April 2009

Roots and Categories

Hagit Borer
USC

1. Introduction
1.1. A general outline of the system
1. There is only one computational component, call it Syntax, which is responsible for hierarchy forming
and constituent manipulation operations in the grammar. These include such operations when they
are associated with word-internal structure (and specifically, for our purposes, operations which create
complex word derivatives such as trans-form-ation-al or kitchen towel rack (i.e., operations of so called
derivational morphology).
2. A substantive Word, as conventionally conceived, is a syntactic constituent which (happens) to
correspond to a phonological unit of a given size (e.g., for the assignment of primary stress). While it is
likely that there are some universal constraints on what syntactic constituents can correspond to such
phonological units, beyond that, the mapping is language specific, and syntactic constituents of equal
complexity may or may not be phonological-stress units. Crucially, then, substantive Words are not
syntactic primitives or atomic in any meaningful sense.
3. Roots (listemes, in Borer (2005a,b) do not have any internal grammatical structure, and are devoid of
any syntactic value:
a. No category
b. No morphological marking
c. No argument structure (either external or internal).
4. Roots need not correspond to well-formed phonological words.
5. Grammatical Functors (both FL and FF ) are members of the Functor Lexicon (=vocabulary items in
DM).
a. FF : a grammatical functor that projects and licenses a functional node: theD, three#, willT, <pst>T,
<pl>DIV etc.)
b. FLs: a grammatical functor that projects a lexical node: -alA, -ationN, -izeV, -lyAdv, onP
6. English derivational suffixes, instances of FL, merge syntactically with their semantic (functorial) and
syntactic properties. At times, such a bundle of syntactic and functorial properties defines a unique
affix in a language, e.g., English –{i/a}ble, with the result that whether or not it is inserted with a
phonological index or not is immaterial. At other times, such a bundle defines a set of morphemic
allomorphs (e.g. deverbal nominals of the {–ation class}: -ation/-ion, -ment, -ance/ence, -al).
7. Extended Projections must have an L core (definitionally).
8. Roots merge as phonological indices. Phonological indices are exactly specific enough to ensure
phonological faithfulness in the syntactic derivation, thereby excluding the derivation of show from
see or an event-denoting noun such as lesson from some abstract non-existing verbal entry. A
complete phonological matrix for a root is inserted on the basis of the index in the syntactic context
created by the derivation.
9. Roots merge with functors (FL or FF ), and are categorized contextually by their merger environment.
Consequently, bare roots, without a category label, are not a syntactic option.
 Roots and Categories

10. English does not have (productive) -FLs (zero n, v, a etc. as categorizers1).

11. Roots merge without meaning.

b. All non-compositional substantive meaning (henceforth Meaning), including that of roots, is
assigned by a single component, henceforth the encyclopedia.
c. The encyclopedia searches (en-searches) post-syntactic bracketed phonological representations
and returns Meaning.
d. En-searches are cyclic and local: FF brackets delimit en-searching (caveat: DIV may be,
simultaneously, an instance of both FL and FF, as may some aspectual heads).
1.2. Some entailments and consequences
12. a. If root meaning is only available on the basis of phonological representations, it follows that it is
syntactically inactive (including in Logical Form). It further follows that roots cannot have internal
arguments (i.e. (‎3c)follows from (‎11)).
b. Silent copies are not assigned encyclopedic meaning. It therefore follows that for any one
given derivation, non-compositional derivatives need not be ambiguous.
c. Identical syntactic derivations with distinct, multiple phonological spellout possibilities for FL
vocabulary items may result in Meaning differences.
d. All grammatically active material must be encoded either structurally or through the properties
of functional vocabulary
e. Non-local non-compositional Meaning assignment (e.g. phrasal idioms) requires multiple, possibly
templatically-linked en-searches.
1.3. In this workshop:
13. What I will mostly talk about:
a. Arguments for post-syntactic non-compositional substantive meaning assignment
b. Arguments for the sensitivity of non-compositional meaning to phonological representation
c. Arguments that en-searches are stopped by FL brackets.
14. What else is in this Handout:
a. Evidence that Roots must be inserted with a phonological index.
b. Arguments for contextual categorization and against zero FL

2. One simple argument for (some) early root phonology (cf. Borer 2003)
15. a. EAT ]  /eat/
b. [V CAUSE EAT ]  /feed/
16. AS-nominals (Grimshaw’s 1990 Complex Event Nominals) are nominalizations of a verbal structure
(Hazout, 1991, 1995; Borer 1993, 1999, 2003; Engelhardt, 2000; Alexiadou, 2007; Harley 2006; Sichel
2007 among others, differing executions immaterial here.)
17. There are no AS-nominals in English (Hebrew, Greek) whose head is not morphologically
(transparently) derived from a verb (or adjective).2

1
To be distinguished from, e.g. v as an event-structure node, which the present claim is silent about.
2
With English aggression being an exception that clearly proves the rule.

Borer 2/23
 Roots and Categories

18. But if roots do not contain at least some phonological information, AS-nominals without a verbal
source cannot be excluded.
19. a. The lesson lasted several hours
b. The lesson took place from 7am to 8am
20. a. *the lesson of geometry by an incompetent teacher
b. *the lesson of geometry for several hours
c. *the lesson of geometry in order to understand the most recent proof
21. a. ha-šinui šel merkaz ha-'ir 'al yedey ha-'iriya
the-transformation/change of center the-city by the-city administration
b. ha-šinuy haya madhim
the-change/transformation was amazing
22. a. *ha-transformacia šel merkaz ha-'ir 'al yedey ha-'iriya
the-transformation of center the-city by the-city administration
b. ha-transformacia hayta madhima
the-transformation was amazing
23. All and only phonology that impacts the derivation (see Acquaviva 2008 for a relevant discussion).3

3. Constraints on non-compositional meaning assignment

3.1. Some general considerations:
24. a. The conceptual issue –
i. Restricting the search
ii. Avoiding a redundant search
b. The empirical issue: is there a demonstrable en-searchable domain?
25. a. A domain too big: idiomatic meaning is sometimes discontinuous and may involve multi-clausal
structure:

SUBJ4 (aux) cross(TENSE) DEM (A) bridge3(PL) when pron4 (ADV) come(TENSE) to pron3

Unless the domain of en-searches is delimited, even if [[AB]C]D] is compositional, it cannot

dismissed from short term memory because [[[[[AB]C]D]..]..]Z] might have a non-compositional
meaning.
b. A domain very small: AS-nominals are always compositional (Alexiadou, 2007, based on Marantz
2001).
the field’s transformation
26. a. *the linguist's transformation of the structure (but try the field)
(cannot mean the linguist having performed a grammatical transformation on the structure)
b. *The patient's transference of his feelings (but try his files)
3.2. The Marantz-Arad approach
27. “Locality constraints on the interpretation of roots: roots are assigned an interpretation in the
environment of the first category-assigning head with which they are merged. Once this interpretation

3
The logical conclusion is that suppletive pairs such as go/went constitute two, rather than one, roots with
phonological gaps.

