rHA!

'Ti:R 7

Sociocullur.il F:Jt:lor>

tricd to
Think of one or t\VO languages you re, ..tn11.t.u- wtrh or vou\:e
.
lcarn. How <lo you feel ahout tlle peo ple of rhe culrure of that languagc~
Any tnxcd feelings?
,
Look at iteni 4 011 page 203 and v,rrite ahout an cxan1ple oi. one or_ inure
nf J-Iofstede's categories in your O\Vn current or past exper1cnces in lan+

guage classroon1s.
.
. . 1
Do \'OU personally think the spread of English in the colon.1al. a h,t(.
inip~rialistic overtones? llow c:1n you as an English teacher in thts llC\V
n1ille11niu1n avoid such cultural in1peralisn1?
lvlake a Hst of words, phrases, or language rules in your foreign languagc.
~.

11 si T'1ke two or three

thar are good exainples of the w
w 101 t1.u1 1ypo 1e s. . uf.
those a~d write about whether or not you think the language 1t,<,el
shapes the \\tay speakers of that language think or feel.

e:

--"<RG12 LA;N6uA6C
L..~-,~~ ...q ~~"-"~"""""~~~

CRQSS--_LIN G UISTIC.~-

lNFLUENCE AND
LEARNER LAl\LGUAGE_ _~

Cci-.,;t~S'li\IC: ~Nf~\.y,S".

~6UA.z~

1
1

PTO this point in the trcattnent of principies of se-cond language ac.quisirion,

focus has been on the psychology of language learning.
of second language acqu!si.tir1n forn1 the founc!ation stones for building a con1prehensh'e underst-lnding uf rhe accuisilion
of rhe linguistic system. In this chapter '1.1e 1vil1 take a differenr direction a:-.
we begin ro exa1nine the n1ost sa!ient con1ponent of secon<l languagi::
acquisition: thefJ;fnguage~[jThis treatJnent ~-ill first consider, in ~
ical progressiorl, an era of preoccupation \Vith studies of c.ontr<J.Sts bet\Veen
tJ1e_ na,riveAanguage aJllj the t~~Q;et ,l;lnguag~ ( C?lJ!r:1':.I,~2D. ~l.1l::ii") 1ncl _thc
effect of native on.t~~get langu;a.,e (now callCd ''cross-!ingni<>tic influenCe"f
\-Xfe \Vill then see houl the -~!21_9J_~:_s~n_tra<>tlve an:1lysis gave v:ay toan era or"
t;rror. a.naly~-i~, 1.yirh its gulingcQ!)CTptf~~rElri:gclJge ~.Jli2iv also \viciely
r:efen::ed o as "learner l:inuag~_~,'~Finaiiy. qu.estions ah;);.rt the effect of classroon1 instruction and error treatl11ent \Yil! be addn:s.sed, "\vit11 ::.on1t practica! in1plications for the language teacher.

our
U
Psychological principies

THE CONTRASTIVE AL"lAIYS!S HYPOTHESIS

In ihe 1;i(!C()ffhe""CTVe!1t'ieth:~centu:V::J)ne of the .rnost pnpLir pur:;1iirsJ
for :tpplicd Hngui~~:.'i \',~s the "tudy of t"\YO 1:1ng\1;ige:s in contr:1st. l:Yei1 tun~
rile stoc.kpilc of c<:n1p-:ir:HiYe and contr;1stivt dar on a inultitudt' of pairs of
L1ngu:<ges y\cldcd \\'lu.t comn1on!y c:in1e to he knn\\'n as E:: Cont:rastlve
An.:ii'ysis flyfn.1tl1esiS (CAH). Deeply
hehaviodstic and srrtH>

