7  February  2007.  

Spitball.  http://spitball.com/2009/04/how-­‐we-­‐collaborate-­‐on-­‐a-­‐screenplay  -­‐  

accessed  28  August  2012.  

Stapleton,  Christopher,  and  Dana  Mott.  Making  Memories:  Linear  Versus  

Interactive  Storytelling  Collaboration  in  Mixed  Reality  Space.    
http://www.rm.is.ritsumei.ac.jp/MR-­‐PreVizProject/pdf/stapleton.pdf  -­‐  accessed  

Stewart  McKie   -­‐  332  -­‐   As  of  15/08/2014  

07  February  2007.      

StoryVisualizer.  http://www.irit.fr/recherches/SAMOVA/ERCOLESSI/StoViz/#    -­‐

accessed  January  15  2013.  

Strapparava,  Carlo  and  Mihalcea,  Rada.  Learning  to  Identify  Emotions  in  Text  in  
SAC’08  March  16-­‐20,  2008,  Fortaleza,  Ceara,  Brazil.  
http://web.eecs.umich.edu/~mihalcea/papers/strapparava.acm08.pdf  -­‐  
accessed  12  May  2014.  

Tauberer,  Joshau  (2006).  What  is  RDF?  XML.com  

http://www.xml.com/pub/a/2001/01/24/rdf.html  -­‐  accessed  10  November  
2008.    

TechAmerica  Foundation.  Demystifying  Big  Data.    

http://www.techamericafoundation.org/bigdata  -­‐  accessed  07  June  2013.  

Tessman,  Kent.  http://www.kenttessman.com/2012/02/open-­‐screenplay-­‐

format  -­‐  accessed  28  August,  2012.  

Velikovsky,  J.  T.  CREATIVE  PRACTICE  THEORY:  The  Model,  

http://storyality.wordpress.com/creative-­‐practice-­‐theory/  -­‐  accessed  09  July  
2013.  

Walker  M.A.,  Grace  L.I.,  Sawyer  J.E.  An  Annotated  Corpus  of  Film  Dialog  for  
Learning  and  Characterizing  Character  Style.  Proceedings  of  LREC  2012.  
http://www.lrec-­‐conf.org/proceedings/lrec2012/pdf/1114_Paper.pdf  -­‐  accessed  
12  May  2014.  

Watching  the  Script.  http://digitalplaybook.humviz.org  -­‐  accessed  23  September  

2008.  

Watson,  Mark.  http://www.markwatson.com/opensource/  -­‐  accessed  02  April  

2012.  

Stewart  McKie   -­‐  333  -­‐   As  of  15/08/2014  

Wenger,  Etienne.  http://wenger-­‐trayner.com  -­‐  accessed  05  February  2014.  

Writers  Guild  of  America  Schedule  of  Minimums  (2004).  

http://www.wga.org/subpage_writersresources.aspx?id=68  -­‐  accessed  22  
December  2013.  

Wordseye.  
http://www.cs.columbia.edu/~coyne/images/wordseye_siggraph.pdf  -­‐  accessed  
07  February  2007.  

XML  formats  -­‐  http://screenplayxml.org/  http://openscrfmt.sourceforge.net/  

http://freecode.com/projects/screenplay-­‐xml  -­‐  accessed  23  January  2012.  

XBRL  example.  http://xbrl.squarespace.com/journal/2008/12/18/hello-­‐world-­‐

xbrl-­‐example.html  -­‐  accessed  6  June  2013.  

Zhi-­‐Qiang,  Liu,  and  Leung  Ka-­‐Ming  (2005).  ‘Script  Visualization  (ScriptViz):  a  

Smart  System  That  Makes  Writing  Fun  in  Soft  Computing  10  (2005),  34-­‐40.  
http://www.springerlink.com/content/m1811w1261160627/fulltext.pdf    -­‐  
accessed  07  February  2007.          