Borer 3/23
 Roots and Categories

is assigned, it is carried along throughout the derivation” (Arad 2003, based on Marantz 2000, 2001).

note conceptual affinity with Lexical Phonology and Morphology where meaning changes are
restricted to level I affixation, which, by definition, is 'inside' level II affixation which cannot change
meaning).
28. a. A root inventory with basic meaning
b. A mechanism allowing the assignment of non-compositional meaning in a restricted syntactic
environment (for Arad, op. cit: at categorizing, possibly defining a phase).
29. Pairs such as transmission (the act of transmitting) vs. transmission (car gear), have distinct structures
(cf. Embick and Marantz 2008):
a. [[[transmit] v]-tionn] b. [[transmit] -tionn]
3.2.1. Some general queries:
30. What is the status, in this system, of the 'basic meaning of roots'?
a. If they do have a basic meaning, is it reanalyzed when the root+category combination is assigned a
distinct meaning (e.g., transmission)?
b. If they don't have a basic meaning altogether, what does it mean to claim that roots take internal
arguments?
31. Whenever FL attaches to an uncategorized root, it only does so providing in some (parallel) derivation
that root is categorized as a specific category type, and with a zero FL. (e.g., whenever –tion attaches
to a root to give rise to a 'non-compositional' meaning, it can also attach to that same root, providing it
has been verbalized with a -FL). That is a rather curious coincidence.
3.2.2. Empirical problems
32. Arad’s (2003) evidence: Hebrew verbs derived from roots may have idiosyncratic meanings.
Denominal verbs have predictable meanings.
33. a. Root: X.Š.B (roughly cognitive process)  [N xešbon ]  [V hitxašben]
'arithmetic' 'settle accounts'
'account' ‘retaliate’
'calculus' (Arad, op cit), .
 But Hebrew xešbon ‘account’, does not have the ‘settle account/retaliate’ reading (e.g. with
any light verb)! This may be literally translated from the English idiom, rather than derived from
the Hebrew noun.4
b. Underived noun: [N bayit]  [V biyet]
5
'house'  'domesticate'
 but also: [V hitbayet]
'focus in on something'
Again may be a literal translation from English ‘home in on’, but the noun bayit ‘house’ is
impossible in that meaning context (e.g., no 'homing pigeons'-type of expression).
34. [[Vtrans[form]]-ation]
[N [N[Vre[act]] tion] ary]]
[N[A [glob] al] ize ]-ation]

4
ose (li/iti) xešbon(ot), lit. ‘makes (to me/with me) account’ means ‘keep record of wrongdoings’, not 'retaliate'
5
See Bat El (1994) inter alia for extensive argumentation that bayit ‘house’ is not derived from a root.

Borer 4/23
 Roots and Categories

[N[V[A[Nnatur]al]iz]ation] (H. Harley, p.c.)

[N [V [ organ] ize] ation]
[N[A[act] ive] ism/ist]
[N[N[Vprotect]ion]ism]
[N[A[A[Vexist]ent]ial]ism]
[A[A[Vdiffer]ent]ial] (math)
[N[A[Nfest]iv]al]
[A[N[vexcept]ion]al] (excellent), [un[A[N[Vexcept]ion]al]] (ordinary)
[un-[A[(N)Americ]an]]
3.3. An Alternative
35. One list, one access point
a. No initial assignment of root meaning
b. No meaning adjustments at any syntactic intermediate point
c. An encyclopedia that matches Meaning with Sound (=phonological representations) in a restricted
domain (see also Marantz, 1996). Within that domain, the arbitrariness of meaning assigned to,
e.g., form or pie is no different from that assigned to reactionary or to last judgment and fellow
traveler.
36. What domain? A clue from derived nominals:
Both R-nominals and AS-nominals are derived from verbs (Borer, 1999, 2003; Alexiadou, 2007) through
incorporation. In both cases, the incorporation creates string adjacency between V and its
nominalizer:

a. R-nominals: [N [verb] N [L=V verb ]

b. AS-nominals: [N [verb] N [FF1.. [FF2… [L=V verb ]]]
37. a. [[ formL=V ] –ationN] a possible single phonological word; possibly non-compositional
b. [[[[formL=V ]FF2]FF1] –ationN] a possible single phonological word; must be compositional
38. A locality constraint on en-searches:
a. En-searches match Meaning with sound cyclically, using L as an anchor and operating on bracketed
strings.
b. FF stops en-searches (differently put, encyclopedic entries may not contain FF. )
39. Some consequences and elucidations:
a. Reanalysis of root meaning is never necessary
b. As silent copies are not en-searchable; if an en-searchable domain was formed by movement (e.g.
incorporation), there is no reconstruction and compositional meaning is not formed alongside a
compositional one for any single derivation.
40. [1ACT] .
[ RE [ACT ] ] [2 REACT]
[ [ RE [ACT ] ] TION ] [3 REACTION]
[ [ [ RE [ACT ] ] TION ] ARY ]; [4 REACTIONARY]

(Single) en-searchable domains: [1ACT] , [2 REACT], [3 REACTION], [4 REACTIONARY]

Compositional readings: all single en-searchable domains plus the inherent meaning of any FL within a
given cycle.