207

208

Cl-nPTER 8

Cross-Linguistic lnfluence and l..e,irner l..anguage

turalist approaches ofthe Uay, the CAEL~ that thc princip;tl bal'rier to
second language acquisition is the'iifterfereni..::c of the fh:st. language system
'W'itlrthe second language system, and that a scientific, strucrural analysis of
the two languagcs in qui;stion Yvould yie!d a taxono1ny of linguistic contrasts between the-m which in turn would enable the linguist to-preclict the
difficulties a learner woul<l encounter.
-It 'vas at that tne considered feasible that the tools of structur.1! linguistics, such as Fries's (1952) slot-filler gran11.11ar, would enable a linguist to
accurately describe the two languages in question, and to match those t"\-VO
descriptions against each othcr to deternline valid contrasts, or differences,
bctween them. BehJ.vioristn contributed to the notion that human behavior
is the sum of its s111allest parts and cornponents, and therefore that language lc:i.rning could be dcscribed as the acqu;ition of all of_ those discrete
units. i\loreover, hurnan it.:arning theories highlighted i!tte1feritig'.;elements
of learning, coaclud-ing that \Vhere no interference ccilt1d be pre-dicted, no
.lifficulty 1vould he experienced sincc one couid transfer positively ali
other ite1ns in a lrtnguage. The iogical conclusion frurn these various psychologcal and linguislic as.sumptons ~vas that second language learning
basically involved the overconling of the differences benveen the two linguistic systems-the nat.ive and target languages.
Intuitively the CAH has appeal in that we comn1onlv observe in
scco~d language l~ar~ers a plethora of~attributab:e to..the negative
transfer of the nat1ve 1anguage to the target language. It 1s quite common,
t~)f example, to detcct ccrtain foreign acccnts and to be able to infer, fro1n
the speech of the learner alone, where the learner con1es fron1. Native
English speakers can easily identify the accents of English language
lcarners fion1 Gerrnany, France, Spain, and Ja pan, for ex~unplc. Such accents
cJ.n even be repn:sented in the written word. Consi<ler Tv1ark Twain 's The
Inuocents Abro<-t.:l (1869: 111), in which the French-spcaking guide introduces himself: "If zc zhentlcmans "'ill to me n1ake ze grande honneur to
n1e rattain in hces scrveece, I shall show to him everysing zat is magnifique
to lcok 11p0n In z:::: beautif11l Paree. I speaky ze Angleesh oirf:.liun;iw." Or
\'/illiam E. CJ.Uahan's Juan C:Istaniegos, a young IVJe,""(ican in .A/raid of the
Dark, who says: u:riclp n1e to leave from thees piace. But, Seor Cap~tn, me,
rave do notheeng. Nuthceng, Seor Capitn." These e.xcerpts also capture
the transfer of vocabulary and gran1matical rules from the native language.
Son1e rather strong cL'lllns were ma<le of the CAH by language
teaching experts and linguists. One of the strongest was n1ade by Robert
Lado (1957: vii) in the preface to Li11guistics Across Cultu1es: "1'he plan of
the book rcsts on the assumption that we can preJct an<l describe the patterns that 'viU cause <lli11culty in lcarning) and thosc lhat \Vill nol cause difficulty, by cun1pasing systcn1alically the la11guage and the culture to be

Ci-1;.PnR 8

Cross-Linguistic lnfluence and Learner Language

209

learneJ '\Vith the native language and culture of the student." Then, in the
first chapter of the book, Lado continues: "in the con1parison benveen
native and foreign languagr: lies the key to ease or difficulty in foreign language learning .... 'fhose elements that are similar to [the learner's] native
language will be sitnple for hin1 and those elements that are <lifferent "'"ill
be clifficult" (pp. 1-2). An equally strong claim was made by Banathy,Trager,
and \Vaddle (1966: 37): "The change tha_t has to take place in the language
behavior of a forcign language student can be equated with the differences
between the structure of the student's native languagc and culture and that
of the targct language and culture."
Such claims were supported by what sorne researchers clain1ed to be
an empirical n1ethod of prediction. A well-known model was offered by
?~_(j-~-~~l_l_,_-B~?~' ~nd-Mrii.-_(1965),.who p_o_site_d what they called a hier:.
-aC-l1y f difficulty by which a teacher: or lriguist could make a prediction
of the relative difficulty of a given aspect of the target language: For phonological systems in contrast, Stockwell and his associatcs- suggested ight
possible <legr<::es of difficulty. These degrees were Based upon the notif1s
of trahsfer (positive, negative, and zero) and of optional ancl bligatOry
choices of _certalll phonemes in the two languages in contf'J.St. Through a
Yery careful, systematic analysis 0f the properties of the Gvo languages in
reference to the hierarchy of difficulty, applied linguists were able to derive
a reasonably accurate inventory of phonological difficulties that a second
language learner would encounter.
Stock'ivell and his associates also constructed a hiera:tchy of difficulty
for granun-atical stiuCtures of two language:;; in contrast. Their grammatical
hierarchy included sixteen levt.:ls of difficulty, based on the sanie notions
uscd to construct phonological criteria, \Vith the addcd dirnensions of
"structural correspondence" and "functional/semantic correspondence.'
Clifford Prator (1967) captured the esscnce of this gra1nmatical hierarchy
in sL"t categores of difficulty Prator's hierarchy \Vas applicable to both
gr:ammatical ad phonological features of language. The six categories, in
ascending order of difficulty, are Iisted below. Most of the examplt1S are
taken from English and Spanish (a native English speaker lcarning Spanish
as a second language); a few examples illustrate other pairs of contrasting
languagcs.