Zhuang,  Li,  Feng  Jing,  and  Xiao-­‐Yan  Zhu  (2006).  ‘Movie  Review  Mining  and  
Summarization’  (2006)  in  CIKM  '06  Proceedings  of  the  15th  ACM  international  
conference  on  Information  and  knowledge  management,  43-­‐50.  
http://www.csai.tsinghua.edu.cn/~zxy/paper/cikm249_zhuang.pdf  -­‐  accessed  
07  February  2007.  

Stewart  McKie   -­‐  334  -­‐   As  of  15/08/2014  

Appendix  A:  Hoffman’s  Hello  World  XBRL  Example  

This  is  a  very  simple  example  of  using  XBRL  as  provided  by  Charles  Hoffman  
(aka  the  ‘father’  of  XBRL)  for  learning  purposes  –  see  
http://xbrl.squarespace.com/journal/2008/12/18/hello-­‐world-­‐xbrl-­‐
example.html  -­‐  accessed  6  June  2013).  Note  that  all  data  content  is  enclosed  
within  <start>  and  </end  >  tags.HelloWorld  Instance  File  

The  initial  lines  of  HelloWorld.xml  provide  links  to  essential  reference  files  
managed  by  the  global  XBRL  authority  at  xbrl.org  

<xbrl xmlns="http://www.xbrl.org/2003/instance"
xmlns:xbrli="http://www.xbrl.org/2003/instance"
xmlns:link="http://www.xbrl.org/2003/linkbase"
xmlns:xlink="http://www.w3.org/1999/xlink"
xmlns:xsi="http://www.w3.org/2001/XMLSchema-
instance"
xmlns:iso4217="http://www.xbrl.org/2003/iso4217"

The  next  lines  refer  to  the  online  location  of  this  specific  instance  file  and  the  
location  of  the  XBRL  schema  this  instance  file  references  -­‐  HelloWorld.xsd.    
Typically,  many  instance  files  point  to  one  schema  file,  which  is  why  an  XBRL  
schema  can  be  considered  ‘mutually  agreed’.  

xmlns:HelloWorld="http://xbrl.squarespace.com/Hell
oWorld" xsi:schemaLocation=" ">
<link:schemaRef xlink:type="simple"
xlink:href="HelloWorld.xsd"/>

Then  some  contexts  are  defined  for  the  data  points  listed  in  the  file,  which  in  
this  case  include  a  reporting  year,  a  reporting  entity  and  a  reporting  period  
date:  

<!-- Contexts -->

Stewart  McKie   -­‐  335  -­‐   As  of  15/08/2014  

<context id="I-2007">
<entity>
<identifier
scheme="http://www.ExampleCompany.com">Example
Company</identifier>
<segment></segment>
</entity>
<period>
<instant>2007-12-31</instant>
</period>
</context>

In  this  instance,  there  are  two  contexts  representing  the  reporting  years  
2006  and  2007.  Then  the  reporting  units  are  defined  as  monetary  and  
denominated  in  US  dollars:  

<!-- Units -->

<unit id="U-Monetary">
<measure>iso4217:USD</measure>
</unit>

Now  that  the  overall  data  context  has  been  defined,  the  file  lists  a  series  of  
data  points  within  that  context  and  their  respective  attributes,  known  as  
‘facts’,  that  comprise  a  title,  a  year,  a  decimals  identifier  and  an  amount  (one  
fact  each  for  reporting  years  2006  and  2007):  

<!-- Fact values -->

<HelloWorld:Land contextRef="I-2007" unitRef="U-
Monetary" decimals="INF">5347000</HelloWorld:Land>
<HelloWorld:Land contextRef="I-2006" unitRef="U-
Monetary" decimals="INF">1147000</HelloWorld:Land>

A  number  of  facts  are  listed  in  the  file,  the  above  example  simply  reporting  
the  data  fact  for  ‘Land’  for  each  of  the  two  context  years:  2006,  2007.    