Borer 5/23
 Roots and Categories

41. *The linguist’s transformation of the structure

[1FORM]
[TRANS [FORM] ] [2TRANSFORM]
[[TRANS [FORM] ] FF1] [3 ] (??)
[[[TRANS [FORM] ] FF1] FF2] *[4 ]
[[[[TRANS [FORM]] FF1] FF2] ATIONN] *[5 ]
42. A successful grammatical transformation

-ationN  -ationN
2 2
-ationN [L=V transform] -ationN [L=V transform]
2
(unordered merge) [L=V transform] -ationN
(English adjoins to the left; by Uniformity what moves cannot project;
suffixes are listed phonologically as such)

[1FORM]
[TRANS [ FORM] ] [2 TRANSFORM]
[[TRANS [ FORM] ] ATION] [3 TRANSFORMATION](AL-IZE-ATION)
43. a. Only non-compositional constructs in Hebrew are systematically consistent with the grammatical
properties of incorporated structures
b. Only non-compositional constructs can head another construct (i.e., create a left branching
structure).
Non-compositional Meaning requires incorporation thereby creating an en-searchable domain (cf.
Borer, 2008)
44. [beyt-sefer] sade; [begged-yam] meši; [yošev-roš] mo'aca [yom huledet] šlošim6
house book field suit sea silk sitter head council day birth thirty
'field school' 'silk bathing suit' 'council chairman' ’30th birthday’
45. *[beyt ‘ec] sade; *[xulcat (ha-)mora] meši;
house wood field blouse (the-)teacher silk
(field wooden house) (silken teacher’s blouse)
46. A very simplified structures for construct and compounds in Hebrew:
a. [DP [#P beyt [FF [DP mora] beyt [DivP beyt [NP beyt ] ‘a teacher’s house’
house teacher

b. [DP [#P beyt [FF [NP 'ec] beyt [DivP beyt [NP beyt ] ‘a wood house
house wood
47. [DP [#P [Nbeyt [N sefer]] [FF [NP sefer] [DivP beyt [NP beyt ] (right adjunction) ‘school’
house book book

6
Cf. šnat alpayim 'year 2000', where the feminine form šana 'year' takes a construct form, for evidence that
numerical of this sort require a construct formation.

Borer 6/23
 Roots and Categories

48. And note, as well: begged, 'suit' 'clothing item' is already categorized (as N, by the vowel combination);
yošev, a participle, ‘sitter’, is already categorized. Yet they give rise, in compounding, to non-
compositional meaning, in turn associated with a single N dominating constituent.
3.4. Evidence or phonological representations feeding the encyclopedia
49. a. The selective transmission of historical documents
b. scanning and transmittal of documents or parts of documents (US Patent office publication)
c. A camera system for processing documents for measurement of reflectance and/or
transmittance of documents
50. a. Deferment of student loans
b. The department shall grant a deferral of interest and principal payments
51. a. Several groups … monitor the sale and transportation of seed
b. "The transportal of seeds in the wool or fur of quadrupeds." –a C. Darwin article
52. a. the slight transference of red pigments from the skins
b. "Transferral of bread “sponge” from dough mixer to trough prior to fermentation"
53. a. public transportation
b. *public transportal
54. a. the car's transmission
b. *the car's transmittal
c. *the car's transmittance
55. a. "Understanding transference and counter transference"
b. *Understanding transferral and counter transferral

4. Some consequences and some open issues:

56. Plural marking, if head of a count phrase (classifier; DivP) is not separated from L with an FF bracket,
and hence may be included in en-searches. Thus [pl[N]] combination may return a non-compositional
meaning while still being plural.7
a pluralia tantum: scissors, glasses, trousers, pants, news etc. (typologically very common)
b. compounds with plurals including non-compositional (in English mostly with irregular plural but
not so in, e.g., Hebrew and Arabic both of which also have pluralia tantum)
57. Phrasal idioms cannot be the output of a single en-search. Speculatively, the encyclopedic entry for
cross will activate a lookahead en-search for bridge, the entry for bridge will activate a lookahead en-
search etc. (and see Harley and Noyer, 2000 for an explicit suggestion along somewhat similar lines).
Likewise, coat of arms, kick the bucket and similar cases must involve multi en-searches
58. a. I hammered the nail in with my sandal (derived from root, hence flexible)
b. I painted the wall with lacquer
59. a. *I taped the picture with pushpins (derived from word, and hence rigid)
b. *I lacquered the wall with paint
(argument in Arad, 2003, based on Kiparsky 1997)

7
Possibly also some aspectual markers (e.g. Slavic perfective prefixes), which may contribute to the formation of
non-compositional meaning without losing their grammatical function.

Borer 7/23
 Roots and Categories

60. a. Lola taped the poster to the wall with band aids/mailing labels
b. screw the fixture on the wall with nails Kiparsky, 1982) – ok if nails are twisted to affix the fixture
(Harley and Haugen, 2007)
61. Bento boxes of the week: Lacquered with bitter persimmon juice (Google search)
62. a. I taped the picture with a ??(duct) tape
b. I hammered in the nails with a ??(big) hammer (Harley and Haugen, 2007)

5. Synthetic Compounds
5.1. The Conundrums
63. a. If compounds are en-searchable and may return a non-compositional meaning, they cannot (when
non-compositional, at the very least) be derived from structures containing FF.
b. Synthetic compounds (or things that look an awful lot like them) most certainly can be non-
compositional: crystal-gazing, card carrying, line producing; globe trotting, sharp shooting, head
hunting, face lifting, shop lifting, sleep walking, home making, war mongering
c. A conundrum for some models (Borer, 1999, 2003, 2005; Alexiadou 2007, Sichel, 2007): if internal
arguments come from event structure, then the(apparent) internal argument in synthetic
compounds must come from functional event structure. But synthetic compounds do not have an
event interpretation (cf. ‎63(‎64)-(‎65)). So where does the internal argument interpretation come
from?
d. A conundrum for the present proposal – if internal arguments come from functional event
structure, how can synthetic compounds have a non- compositional meaning?
e. Another conundrum for some models (Marantz, 1997; Alexiadou, 2007): if –ing, in derived
nominals, triggers the projection of some particular functional structure (v, for Marantz; VoiceP for
Alexiadou) which effectively forces –ing nominals to be AS-nominals and excludes them as R-
nominals, how are synthetic compounds with –ing but without an event reading possible?
f. Suppose internal arguments are properties of roots, and not of structure (Marantz, 1997,
Harley, 2006, 2008). A number of prima facie conundrum nevertheless arises: first, if roots have an
internal argument (in which case, note, they do have meaning), where does it disappear to in R-
nominals? And wouldn't having an argument mean, effectively, that the root is a verb? (cf.
Acquaviva, 2008). Second, derived verbs have internal arguments (e.g. verbalize), which could
hardly come from the root. But if internal arguments are licensed structurally for verbalize, why
can't they be licensed structurally across the board without root-related information?
64. a. The breaking of the door by Mary in seven minutes in order to collect the insurance
b. (I watched) the door breaking (*by Mary) (*in seven minutes) (*in order to collect the
insurance)
65. The breaker of the door (*in seven minutes) (*in order to collect the insurance)
5.2. Harely (2008)
66. Synthetic compounds are the result of the incorporation of a complement into an a-categorial root
head, with subsequent categorization applying to the result of the incorporation:

Borer 8/23
 Roots and Categories

truck driver; truck driving (different n

types for –er and –ing)

67. Assumptions:
a. roots are listed with complements (also possible: [Pact fast]) for fast acting
b. complements are always realized prior to categorization of the root
c. the root incorporates into the categorial node (n, v)
68. But derived verbs give rise to synthetic compounds. It therefore follows that synthetic compounds
cannot be derived (just) from a root plus its complement.
69. Alas, it is exactly the internal argument that cannot form a synthetic compound with the verb unless
the verb has its transitive meaning:
a. ship sinking – transitive reading only
b. tomato growing – transitive reading only (and compare with tomato growth)
c. *tree falling
d. *train arriving (and compare with train arrival)
5.3. Back to Basics: The First Sister Principle?
70. a. truck driver truck driving man driven
b. letter writer letter writing secretary written
c. bread eater bread eating moth eaten
71. a. fast acting; strong smelling; happy looking
b. quick-fried, slow-roasted
c. pan frying; church going
72. a. *chef maker/making (of cakes); *man driver/driving (of trucks)
b. *cake baked; *letter written; *church gone
73. First Sister Principle (Roeper and Siegel, 1978):
All verbal compounds are formed by incorporation of a word in first sister position of the verb
(presupposed demotion of the external argument and promotion of the internal argument in passive)8
74. a. vacuum cleaner; breech loader; line producer;
b. *fast actor; *strong smeller; *beautifully dancer; *quick fryer
c. *beautiful dancing; vs. quick frying;
d. home-made by hand; sun-baked (too) quickly by us; fire-brewed by boy scouts
e. mountain-grown; garden-tested; field-harvested; night stalker
f. pan fried/pan frying; oven roasted/oven roasting; mountain growing; garden testing; field
harvesting; sun baking; fire brewing

8 The proposal has a number of execution problems if we try to adapt it to present day accounts, e.g., the fact
that in pan frying pan is an adjunct and not a complement. We set these aside, however, for the sake of streamlining
the discussion.

Borer 9/23
 Roots and Categories

75. a. war mongering war mongerer *to monger war

b. baby sitting baby sitter *to sit baby
c. breech loading (weapon) breech loader *to load breech
d. home making home maker *to make home
e. line producing line producer *to produce (a) line
f. sharp shooting sharp shooter ??to shoot sharply
g. crystal gazing crystal gazer *to gaze (into) a crystal
h. church going church goer *to go church
i. shop lifting shop lifter *to lift shop
76. a. orex din arixat din arax sefer/##din
editor law editing law [Vedited] book/##law
‘lawyer’ ‘lawyering’

šomer saf šmirat saf šamar bayit/##saf

guard threshold guarding threshold [Vguarded] house/##threshold
‘gate keeper’ ‘gate keeping’
77. a. -ing and –er synthetic compounds exclude an understood actor interpretation for the non-head
b. -en compounds synthetic compounds exclude an understood undergoer interpretation for the non-
head
c. Any other interpretation of the non-head appears licit if sufficiently contextualized.
5.4. Proposal: it is all in the affixes9
78. a. –er, a FL, means an originator/instrument.
b. *man driver (with man as originator)  two originators in the same expression. Only possible if
one of them modifies the other (e.g., a driver who is a man).
79. a. -ing a FF, means an activity with an incorporated originator (activity+er)
b. *man driving (with man as originator)  two originators in the same expression. Only possible if
one of the modifies the other (driving by a man(ly) agent).

6. -ing
6.1. More evidence that synthetic –ing compounds do not derive from –ing AS-nominals
80. a. *the mongering of war
b. *the sitting of (a) baby
c. *the loading of breech
d. *the lifting of faces
e. *the making of (a) home
f. *the producing of (a) line
g. *the sharp shooting (of the target)
h. *the gazing of crystal (compositional only)
i. *to going of church

9
Discussion of –en, an adjectival head, is set aside here. We further set aside cases of –er which are not
agentive, and which, to the best of my knowledge, are not attested with so-called synthetic compounds. Marantz
(p.c.) notes that these are not licit, typically, in non-modified contexts (e.g., roaster vs. oven-roaster), providing us with
some test for distinguishing between the occurrences.

Borer 10/23
 Roots and Categories

81. a. *the acting fast

b. ??The frying in the pan
c. The pan frying of the pasta (note clear violation of FSP, if this is the source)
6.2. Does –ing always head an AS-nominal?
82. a. Grimshaw (1990): unlike –ation and kin, -ing only assigns Ev and hence cannot be an R-nominal.
b. Alexiadou (2007): -ing (always) projects an (active) VoiceP. As a result, it always has event structure
(i.e., is always an AS-nominal). R-nominals with –ing are impossible.
83. But synthetic –ing compounds are R-nominals!
84. a. a good living, a strong craving, a strong beating, a reading, a (leftist)leaning, (good) standing, (one)
sitting, etc
b. "Women are reared not to feel competent or gratified by the questing, the competing, the
outbidding that collecting … demands." S. Sontag, Volcano Lover, p. 138
c. (this kind of) fighting, fraternizing , parenting, writing etc.
cf: this kind of picture ; this kind of story; this kind of destruction (*of a city)
85. –ing R-nominals are possible, and by and large they all have an activity (simple event) interpretation
6.3 –ing is atelic (Snyder, 1998; Alexiadou, 2001; Borer, 2005b)
86. a. Kim's (*gradual) formulating of several procedures {for the past few weeks/*in few weeks/??twice}
b. Pat's (*gradual) forming of many committees {for three months/*in three months/??twice}
c. Robin's (*gradual) dissolving of these chemicals {for three hours/*in three hours/??twice}
d. Inny's (*gradual) writing of the letter {for three hours/*in two hours/??twice}
87. a. Kim's (gradual) formulation of several procedures {twice/in two weeks}
b. Pat's (gradual) formation of many committees {twice/in two minutes}
c. Robin's (gradual) dissolution of these chemicals {twice/in two hours}
88. a. */#Kim's reaching of the summit
b. */#Pat's ending of the flood
c. */#Robin's finding of (the) oil
d. */#The bulldozer's hitting of (the) bedrock
e. */#The balloon's noisy exploding
f. */#The rabbit's mysterious appearing (and compare with appearance)
89. a. */#the erupting of Vesuvius
b. */#the exploding of the balloon
90. a. */#Vesuvius' sudden erupting
b. */#the balloon's noisy exploding
91. a. the eruption of Vesuvius
b. the explosion of the balloon
c. the appearance of the rabbit
92. a. Vesuvius' eruption
b. the balloon's explosion
c. the rabbit's appearance