Leve! O - T::t'a[i_S,f~r'.:i No difference or contrast is present bet\veen the

two 1anguages. The:learner c-an simply transfer (positively) a
sound, structure, or lcxical item frotn the native language to the
target language. Examples: English and Spanish cardinal vo\vcls,
\vord order, and certain \Vords (ntortal, inteligente, arte, an1ericanos).

Ci-/Af'T[R 8

Cross-Linguistic fn(/1f'nce ;!lld Learner Language

~~vet-1:;_'- C()ales_ce_nce&:Two iten1sin the native language hecon1e coa

lesced in to essentaUy one iten1 in tlle target_ language. This
requires that learners over!ook a distinction they haYe gro\vn

accuston1ed to. Exan1ples: English third-person posses~Jives require

gender distinctioo (1Jis/her), ;ind in Sp::tnish they do not (.su); an
English speaker learning Fr~nch mt1st overlook the clistinction
berween teacb and learn, and use just the one '\vord apprendre in
French.
Level_2
D:det<liff'erntiatiot'.tAn item in the-na1i\"e language is
absent in the larget language.The learner 1nust avoid thar iten1.
Exa1nples: English learners of Spanish inust "forget" such iten1s as
English do as a tense carrier, possessive forn1s of Il'!J- \Vords
(zvbose), or the use of -~n1ne with mass nouns.

Level 3 - --~~interpretation:An item that exists in the native language

is givcn a new Shape or distribution. Exan1ple: an English speaker
Jearning I4 rench must learn a nev distribution for nasalized
- vowels.
Leve! 4 - Vefdiff-ntiatiori~ A-:riew item entirely. bearingJittle -f
any siri-{.l-rY- to the native ianguage iten1, nn1~t be learned.
Exan1ple: an English speaker learning Spanish n1ust learn to
include dererminers in generalized nominals (Nlan is 1nortal/El
hon1/Jre es 1nortal), or, most con1monly, ro learn Spanish gram1natical gender inherent in nouns.
Level S - ~plit.'-'one item in- the native language- becomes tv.ro or more
in the target language, requiring the learner to n1_ake a nev distinction. Exa1nple: an English spe;iker_ learning Spaq,ish n1ust learn
the distinction- between ser and estar (to he), or the distinctioo
between Spanish inclicative and subjunctivc 111oods.
Prator's reinterpretation, and Stockwell and his assocates original hierarchy of difficulty, were based on principies of htunan learning. The first, or
"zero," degree of difficulty rcpresents co_n1pl~te one-to-one correspondence
and .transfer, while the fifth clegree of difficulty was the heighr of intcrference. Praror an<l Stockwell both claned that their hierarchy could be
applied to virtually any two languages and 1nake it possible ro predict
secon<l language learner <lifficulties in any language with a t~tir dcgree of
certainty and objectivit)''.