Stewart  McKie   -­‐  336  -­‐   As  of  15/08/2014  

The  HelloWorld.xsd  schema  also  references  some  key  files  like  the  instance  
file  but  mainly  functions  to  define  the  behavior  of  the  ‘elements’  that  the  
‘fact’  data  is  subject  to,  for  example  the  element  for  the  ‘Land’  fact:  

<element  
name="Land"  
type="xbrli:monetaryItemType"  
substitutionGroup="xbrli:item"  
xbrli:periodType="instant"  
 />  

Stewart  McKie   -­‐  337  -­‐   As  of  15/08/2014  

Appendix  B:  Selective  Scenepad  Feature  List  

Screenplay  as  Data  

-­‐  All  data  and  metadata  stored  in  the  MySQL  database  

-­‐  All  views  of  script  content  ‘assembled’  by  querying  multiple,  related  tables  

-­‐  Content  views  (PDFs)  at  script,  scene,  location  (action)  and  role  (dialog)  

-­‐  Data  granularity  to  snippet  (i.e.  dialog  or  action  text)  level  

-­‐  Additional  metadata  for  all  entities  (e.g.  role  gender,  location/scene  type,  
script  genre  etc.)    

Screenplay  Analytics  

-­‐  Word  frequency  clouds  derived  from  whole-­‐script,  dialog-­‐only  and  action-­‐
only  text  

-­‐  Word  frequency  clouds  based  on  parts  of  speech:  Nouns  and  verbs  

-­‐  Data  and  metadata-­‐based  charting  at  script,  scene,  location  and  role  levels  

-­‐  Timeline  visualization  of  scene  grouping  at  sequence  and  storyline  levels  

-­‐  ‘Dial’  visualization  of  script  scenes  by  scene  type  metadata  

-­‐  Character  thrulines  by  character  and  role  (e.g.  protagonist/antagonist)  

-­‐  Sentiment  analysis  of  snippet  content  at  script,  action  and  dialog  levels  

Screenplay  as  Social  Network  

-­‐  Multi  user  application  and  ability  to  create  script  groups  (i.e.  communities  
of  interest)  to  share  scripts  with  

Stewart  McKie   -­‐  338  -­‐   As  of  15/08/2014  

-­‐  Script  sharing  (e.g.  private,  group  and  public)  

-­‐  Role  access  rights  (i.e.  read-­‐only  and  read/write)  

-­‐  Content  hubs  at  script,  scene,  location  and  role  levels  for  users  to  
contribute  images  and  links  

-­‐  Scene  commenting  

-­‐  Access  to  a  ‘read-­‐only’  mobile  site  to  review  the  script  content  either  via  a  
web  link  or  a  QR  code  

-­‐  link  Scenepad  entities  (e.g.  role  or  location)  to  images  uploaded  from  a  
user’s  mobile  device  and  stored  on  a  third  party  website  (e.g.  Kee.ps.com  or  
Photobucket.com)  

-­‐  link  Scenepad  entities  (e.g.  script)  to  tweets  posted  to  Twitter.com  with  the  
appropriate  hashtag    

Screenplay  as  Design  

-­‐  Populate  new  blank  script  with  sequence  and  role  metadata  based  on  
Vogler’s  Hero’s  Journey  

-­‐  Group  scenes  into  structural  devices  such  as  sequences  and  storylines  

-­‐  Attach  images  at  script,  scene,  location  and  role  levels  to  help  to  visualize  
the  design  of  the  screenplay  

Business  of  Screenwriting  

-­‐  Attach  coverage  reports  to  the  script  

-­‐  Post  time  and  expense  to  a  script  development  project  

-­‐  Track  script  submissions  to  review  sites  and  competitions  

-­‐  Log  ownership  of  script  content  at  snippet  level  for  authorship  credits  
Stewart  McKie   -­‐  339  -­‐   As  of  15/08/2014  
Appendix  C:  Screenplay  Analytics  –  A  Use  Case  

Screenplay  analytics  of  a  datafied  script  enables  the  use  of  quantitative  
analysis,  via  machine-­‐reading  of  the  screenplay  content,  as  a  supplement  
and  complement  to  the  qualitative  analysis  that  can  be  gained  from  human-­‐
reading  of  the  script.  Using  my  ScriptFAQ  application  (accessible  at  
http://www.screenplayanalytics/scriptfaq/login.php)  I  have  imported  a  
Final  Draft  .fdx  file  in  order  to  ‘datafy’  the  screenplay  content,  by  which  I  
mean  parsing  the  screenplay  text  in  the  file  and  ‘shredding’  it  to  store  in  a  
MySQL  database.    