Borer 11/23
 Roots and Categories

93. a. The sinking of the ship (under intransitive reading)

b. The falling of stock prices
c. The slipping of standards
d. The laughing of the boys
e. The jumping of the cows
f. The dancing of the fairies
94. And contrast with gerunds:
a. Kim('s) formulating several new procedures {for several weeks/in two weeks/twice/gradually}
b. Pat('s) forming many committees {for three minutes/in two minutes/twice/gradually}
c. Robin('s) dissolving these chemicals {for three hours/in two hours/twice/gradually}
d. Inny('s) writing the letter {for three hours/in two hours/twice/gradually}
95. a. Kim's reaching the summit
b. Pat's ending the flood
c. Robin's finding (the) oil
d. The bulldozer's hitting (the) bedrock
e. The balloon's exploding noisily
f. Mary's disappearing
96. And the progressive:
a. Kim is reaching the summit (as we speak)
b. Pat is ending the flood (as we speak)
c. Robin is finding oil (as we speak)
d. ?The bulldozer is hitting bedrock (as we speak)
e. Mary is disappearing
97. a. The (enthusiastic) formulations/*formulatings of many procedures (by newly appointed bureaucrats)
b. The (occasional) salutations/*salutings of an officer (by his juniors)
c. The (gradual) promotions/*promotings of these incompetent functionaries (by their superiors)
d. The (frequent) replacements/*replacings of many humans with few machines (in thirty years)
e. The appointments/*appointings of three musicians to permanent positions (by the management)
f. The arrivals/*arrivings of the trains
98. a. A formulation/*formulating of many procedures (by newly appointed bureaucrats)
b. A salutation/*saluting of an officer (by his junior)
c. A promotion/*promoting of an incompetent functionary (by his superior)
d. A replacement/*replacing of a worker with a machines
e. An appointment/*appointing of a musician to a permanent position (by the management)
f. An arrival/*arriving of a train
6.3 -ing is not stative (subject is understood as originator)
99. a. Kim loved Pat (stative reading (preferred); eventive-agentive reading)
b. Charles felt the coat (stative reading; eventive-agentive reading)
c. Jenny smelled the stew (stative reading; eventive-agentive reading)
d. Corrine touched Gil (stative reading; eventive-agentive reading)
e. The wall touched the fence (stative reading only, under normal circumstances)

Borer 12/23
 Roots and Categories

100. a. The loving of Pat (by Kim) (eventive-agentive reading only)

b. The feeling of {#the cold/the coat on his shoulders} (by Charles) (eventive-agentive reading only)
c. The smelling of the stew (by Jenny) (eventive-agentive reading only)
d. The touching of Gil (by Corrine) (eventive-agentive reading only)
e. The touching of the fence (#by the wall) (eventive-agentive, abnormal under normal circumstances)
101. a. Kim's loving of Pat (eventive reading only)
b. Charles' feeling of {#the cold/the coat on his shoulders} (eventive reading only)
c. Jenny's smelling of the stew (eventive reading only)
d. Corrine's touching of Gil (eventive reading only)
e. #The wall's touching of the fence (eventive, abnormal under normal circumstances)
102. a. The wall's (persistent) adherence/*adhering to the fence
b. Dennis' (tenacious) endurance/*enduring of the noise
c. The stain's (sad) resistance/*resisting to cleaning

7. Back to synthetic nominals – some more on why synthetic nominals are not AS
nominals
103. Compounds: transitive reading only AS Nominals: intransitive reading possible
tomato growing the growing of the tomato
ship sinking the sinking of the ship
#guest departing the departing of the guest
#flower wilting the wilting of the flower
#standard slipping the slipping of standards
104. fence-touching *wall *stative, activity
stew smelling *stative, activity
(under relevant reading. ‘smells like stew’ is ok)
105. tomato grower
ship sinker
fence toucher
stew smeller
106. This kind of growing, sinking, dropping, disappearing, all allow, indeed appear to prefer, an
intransitive, but activity reading.
107. It cannot be that nominal -ing always forces a little v or an active VoiceP reading.
108. Transitive reading only for synthetic compounds - problem for:
a. The assumption that the transitivity of, e.g., grow derives from a projected v or CAUSE
b. Any account based on the First Sister Principle which subscribes to the unaccusative hypothesis
c. Any account based on the First Sister Principle which subscribes to the view that causative are
derived from adding structure to inchoatives (acknowledged by Harley, 2008)
109. The growing of the tomato
a. -ingoriginator [EP (expl) [FsP the tomato ]] transitive, activity
participant
b. -ingoriginator [EP the tomato ] intransitive, activity
originator2

Borer 13/23
 Roots and Categories

8. Late insertion of meaning - summary

8.1. One substantive meaning list, once accessed:
110. There are good empirical reasons to assume that roots are inserted with at least some phonological information,
111. Empirical reasons for assuming that roots are inserted with meaning appear dependent on a fortiori
conceptual distinctions motivated by grammatical phenomena. These are hard to evaluate without a
worked out cognitive theory of concepts and the restrictions on their vocabulary correlates
112. It is not clear what, if anything, is explained by the assumption that roots have internal
arguments/complements.
113. If there are no good reasons to assume multiple substantive meaning lists or multiple access, by
Occum’s razor, one list/one access suffice. Since all en-searches for a specific constituent happen
simultaneously and silent copies are not en-searchable, reanalysis is unnecessary. It raises the
possibility that some simultaneous assignment may co-exist, context allowing (e.g., the meaning of the
root š.m.n ‘to do with fat’ and the meaning of the noun šamenet – ‘cream’).
8.2. One more point: the encyclopedic factory
114. a. [war monger]  to war monger
b. [baby sit]  to baby sit attested
c. [breech load]  to breech load
d. [wool gather]  to wool gather
e. [home make]  to home make
f. [house keep]  to house keep
f. [line produce]  to line produce attested
g. [sharp shoot]  to sharp shoot attested
h. [crystal gaze]  to crystal gaze attested