TC\)-

(<._

U-f,\i'fi-'R B

Cross-L1nguistic lnr.luence ,1nrl Learner /.an:udge

211

FROM THE CAH TO CU (CROSS-UNGVISTIC

INFLUENCE)
..
Pre<liction of difficulty by n1eans of contrastive procedures \\1as not
\Vithout glaring shorrcomings. For one thing, the process \V:1s oversimplified. Subtle ph{inetic, phonological, and gran1n1arical {lhtinctions were not
carefully accounted f(Jf. Second, it \Vas very difficult, even '\Vltb six caregories, to <letern1ine exacrly which category a p1rticuLrc contrast fit into.
For exan1ple, when a ]:1parH.:s~ speaker learns the Fnglish /r/, is ita case of
a level O, l, or 3 <lfficulty? J-\ ca:>e can he n1ade for all three.The third and
1nosr proble1natic issue centered on the larger qut'stion o whether or not
predictions of diffi.culty Jevels \:vere actually verifiab!e.
The atten1pt to pre<lict difficulty by n1eans nf contrastive analysis is
what Rp,n;tli;J. \-V:;u-ufia-i1g-b (1970) cal!ed the[5trong -versior1Jof rhe CAI-i, a
version rhat he - believed "\VlS guite __ -t_i_nrc-;!Ji:,t_ic anti irnpracti-cab!e.
\V'ardh:iugh no'ted (p. 125) that 'ar the very least, this version dc1nands. of
lingusts tLat they have :nai1able a set of !ingui<>tic unh'tT-;als formulated
within a comprehensiYe linguistic theory which clt:als adequately with
syntax, semantics~ and phono!ogy." He went on to point out rhe difficulty
(p. 126), already noted, of an adequate procedure, bnilt nn snt1nrl theory. for
actually contrast~g the fonns of languages:"Do lingl1ists have avaibble to
the111 an overall contrasth-e systen1 \Vithin which they can relate the tvo
ianguages in tern1s of mergers, splits, zeroes, over-diffcrentlatinns, underdifferentiations, reinterpretations?" And so, \\'hile many Hngu~ts clain1t:d
to be using a scientific, e1npirical, and theoretically jll'>tified tool in contrasrive analysis, in actualiry they were operating n1ore out of n1entalistic
subjectity.
\Y:.~rdhaugh- notcd, hO\\-'ever (p. 126), that contr;istive ;in:llysis ha<l inruitive appeal, and that teachers and Unguists had succcs.sfully used 'the best
linguistic knowledge av-ailable
in order to a-z:count for obscrYcd difficulties in second language learning: l-Ic termed such nhscr.::itinnal use of
cont1-;i~ti-ve :in:1ly<~ rhe ~e:ak---ve-rsion of-1he:C/"'l-L Th~ \\Tak versinn <loes
not in1ply the n ;rfnri prcdiction uf certain dep:rees of difficulty. It. recag-_nizes-the<s:ignittcance of:interference-tic10ss-Jn1:Igu11ges:vtl1c fact that sl1ch
nterterence cines exst :111l..l can cxpl:iin diCficnltic~. hut it 11"1 recng\'izes
th;:t linguistic difficu!ties can be rnore profitahly expl:1tned u posteriorf~.;t-J)c. E1c1 i\S learners are learnin ~ rhe T:lg;lge ~'.U1d errors appear.
teaChcrs e.in uti 1ze t 1eir knO\Yledge of the t:1rgcr anc n:HYe languages to
understand sources of error.
The so-c:tlled \\'e:ik ,-ersion of rhe CAI--1 is vh;ir rt"n1:1ins toLhly under
the lahe! cross-lin.guistic b:U:111e11ce (_(L1), suggesting th:it \Ve atl rt:Cog~
nize the -signi.ficant rok: th:H prior experience playt> in any lcarning act. and