Here  I  present  a  simple  use  case  for  how  screenplay  

analysts/authors/readers  can  use  screenplay  analytics  to  help  them  
understand  the  likelihood,  not  the  certainty,  that  a  script  contains  a  strong  
protagonist  by  using  basic  screenplay  analytics  and  by  supplementing  the  
screenplay  data  with  additional  metadata  added  by  the  reader/analyst.  This  
analysis  also  shows  how  the  addition  of  metadata  (data  about  the  screenplay  
data)  by  the  analyst/author/reader  can  help  to  drive  more  analysis  that  may  
not  be  easy  to  get  from  the  screenplay  data  alone.  

I  used  ScriptFAQ  to  import  and  analyze  a  screenplay  called  Reconciliation.  

This  is  not  a  script  that  any  reader  will  be  familiar  with  as  I  wrote  it  and  it  is  
unpublished.  But  this  unfamiliarity  ensures  that  readers  can  follow  the  
analysis  below  without  any  pre  conceptions  that  may  be  present  if  a  well-­‐
known  movie  script  were  used  instead.    

As  soon  as  the  script  is  imported  we  can  go  to  the  script  ‘home’  page  and  see  
that  some  basic  statistics  about  the  screenplay  content,  namely:  

-­‐  It  has  71  Scenes  involving  17  roles  across  30  locations  
-­‐  It  has  1369  Snippets  comprising  510  dialog  and  342  action  snippets  

Stewart  McKie   -­‐  340  -­‐   As  of  15/08/2014  

  

Figure  C1  –  Script  Home  Statistics  

If  we  go  to  the  role  list  we  can  see  a  list  of  roles  identified  from  the  script  
import.  Notice  some  metadata  –  e.g.  gender,  type  and  logline  –  that  would  
not  have  come  from  the  script  import  but  was  added  by  the  author  later.  

Stewart  McKie   -­‐  341  -­‐   As  of  15/08/2014  

Figure  C2  –  Script  Roles  

If  we  click  the  ‘Home’  icon  next  to  the  role  called  ‘Maggie’  we  can  already  see  
some  basic  statistics  about  this  role  that  may  be  interesting  such  as:    

-­‐  The  number  of  scenes  and  snippets  the  role  is  linked  to  
-­‐  The  role  rankings  vs.  other  characters  in  the  script  

Figure  C3  –  Role  Home  Statistics  

So  of  our  17  roles  who  looks  most  likely  to  be  the  protagonist?  We  go  to  the  
Role  FAQs  page  to  investigate.  

Stewart  McKie   -­‐  342  -­‐   As  of  15/08/2014  

  

Figure  C4  –  Role  FAQs  

Here  we  see  there  are  9  Role  FAQs  to  choose  from  –  clicking  the  button  on  
the  right  of  the  question  displays  an  answer.  

1.  On-­‐Screen  Time  

One  indicator  of  a  strong  protagonist  is  likely  to  be  a  character  that  benefits  
from  significant  ‘on-­‐screen’  time  -­‐  in  relation  to  other  characters.    On  screen  
time  is  can  be  determined  by  at  least  three  factors:  

-­‐  The  number  of  scenes  the  character  appears  in  

-­‐  How  much  dialog  they  speak  
-­‐  The  ‘longevity’  of  the  character  within  the  screenplay  narrative  

In  the  Role  FAQ  section,  ScriptFAQ  presents  3  simple  charts  to  help  
understand  on-­‐screen  time  as  shown  in  figures  5  and  6  below.    