Appendix: Categorizing and against  categorial affixation

A1. Introduction
115. Grimshaw (1990): English -nominalizers do not allow for event structure and are hence, by and large,
excluded as AS-nominals.
116. change (and exchange), release, use, (among others)
117. A run, a stand, a sit in, a lie down, a walk, a ride, a dance, a turn, a twist, a smoke, a smile, a laugh, a
frown, love, hate, a kiss, a lift, a roll, a rock, a hold, a climb, a descent, a kill, a raid, an arrest, a follow-
up, a chase, an export, an import, a think…… (and see below for more examples)
118. *the walk of the dog for three hours
*the dance of the group for a whole evening
*the kiss of the beloved in order to seal the marriage

Borer 14/23
 Roots and Categories

119. a. T b. D c. Vize
3 3 3 33
FUT [L=VFORM] det [L=NFORM] Aal Vize aal
PST 3
[L=N√FORM] AAal

120. Nothing formally merges directly with a root, nor are roots, as such, well-defined syntactic objects. As
the root is, definitionally, a member of a specific category in the context of the functor selecting it, be
it an L-functor or an F-functor, all mergers are, posteriori, minimally, to L.
121. English does not have (productive)  FLs. There is, in other words, no  correlate of '-ize' or a 
correlate of '-ation', such that when they attach to a root they convert it, effectively, to a noun or to a
verb without any overt affixation to mark such a change.
122. Contra Kiparsky (1982, 1997), and contra DM.

Borer 15/23
 Roots and Categories

A2. -categorizers (FLs) in DM

a. T b. D
123.
3 3
FUT v det n
PST 3 3
FORM v FORM n

a. vize b. vize
124.
3 3
aal vize aal vize
3 3
n aal √FORM aal
3
√FORM n
c. nation d. nation willT
3 3 3 3
v nation √FORM nation
willT v
3 3
√FORM v
e. n willT
3 3
v n willT v
3 3
√FORM v (??)

125. Non-compositional meaning is available for (‎124d) but not for (‎124c) (is this a falsifiable claim?)
126. form, a noun, is as morphologically complex as formation (Kiparsky, DM);
form, the verb, is as morphologically complex as liquefy (DM, but not Kiparsky).
a. [N*√FORM+ n] [N*√FORM+ nation]
b. [V*√FORM+ v] [V*√LIQUI+ vify]
127. In XSM, neither the noun nor the verb form are morphologically complex.
128. government, in DM:
a. n  -ment / √GOVERN ____
[V*√GOVERN] v] ____
b. -ment only attaches to v: *[V*√GOVERN] v-] (e.g., vify, vize, vate etc.)
Query 1: why does –ment only attach to roots which may otherwise be verbalized with a v? Why
not to roots that are otherwise nominalized or adjectivized?
Query 2: why should the root GOVERN and the verb GOVERN systematically select the same
nominalizer?
129. In XSM:
a. N  -ment / [LGOVERN] ____
b. ment verbalizes its complement.

Borer 16/23
 Roots and Categories

130. a. a salute to salute

a form to form
a chair to chair
a floor to floor
a lamp to lamp
b. a salutation *to salutation
a formation *to formation
an arrival *to arrival
a friendship *to friendship
a neighborhood *to neighborhood
131. a. a dance to dance
a kiss to kiss
a run to run
a walk to walk
b. *a verbalize to verbalize
*an instantiate to instantiate
*a liquefy to liquefy
*an encase to encase
*a fatten to fatten
132. a. a wardrobe to wardrobe
a blackboard to blackboard
chicken wire to chicken wire
b. history teacher *to history teacher
baby sitter *to baby sitter
government policy *to government policy
133. How to exclude (‎130)-(‎132) in a grammar with zero categorizers:
a. If a root is verbalized by any affix other than v, it cannot be nominalized with n
b. If a root is nominalized by any affix other than n, it cannot be nominalized with v
134. Alternatively: zero-affixation is restricted to roots, a problematic statement for a system with late
insertion of functors (as well as unfalsifiable), where no generalizations based on phonological content
are expected, let alone across categories.
135. And, not to forget:
n cannot merge with a functional event structure (and as such is distinct from nation, nence, nment, nal,
and nine).
136. A theory internal problem for DM: English default past tense form is the only one that attaches to
overtly derived verbs, a fact that should derive directly from locality – exceptional information
specified only on roots – but in (b) roots are not any more accessible than in (a). Similar issues emerge
with irregular plurals:
a. [V*√INSTANT+-iate]; [V* √LIQUI + –ify]; [V*√HARMON+ –ize]; [V*√FAT+ –en]; [V en- *√CASE+ +
b. [V*√STAND + -]; [V*√SING + -]; [V*√BEND + -t]
137. a. [V[L=N√INSTANT+-iate]; [V[L=N √LIQUI + –ify]; [V[L=N,A√HARMON+ –ize], [V[L=A√FAT+–en]; [Ven-[L=N√CASE++
b. [T[L=V√STAND++; [T[L=V√SING + + [T[L=V√BEND ++

Borer 17/23
 Roots and Categories

138. Similar problem with –ation (when contrasted with –ence, -al, -ment), which is the only form available
following (non-zero) verbalizers (e.g., -ate, -ify, and –ize), but which, given zero categorizers, are as
complex making a non-local statement necessary.
A2. An Alternative.
139. a. [T [L=V √FORM++; [Asp [L=V √FORM++ etc.
b. [D [L=N √FORM++; [DIV [L=N √FORM++ etc.