212

Cl-!APTER rJ

Cross-Linguistic lnflucnce and Learner Language

that the nfluence of the nalive language as prior experience 1nust not be
overlookcd. l'he diffcrence between today's cmphasis on influence. rather
than prediction, is an important one. Aside fro1n phonoiogy, which re1nains
the most reliable llnguistic category for predicting learner perforn1ance, as
illustrateJ at the beginning of the chapter, other aspects of language
present more of a ga1uble. Syntactc, lexical, and senuntic, in,terference
show far more variatiun arnong learners than psycllun1otor~based pronunciation interference. Evcn presumably snple gr..tn1111atical categories like
vvord ordcr, tense, or aspect have been shown to contain a good <leal of
variation. For exa1nplc, one n1ight expect a French speaker who is beginning to lcarn Englisll to say "I an1 in Ne>v York sincc January"; however, to
prcdict such an ultcr:>.nce fro1.n cvcry French learner of English is to go too
fat:
1'he most cvnvincing carly criticism of the strong version of the CAH
\Vas offcred by\\/hitni:.u1 and]ackson (1972), who undertook to test empirically the cffecti\'CHCSs of contras\e an~1lysis as a tool for predicting arcas
of difficulty for Japancsc learners of English. The pretlictions of four separare contrastive ana1ysis rubrics (inctuding that of Stockwell, I3owen, &
l'vlartin 1965) \.vere :ipplic. lo a fortrite1n test of English grammar to determine, a priori, the relalivc diJficu1ty of the test items for spcakers of
Japanese. 'fhe test was ~uJnnistered to 2500 Japancse learners of English
v.rho did not know the relative predicted difficulty of each iten1. The results
of the test were cornpared with the predictions. The result: \\?hitn1an and
J~ckson tund no support fr the predictions of the contrastivc analyses so
carefully \Vorked out by linguists! They concluded (p. 40) that "contrastiYe
analysis, as represented by the four analyses tested in this project, is inade4uate, theoretically and practically, to pn.:Jict the interference pfoblen1s of
a language learner."
Another llo\v to the strong version of the CAii was delivere<l by Oller
and Ziahos;,einy (1970), \Vho proposed what one might calla "subtle differences'' Yer;;ion of lhe CAH on the basis of a rather ntriguing study of
speHing errors. Thcy found that for learncrs of Euglish as a second language, English spelling proved to be rnore <lifficult for pcoplc \vhosc r.;:,tive
langua.gc used :.l Ron1an sc.ript (for cxan1ple, French, Spanish) than for thosc
whuse native hi.nguage used a nonRoman script (Ar.1bic, Japanese). The
strong form of the C.i\11 \Voul<l have prc<lcted that thc lcarning of an
entirely ne\Y \Yrit.ing systen1 (Level 4 in the hicrarchy of-difficulty) woulJ
be n1ore difcuil than reinterpreting (Level 3) spelling rules. Oller and
Ziahosseiny (p. 186) found thc oppositc to be true, conciu<ling that "wherever patterns are n1innally d.istinct in tOrm or n1caning in one or more systen1s, confusion inay result."
The learnlng of sounJ;,, sequenccs, an<l meanings vill, according to
Oller and Ziahue:.sciny's stu<ly, bt: potcntialiy very ifl1cult \vhere subtle dis-

C/-"IAPTfR 8

CroS<;Linguisric lnfluenr;e and /.earner Language

213

tinctions a.re required eirher bet\veen the target language and native lan- J
guage or "\Vithin thc target language itself. In the case of thcir research on
spel!ing E11glish, there were more differences betwcen non-Roman vriting
and Ro1nan \\ITiting, but learncrs from a non-Ro1nan writing systen1 ha<l tu
makc fewer subtle <listinctions than <lid those from thc !tornan writing
systen1. Exan1ples of subtlc tlistinctions at the lexical leve! inay be seen in
false cognatcs like the French \VOrd parent, which in the singular means
''relative" or"kin, "\Vhile only the plural (paren~) means "parents." Consider
thc Spanish verb enzbarazar, >vhich con1mo.nly denotes "to make pregnant," and has thcrefore bcen the so1.1rce of true 'en1barrassment" on the
part of beginners atte1npting to speak Spanish! In recent years, research on
CLI has uncovere<l a nun1ber of instances of subtle differences causing
great difficulty (Sjoholrn 1995).