Stewart  McKie   -­‐  343  -­‐   As  of  15/08/2014  

  

Figure  C5  –  Top  5  Roles  by  Scene  (Note:  each  dot  represents  a  unique  scene)  

In  figure  5  we  can  see  that  the  role  ‘MAGGIE’  has  a  speaking  part  in  a  
significant  number  of  scenes  (actually  35  out  of  71).  The  next  most  
significant  character  by  this  analysis  is  ‘DOREEN’  who  only  speaks  in  15  
scenes.  

In  figures  6  &  7  we  can  see  that  of  the  five  major  characters,  Maggie  enters  
the  script  first  (in  scene  4)  and  exits  the  script  joint-­‐last,  in  scene  69.    So  she  
enters  the  screenplay  early  and  leaves  late  –  both  of  which  are  likely  to  be  
good  indicators  of  a  protagonist  role.    

Stewart  McKie   -­‐  344  -­‐   As  of  15/08/2014  

  
Figure  C6  –  Top  5  Roles  by  Entry  Scene  (Note:  each  dot  represents  a  unique  scene)  

Stewart  McKie   -­‐  345  -­‐   As  of  15/08/2014  

Figure  C7  –  Top  5  Roles  by  Exit  Scene  (Note:  each  dot  represents  a  unique  scene)  

If  you  did  not  know  that  Maggie  is  the  protagonist  then  you  might  already  
guess  that  she  could  be.    

2.  Dialog  and  Action  

It’s  hard  to  imagine  that  a  strong  protagonist  would  not  say  and  do  a  lot  in  a  
script.  On  screen  time  is  unlikely  to  count  for  much  if  the  character  is  mute  
and  rooted  to  the  spot  as  audiences  are  likely  to  become  bored  quite  quickly  
with  such  a  character.    So  in  the  Role  FAQ  section,  as  shown  in  figures  8  and  
9  below,  ScriptFAQ  provides  2  more  analyses  that  count  how  many  dialog  
snippets  a  character  speaks  and  an  indication  of  how  much  action  they  
participate  in.    In  figure  8  we  see  that  Maggie  speaks  172  dialog  snippets  (of  
510)  and  the  next  most  vocal  character  appears  to  be  Doreen  (with  81  dialog  
snippets).  

Stewart  McKie   -­‐  346  -­‐   As  of  15/08/2014  

Figure  C8  –  Top  5  Roles  by  Dialog  Snippets  (Note:  each  dot  represents  a  unique  dialog  snippet)  

In  figure  9  we  see  that  Maggie’s  role  cue  is  mentioned  in  163  action  snippets  
and  the  next  most  ‘active’  character  appears  to  be  the  Sergeant  (mentioned  
in  47  action  snippets).  We  know  from  figure  7  that  the  script  as  a  whole  
contains  342  Action  snippets,  so  Maggie  is  involved  in  almost  half  of  them.  

Figure  C9  –  Top  5  Role  Cues  mentioned  in  Action  Snippets    

Again,  Maggie  continues  to  exhibit  ‘strong  protagonist’  characteristics.  

3.  What  The  Protagonist  Says  and  Does  

Currently,  ScriptFAQ  does  not  provide  any  sophisticated  textual  analysis  of  
the  content  of  a  character’s  dialog  and  action  snippets  except  for  some  basic  
part-­‐of-­‐speech  analysis,  of  nouns  and  verbs  in  particular,  that  can  provide  
some  clues  as  to  whether  a  character  refers  to  potentially  ‘interesting’  things  
or  does/takes  part  in  potentially  ‘interesting’  action.  In  fact,  figures  10  and  
11  do  not  tell  us  that  much  but  might  if  more  sophisticated  text  analytics  
was  available  to  apply  to  the  script  content.  