In the absence of zero affixation, overt affixation is necessary to turn nouns into verbs and vice verse.
In the absence of such additional overt affixation, embedding, e.g. formalize in a nominal context quite
simply leads to a contradiction: the L is already a verb, but is interpreted, contextually, as a noun.
Conversely, formation, when embedded in a verbal context, is interpreted as V-equivalent, but as it is
already a noun, contradiction and hence ungrammaticality result:
140. a. *[T [N formation]]; *[Asp [N formation]] etc.
b. *[D [V formalize]]; *[DIV [V formalize]]

When formation is embedded in an N-equivalent context, or formalize in a V-equivalent context, they

are interpreted, by virtue of their contexts, as N and V respectively, which in this case adds little
information, but which, given the nature of definitional categorization, has no cost:
141. a. [D [N formation]]; [DIV [N formation]] etc.
b. [T [V formalize]]; [Asp [V formalize]]
142. a. to grandstand, to blackboard, to chicken-wire, to wall-paper, to (non- compositional); to French
kiss etc.
b. (much) white-out, a take-off, a sell-out, a buy-up; a take-over, a teach-in, a show-down, a sing-
along etc.
143. a. [L [L … ] Lhead … ] ]
black board
wall paper
French kiss
b. [L [ Lhead … [ P ]]
white out
take out
show down
144. a. [Asp [L=V [L … ] L…+ +
(to) black board
(to) wall paper
(to) French kiss
b. [D [L=N [L … ] L…+ +
(a) black board
(a) wall paper
(a) French kiss

Borer 18/23
 Roots and Categories

145. a. [Asp [L=V L [ P +… + +

(to) white out
(to) wash off
(to) buy up
b. [D [L=N L [ P +… + +
(a) white out
(a) wash off
(a) buy up
146. a. movie director; law enforcement; word formation; piano recital
b. *to movie director; *to law enforcement; *to word formation; *to piano recital
147. a. [N [L=V√DIRECT+ Nor ]
b. [N [V Ven [L=N√FORCE++ Nment ]
c. [N [L=V√FORM+Nation ]
d. [N [re- [L=V√RECITE++ Nal ]
148. [D [N [L … ] N… ] ]
movie director
law enforcement
word formation
piano recital
149. *[Asp-Q [N [L … ] N… ] ]
movie director
law enforcement
word formation
piano recital
150. N
3
N E
{-ation} 3
-ing Spec 3
E (ASPQ /Fs
3
Spec 3
ASPQ) L(=V)
ADMIT
√FORM
151. D
3
D L=N
the ADMIT 'the admit'
√FORM 'the form'

Borer 19/23
 Roots and Categories

152. a. *the form of 2 special committees (by the board of directors) (in order to oversee the elections).
b. the formation of 2 special committees by the board of directors in order to
oversee the elections.
c. the forming of 2 special committees by the board of directors in order to
oversee the elections
153. a. *the admit of 3 graduate students (in order to just teach Russian) is not a good idea.
b. the admission of 3 graduate students (in order to just teach Russian) is not
a good idea
c. our admitting of 3 graduate students (in order to just teach Russian) is not
a good idea.
A3. Rebutting some Potential Counter Arguments
154. a. a portion, a position, a condition, a proposition, an audition, a ration, a question, a motion, an air
condition
b. to portion, to position, to condition, to proposition, to audition, to ration, to question, to motion,
to air condition
155. a. port-, cond- rat-, mot-
b. to pose, to propose, to audit, to quest)
156. a. the model posed in front of the camera for several hours
b. Mary proposed a solution in two minutes
c. The authorities audited my tax records for several years
d. John quested for love for years
157. a. the posing of the model in front of the camera for several hours
b. Mary's proposal of a solution in two minutes
c. The authorities' auditing of my tax records for several years
d. the questing for love for years on the part of adult males
158. a. the position of the model in front of the camera (*for several hours)
b. Mary's proposition of a solution (*in two minutes)
c. The authorities' audition of my tax records (*for several years)
d. *the question (for years) for love on the part of adult males
= with the intended reading as events associated with ( ‎156a-d)
159. The nouns in (‎154a) have exactly the cluster of properties we typically find with underived nouns:
a. They alternate freely with verbal forms
b. They may not function as AS-nominals.
c. They only nominalize with an overt suffix
160. a. The condition *(ing) of certain responses by certain works of literature
b. John's proposition *(ing) of Mary
c. Mary's question *(ing) of John
d. Kim's audition *(ing) of the candidate
A4. Kiparsky's (1982, 1997) arguments
161. There are two zero categorial affixes in English, one which derives verbs from nouns and which is a
level I affix, and another one which derives verbs from nouns, and which is a level II affix.

Borer 20/23
 Roots and Categories

162. a. Contrary to XSM but in line with DM, there are (two distinct) abstract categorizers in English.
b. Contrary to DM and XSM, zero verbs and zero nouns are not derived independently from
roots, but from each other.
A4.1. Two denominal verbs in English? - cf. discussion above (‎58)-(‎62)
A4.2. Morpho-phonological arguments?
A4.2.1. Productivity
163. Level II affixation takes as its input phonological words and is very productive.
Level I affixation takes as its input either words or non-word stems, and is (potentially) less
productive.

Kiparsky’s Claim: N to V is very productive. Not so V to N – but see (‎117)

164. a. an àdmit (a newly admitted student)
b. an admìt (a record of a positive admission decision or a newly admitted student)
c. give it a think (note the absence of blocking from thought)
d. give it a turn (meaning turn it)
e. a responsive read (proof-reading process, from a magazine editorial board)
f. there are people on that list who deserve a listen
g. A scheduled (court) hear
h. an embed (a journalist 'embedded' in a US military unit)
A4.2.2. Stress shift
165. VN alternations may give rise to stress shift (i.e., torméntV  tórmèntN, admítV ádmit), but not so
NV alternations (i.e., páttern  *pattérn). If Level II affixation is involved in NV, and if stress rules
are applicable only to the output of Level I affixation, the absence of stress effects of the change from
NV would be explained.

But consider (‎164a-b). Stress has shifted in (‎164a) but not in (‎164b)
Two zero-affixes associated with VN?