'fhe couclusion that gi,:~at,-:.dift~'.~9-JE~h;t-B-~-~:-_not necessarily cause g.reat

<lifficuJty --underscores the significan,Ce of ffitralingual '-(\.vithin one language) errors (see subsequent sections in this chapter), which are as much
a factor in secontl language learning as interlingual (across two or more
languagCs) errors. The forms within one languitge are often perceived to be
minimally <listinct in con1parison to the vast differences between the native
and target language. yet rhose intralingual factors can leacl to so1ne of the
greatest difficulties.
To<lay we recognize that teachers must certainly guard against a priori
pigeon-holing of learners before u..e have even given learners a chance to
perfonn ..At the san1e tin1e, we must also under.stand that CLI is an important linguistic factor at play in the acquisition of a second language
(Jaszczolt 1995). CLI implies mu ch nlore than simply the effect of one s first
language on a second: the second language also influences the firsc moreover. subsequent languages in rnultilinguals ali affect each other in various
ways. Specialized research on CLJ in the trn1 of contrastive lexicology,
syntax, se1.nantics, and prJ.gn1atics continues to provide insights into SLA
that musr not be discounted (Shanvood-Srnith 1996; Sheen 1996). Sheen
(1996) found. for examp!e. that in an ESL course for speakers of Arabic,
overt attention to targeted synractic contrasts bct\veen A.rabie and English
reduce<l error rates. Indeed, thc strong form of the CAH was too strong, but
the \.veak forn1 was also perhaps too weak. CLI rescarch offers a cautious
middle ground.

MARKEDNESS AND UNIVERSAL GRAMMAR

Fred Eckman (1977, 1981) Prnposed a useful rnetl1od for determining directionality of difflculty. I-:Iis Markedness Differential I-I}'pothesis (othe1-WiSe
kno\vn as 1narkedness thcory) accounted for relative degrees of difficulty

CHAl'Tl'R 8

CrossLinguistic lnOuence ami l.earner l.anguage

CHAPTER 8

by means of principles of universal gramn1ar. Celce-Murcia and Hawkins

(1985: 66) sum up markedness theory:

Eckman (1981) showed that n1arked items in a language \Vill be n1ore diffi*
cult to acquire than unmarked, and that degrees of markedness will correspond to degree.s of di.fficulty. Rutherford (1982) used 1narkedness theory
to explain why there seen1s to be a certain order of acquisition of morphemes in Engiish: marked stnictures are acquire<l later than un1narked
structures. i\tlajor and Faudree (1996) found that the phonological performance of native speakers of Korean Jearning English reflected principies
of n1arkedness universals.
In recent years, the attention of sorne second language researchers has
expanded beyond markedness hypotheses alone to the broader fran1ework
of linguistic universals in general (i'vlajor & Faudree 1996; Eckinan 1991;

Carroll and Meisel 1990; C.;mrie 1990; Gass 1989). Sorne of these argu
menrs focus on the applicability of notions of universal grammar (l.JG)
to second language acquisirion (\'V'hite 1990; Schachter 1988; an1ong
others). As we sav.r in Chapter 2, n1any of the "rules" acquired by children
learning their first language are presumed to be universal. By extension,
rules that are shared by all Ianguages comprise this UG. Such rules ar a set
of limitations or parameters (Flynn 1987) of language. Different languages
set their p:ira1neters differently, thereby creating the characteristic
gr:i;nmar for rh::.t language. The hope is that hy discovering innate linguistic
principies that govern what is possible in human languages, we may be
better able to understand and describe contra~ts between native an<l target
languages and the difficulties encountered by adult second language
learners. Research on UG ha'5 begun to identify snch universal properties
and principies, and therefore represents an avenue of sotne promisc.
Markedness theory and U(; perspectives provide a n1ore sophisticated
understanding of difficulty in Icarning a second language than we had previouslv fron1 the earlv fonnulations of the CA}l, and fit n1ore appropriately
into c~1rrent studies
t-:=LI. But ,ve do well to ren1en1ber d1at describing
and predicting difficulty amidst ail the variables of hun1an learning is still
an elush-e process. Teachers of foreign languages can benefit fron1 UG and

:,r

215

marke<lness research, but even in this hope-filled avenue of research, an

instant map predicting learner difficulties is not rght around the corner.