Stewart  McKie   -­‐  347  -­‐   As  of  15/08/2014  

  

Stewart  McKie   -­‐  348  -­‐   As  of  15/08/2014  

Figure  C10  -­‐  Most  Frequent  Nouns/Words  in  Dialog  Snippets  Spoken  by  Role  

Figure  C11  -­‐  Most  Frequent  Nouns/Words  in  Action  Snippets  of  Role  

4.  Protagonist  &  Antagonist  Role  Metadata  

Let’s  make  the  assumption  that  Maggie  is  the  Protagonist.  As  the  writer  I  
know  this  already  and  so  I  would  be  expecting  these  kinds  of  results  from  
my  quantitative  analysis  of  my  script.    If  Maggie  had  not  turned  in  these  
kinds  of  results  I  might  have  wondered  why  and  whether  in  fact  Maggie  
would  be  perceived  by  others  (e.g.  a  reader/analyst  or  the  potential  
audience  of  my  movie)  as  the  protagonist  of  this  script.    

So  if  I  had  not  done  so  already,  now  would  be  the  time  to  ‘flag’  Maggie  as  the  
protagonist  by  adding  this  metadata  to  her  role  record  and  while  I’m  at  it  I  
might  add  other  metadata  (e.g.  her  gender)  as  in  figure  12  and  probably  do  

Stewart  McKie   -­‐  349  -­‐   As  of  15/08/2014  

the  same  job  for  the  antagonist  role.    

Figure  C12  –  Basic  Role  Metadata  

Once  you  have  identified  your  Protagonist  and/or  Antagonist  in  your  script  
you  can  use  another  set  of  Key  Role  FAQs  to  analyze  their  relative  dynamic.  

 
Stewart  McKie   -­‐  350  -­‐   As  of  15/08/2014  
Figure  C13  –  Key  Role  FAQs  

5.  Protagonist  Social  Network  

A  strong  protagonist  is  likely  to  have  a  rich  set  of  social  relationships  in  the  
script  -­‐  this  role  in  particular  is  unlikely  to  be  ‘isolated’  from  the  rest  of  the  
cast.  In  other  words  the  protagonist  generally  participates  in  an  extensive  
social  network.  ScriptFAQ  provides  various  analyses  of  role  interactions  as  
shown  in  figure  14  below.  Here  we  see  that  Maggie  (the  ‘central’  role  and  
identified  by  her  ‘protagonist’  metadata)  has  dialog  relationships  with  10  
other  characters.  Maggie  is  at  the  centre  of  a  social  network  ‘hub’  where  the  
thickness  of  each  connecting  ‘spoke’  line  is  indicative  of  the  number  of  
relationships  she  has  with  other  characters  in  terms  of  the  number  of  scenes  
in  which  both  characters  have  a  speaking  role.    

Stewart  McKie   -­‐  351  -­‐   As  of  15/08/2014  

  
Figure  C14  –  Maggie  as  Social  Network  Hub  

6.  Protagonist  Sentiment  

ScriptFAQ  does  not  use  a  particularly  sophisticated  sentiment  analysis  

algorithm  but  it  does  provide  some  basic  sentiment  analysis  to  categorize  
role  snippets  as  positive,  neutral  or  negative.    Conventionally,  we  might  
expect  that  our  protagonist  is  generally  more  positive  than  negative  and  
perhaps  the  reverse  for  our  antagonist.  

Figure  C15  –  Maggie’s  Sentiment  Analysis  

In  figure  15  above  we  can  see  that  Maggie’s  dialog  is  more  positive  than  
negative.  

Stewart  McKie   -­‐  352  -­‐   As  of  15/08/2014  

7.  Protagonist  vs.  Antagonist  

Some  might  argue  that  a  Protagonist’s  strength  can  only  be  determined  in  
relation  to  the  script’s  Antagonist.  This  kind  of  analysis  requires  both  the  
protagonist  and  antagonist  character(s)  to  be  identified  by  role  metadata  
and  that  both  can  be  mapped  to  a  specific  character.  It  may  be  that  the  
protagonist  and  antagonist  roles  are  in  fact  represented  by  more  than  one  
character.  In  this  script,  there  is  one  character  identified  as  the  protagonist  
but  two  characters,  loosely  working  in  tandem,  identified  by  metadata  as  the  
antagonist  or  antagonistic  ‘force’.  