The existence of stress shift in some syntactic contexts but not in others is entirely compatible with
a system that allows for roots that are phonologically underspecified, together with the late
insertion of fully specified phonological information.
A4.2.3. Ordering
166. Kiparsky’s claim: Cases such as (‎154), in which verbs are formed from –ation affixed words, argue for
NV later than –tion affixation. NV in compounds likewise argues for a late NV rule, |but an
early VN rule
But Kiparsky cannot account for the systematic ungrammaticality of “to transformation”, “to
friendship” or “to union brotherhood”, all predicted to be possible by his logic.
167. Underived verbs (within his system) may receive Level I affixation, but not so (zero) derived verbs,
formed by assumption at Level II:
contract-ual, murder-ous, rebeli-ous
By this logic, the noun market is derived from the verb market, and the noun question
from the verb question (which is actually a contradiction in his system):

Borer 21/23
 Roots and Categories

salvageabílity, marketabílity, questionabílity

168. Kiparsky: VNV triplets should occur, but not so NVN triplets:
a. compóundV(1) cómpoundN (2)  cómpoundV (3)
b. permítV (1)  pérmitN (2)  pérmitV (3)

As it turns out, however, NVN triplets do occur (as Kiparsky himself notes):

c. sweatN(1)sweatV(2) sweatN(3) ((1) = substance; (3) = result N)

d. spitN(1)spitV(2)spitN(3)
e. paddleN(1)paddleV(2)paddleN(3)

K.’s explanation: there are two distinct rules deriving nouns from verbs, both applying at Level I:
Rule 1: VResult N (to) sweatV  (have) a sweatN
Rule 2: V Substance N (to) sweatV (smelly) sweatN

Note further the presupposition that sweatN is derived from sweatV, shitN from shitV etc.
169. a. I sank the boat
b. that was quite a sink! (viewing the action in (a)
c. I bought a new sink
d. I am going to (re-)sink the kitchen
170. Ironically, tape and screw may be both a substance and a result:
a I just got my hair cut and asked for a tape up
b. Visualization of the flow through a screw-down valve

But that would mean that they are derived from verbs, in contradiction with the purported
properties of tape vs. hammer.

References
Acquaviva, P. 2008. Roots and lexicality in Distributed Morphology. Ms. University College Dublin/Universität
Konstanz.
Alexiadou, A. 2007. On the role of syntactic locality in morphological processes: the case of (Greek) derived nominals.
To appear in A. Giannakidou & M. Rathert, eds. Quantification, Definiteness and Nominalization. Oxford: Oxford
University Press.
Alexiadou, A. 2001. Functional Structure in Nominals: Nominalizations and Ergativity. Amsterdam: John Benjamins.
Arad, Maya. 2003. Locality constraints on the interpretation of roots: The case of Hebrew denominal verbs NLLT 21:
737-778
Bat El, O. 1994. Stem modification and cluster transfer in Modern Hebrew. NLLT 12: 571-596.
Borer, H. 1991/1993. "Derived Nominals." Ms., University of Massachusetts, Amherst.
Borer, H. 1999. The form, the forming and the formation of nominals. Talk given at ZAS, Berlin.
Borer, H. 2003. Exo-skeletal vs. endo-skeletal explanation: Syntactic projection and the lexicon. In J. Moor and M.
Polinsky, eds. The Nature of Explanation in Linguistic Theory. Stanford: CSLI
Borer, H. 2005a,b. Structuring Sense Vols. I, II. Oxford: Oxford University Press.
Borer, H. 2008. Compounds: The view from Hebrew. To appear in R. Lieber and P. Stekauer, eds. The Oxford Handbook
of Compounds. Oxford: Oxford University Press.
Embick, D. and A. Marantz. 2008. Architecture and blocking. Linguistic Inquiry 39.1.

Borer 22/23
 Roots and Categories

Engelhardt, M. 2000. "The Projection of Argument-taking Nominals", Natural Language and Linguistic Theory 18 (1).
41-88.
Grimshaw, J. 1990. Argument Structure. Cambridge: MIT Press.
Harley, H. 2006. The morphology of nominalizations and the syntax of vP. Ms. University of Arizona
Harley, H. and J. Haugen, 2007. http://www.ledonline.it/snippets/allegati/snippets9002.pdf “Are there really
two different classes of instrumental denominal verbs in English?” Snippets: Issue 15
Harley, H. 2006. The morphology of nominalizations and the syntax of vP. In M Rathert and A. Giannadikou. Eds. DP,
QP and Nominalizations. Oxford: Oxford University Press.
Harley, H. 2008. Compounding in distributed morphology. To appear in R. Lieber and P. Stekauer, eds. The Oxford
Handbook of Compounds. Oxford: Oxford University Press.
Harley, H. and R. Noyer. 2000. Licensing in a non-lexicalist lexicon. In B. Peeters, ed. The Lexicon Encyclopedia
Interface. Amsterdam: Elsevier 349-374
Hazout, I. 1991. Verbal Nouns in Hebrew and Arabic. PhD dissertation, University of Massachusetts, Amherst.
Hazout, I. 1995. Action nominalization and the Lexicalist Hypothesis. Natural Language and Linguistic Theory 13: 355-
404.
Kiparsky, P. 1982. Lexical Morphology and Phonology in Linguistics in the Morning Calm, Hanshin Publishing Company,
Seoul, Korea.
Kiparsky, P. 1997. Remarks on denominal verbs. In A. Alsina, J. Bresnan and P. Sells. eds. Complex Predicates.
Stanford: CSLI Publications.
Marantz, A. 1996. “Cat as a Phrasal Idiom", ms. MIT.
Marantz, A. 1997. "No Escape from Syntax," In A. Dimitriadis, I. Siegel et. al, (eds.) University of Pennsylvania
Working Papers in Linguistics. 4.2., pp. 201-225.
Marantz , A. 2000. Roots: The universality of roots and pattern morphology. Paper presented at the conference on
Afro-Asiatic languages, University of Paris VII.
Marantz, A. 2001. Words and things. Ms. MIT.
Roeper, T. and D. Siegel. 1978. A lexical transformation for verbal compounds. LI 9, 199-160.
Sichel, I. 2007. Agent exclusivity in nominalizations. Paper presented at the Workshop on Bare Nouns and
Nominalizations, Stuttgart, June 2007
Snyder, W. 1998. On the aspectual properties of English derived nominals. In U. Sauerland and O. Percus, eds. The
Interpretive Tract: Working Papers in Syntax and Semantics (MITWPL Volume 25), pp.125-139. Cambridge, MA:
MIT Working Papers in Linguistics.

Hagit Borer
Department of Linguistics
University of Southern California
Los Angeles, CA 90089-1693
USA
[email protected]
http://www-rcf.usc.edu/~borer

Borer 23/23