It distinguishes members of a pair of related f0rn1s or structures

by assuming that the marked member uf a pair contains at least
one more tEature than the unmarked one. In addition, the
unmarked (or neutral) n1cmber of the pair is rhe one with a wider
range of distribution t:han the marked one. For example, in the
case of the English indefinite articles (a andan), an is the n1ore
cotnplex or marked form (it has an additional sound) and a is the
unmarked forro with the wicler distribution.

CrossLinguistic h.{hrnce 2nd Learner Language

~LEARNER

V\

'
1

":f-":;

O""

~e-{

The(.f=f\!:I. ~~.~.e~~y{:ttJ~.~:.h1t~rf~rJJ1,g
eff<~cts": of rhe
,,, ... '
.... '

lan!;ilige Dfi secqncl lan_-

......,. ,,, .....~.,,.,,.,,.... : ..'fa'.'"'""'""""''"'"'''""'

M.:~. ~pe'...:~~~r~ni.n~t and claHned, in its strong form, that Secoi:.l Ianguage
\e.P':
.:i""'"
1'
lear1iTi\g iS -:-f>rinaril). if not e:s.ch1siYely, a proc;ess of acquiri.ng svharever
~ '?~) V )\ o-d5;;.
?o-<&
items ar diffe1'ent from the first langnage. As
notcd above, such a
il
'').?<.;"
~-P'-'v .1.i< ' ..~ v
d" ,,s..,. narrow view of interfcrence lgnoreJ, the intrJ.lingu:<l effccts of learning,
~X:..:-..C-e-"'
Q~<fa .<;~mong other factors. In te. ce.nt years researchers an<l tea.chers have con1e
~
1:-~

O p..:

k""~

LANGUAGE.