Figure  16  shows  there  is  a  roughly  70/30  ‘balance’  in  this  script  between  
scenes  that  the  protagonist  role(s)  speaks  in  (35)  and  the  antagonist  role(s)  
speak  in  (15).    

Figure  C16  -­‐  Script  Balance:  Protagonist  vs.  Antagonist  Dialog  Scenes  

Figure  17  shows  who  speaks  more  dialog  snippets  -­‐  protagonist  or  
antagonist?  In  fact  it’s  172  vs.  81  again  about  a  70/30  balance.    
 

Stewart  McKie   -­‐  353  -­‐   As  of  15/08/2014  

  

Figure  C17  -­‐  Script  Balance:  Protagonist  vs.  Antagonist  Dialog  Snippets  

Finally,  figure  18  shows  the  balance  of  scenes  that  the  protagonist  speaks  in  
in  vs.  scenes  the  rest  of  the  cast  speaks  in  but  the  protagonist  does  not  
(49/35).    

Stewart  McKie   -­‐  354  -­‐   As  of  15/08/2014  

Figure  C18  -­‐  Script  Balance:  Protagonist  vs.  Other  Cast  Scenes  

We  can  also  get  some  idea  of  the  relative  weight  of  the  protagonist  vs.  the  
antagonist  by  comparing  their  social  networks.  Figure  19  shows  the  network  
for  antagonist  Doreen  (vs.  Maggie’s  network  in  figure  14  above).  

 
 

Figure  C19  –  Doreen  (Antagonist)  as  Social  Network  Hub  

Clearly,  on  the  basis  of  these  comparisons,  Maggie  remains  a  contender  for  
classification  as  a  strong  protagonist.  

Stewart  McKie   -­‐  355  -­‐   As  of  15/08/2014  

Conclusion  

Screenplay  role  analytics  clearly  has  a  long  way  to  go,  both  in  theory  and  in  
practice,  but  even  these  basic  analyses  show  that  you  can  learn  something  
about  the  relative  ‘strength’  of  a  protagonist  even  before  you  read  the  script.  
Assuming  the  screenplay  is  ‘datafied’,  additional  metadata  must  be  added  to  
the  role  record  (in  the  database)  to  facilitate  many  of  these  analyses  but  this  
is  a  relatively  trivial  task  for  the  script  author  to  do.  The  figures  above  
illustrate  just  some  of  the  role-­‐based  analytics.  ScriptFAQ  also  includes  many  
more  charts  that  analyze  more  about  the  script,  its  scenes  and  locations.  

For  example,  Reconciliation  looks  like  a  good  script  for  strong  female  roles  
since,  as  figure  20  shows,  female  roles  speak  slightly  more  dialog  than  male  
roles  (264/246  snippets).  In  fact  the  script  does  have  a  very  strong  female  
protagonist  so  this  is  not  surprising  to  me  (the  author).  

   

Figure  C20  –  Dialog  Gender  Balance  

Of  course  it  is  your  qualitative  analysis  from  reading  the  script  that  will  help  
to  confirm  or  deny  the  validity  of  these  quantitative  analyses.  This  kind  of  

Stewart  McKie   -­‐  356  -­‐   As  of  15/08/2014  

analytics  correlates  data;  it  does  not  indicate  causality.  A  protagonist  
character  who  dominates  the  screen,  relates  to  everyone  and  overpowers  
their  antagonist  with  the  weight  of  their  dialog  and  action  may  still  be  flat  
and  boring  when  it  is  clear,  for  example,  exactly  what  they  say  and  do  and  
how  they  relate  to  others  by  reading  the  script.  In  fact  their  quantitative  
‘weight’  may  work  against  them  when  compared  to  your  qualitative  
assessment.    

I  hope  this  simple  analysis  has  shown  that  automated  quantitative  

screenplay  analytics,  based  on  a  datafied  script,  is  a  useful  and  interesting  
companion  to  traditional  qualitative  analysis  via  script  reading.  

Stewart  McKie   -­‐  357  -­‐   As  of  15/08/2014