\. J

~ cJ".i'> ,"''!> \~

"e-- c.')\; i110~7 ,,and. more to und~rstn~.. f~;~~ ~.;~~~.;::1.~~.1:1?:1a~e.leatning is a .process of

~~~c-1>1'-/'""""' .,~.i--.;:; 0 i the cr~~~~Y~.:,E.~:,1~~~t~!~.t.191~ ,nt:a ~ys~em 111 wh1ch learner.s ..a.r.~ ~ot~.~ciou.sly

,.,,,... ~
0 :/c- #- o:>__, < ~<V testi~g hypor1~~.~~~....ab()t1.t: tli,e.. ta~g~~.. 1.atls.u.a.se. t~~9,.f!~. a ..t1.9.1nP.t;.~ ..9J._pQss.ible
e-&v..~ o-"""'d,;; .~,'> C<Y <fa;:o\e.,, s~~1~ces.,,_qf ln'lowle.~l~.~:. kn~~~.!~~~.~:. 9fr.\1~..n..tive lungnage, u.1!1ft:~...~
d_"'_"..,.. "f'_, v--\,.e-'O'~,...o""- <.<.
e.dge of_the target langua.ge tself'. kno~"!edge .f the corn11111nicatiYe fr1nc-,,. -. ~..cb~~ "'"' e
ttons o~ l:1nguage, k_1~~~}:Y,l.e<lge about l~nguagi;:~.:1~ ge.e.~~!, ancl ~~
-~../
,.. y !(""
about bfe, hun11n be1ngs, and the unrverse. The lerners., in lcting upon
-.=

"'Jc-o-<(.~e..J?f;, ~ ...,,.<'
c,p?

their enYirorunent,

~onstruct

what to ~. ~1em is a

legiti~te

svsteJlLQ.l!!l

$~age in its own right-a structured set of rules th::it fl)r the time being

bring sorne order__ to. t_he linguistic chaos that confronts them.
r"'By the late ',J.96os:'sLA hegan t:o be examined in mucb. the S;ime waY
that~~first language acquisition had been studied for sotne tiine: learner's
were looked on not __a~.Produ.cer~. 9f,_r:1Jfor.rned, L']lperfect iangui"lg~ replete
with ntistakes bur as intelligent and creative beings proceeding through
logical, systen1atic stages of acquisition, creatively acting upon their linguistic environ1nent as they encountered its forms :ind functions 1 mean~
ingful contexts. By a gradual process of trial and error and hypothesis
testing, learners slowly and tediously succeed in estahlishing closer and
closer approxin1ations to the system used by n3.riYe speakers of the Janguage. A number of terms have been coined ro de.,crihe thc perspectlve
that stresses the l~g.itim~.cy o.f 1.e~~:ners' second b.ng11::ige sy:.,tem.s. 1'he best
known ofthese istftlt_~f.l~~.ff~i~j:? tcrm that SelinLcr (1972) :i.dapted from
\Veinreich's. (1953) term winterllnguaL" Interbngu1ge refe1s to the sepa-

r~t.e~1e;~s.. f?f . .~ . s~:~.<?,.11~1 ,1.ang~1:~~~...l?~~n.~t:s ..~r~r.t:~' n syste.ITi thar :.has a struc-.

tui.111y intefmE:dfrttC sta.tri."I bem~e{::n t11e nattve
taxget lrrng1ages;
Ne111ser (1971) referred to.the sanie gene1"'.ll phenomenon in second
language learnlng but.srres~e~ . .r~.,e s_u~c.:s.sive approxi.ni.:1.ton to the target
1~u<1:.g_ein.hister1n\i~ii'PiO.Xillia11V<f system::rnrdcr (1971: is 1) usett rhe~
ten1l .iTio_~ynU'.'.Ll.fiC..:iialCCt' l co1~~1CHe 'i_Et;'"idca th;it rhe lcarncr's tanguagc
is urri:'que to a par1icuLu indh'"idn:1l, th:i.t the rules of tbe lc:1rner's language
are peculiar to the language of that tndi1'idu:li alone. \V11ile each of these

216

CHAPTER 8

CroS':>-Linp,uistic Jnflul.'nce and Learner Langu,1ge

on.OTER 8

(JJ3Sl.inguistic lnluence and Learner LanguJge

217

guistic systen1 \Vithin \vhich children operare, but, by carefully processing

feedback from othcrs, children slo1,.\dy but surely learn to produce \vhat i::;
acceptab!C Speech in their natiYt: !anguage. S~:C0~1.d_ lan.guage ~.Cilfning is., a:
process that is clearly I)Ot unlike first langu.l.ge Jeafning.i;n its trial-an<l-error
nature. Inevitably learners will make mistakes in the process of acquisition.
and that proccss will be npe<led if they do not comtnit errors and then
benefit from various fonns of feedback un those errors.
Researchers an<l teachers of second languages carne to realize that the
mistakes a person 1nade in this process of con.structing a ne"' systen1 of language needed to be analyzed carefully, for they possibly held in them sorne
of the keys to the under.standing of the proccss of second language acquisition.As Corder (1967: 167) noted: "A learncr's errors ... are signil1cant in
[thatl they provi<le to the researcher evidcnce of hO"\V language is learned
or acqui.red, what strategics or procedures the learner is etnploying in the
<liscovery of rhc language."

designations en1phasizes a particular notion. they share thc concept that

secun<l ianguage !earners are fornling their o\vn self-contained linguistic
systcn1s. This is neither the system of the native language nor the systern
uf the target language, but a system based upon thc best atte1npt of learners
to bring on.ler and structure to the linguistic stin1uli surrounding them. The
interlanguage hypothesis led to a \vhole new era of second language
research and te<.i._;lng and presented a significant breakthrough fron1 the
shack!cs of the CAH.
111e n1ost obvious appro;:~ch q.J analyzng intedanguage is t. study tht.~
speech. and \Vriting of iearners, or \vhat has corne to be c::~Hed~l,~~.Pne.:r la.nguag~ (Lightbcn.vn & Sp;1da 1993; C. James 1990). Production data is publiclr observable and is prcsun1ably reflectivc of a lcarner's underlying
con1petence-production competence. that is. Comprchension of a sccond
language is n-iorc diJficulr to study since it is not directly observable and
must be inf\..Tred fn.Jn1 ovcrt verbal and nonverbal responses, by artificial

instrumcnrs, or by thc intuition of the teacher or researcllcr.

It follo,vs thJ.t the study ofthe speech and writihg.oflearners is largely

thc slu<ly of the errors of learners. "Corrcct,. prouction yields little nformation about thc actu;11 linguistic system of learners, only information
about thc t~trgcr languagc system that learners have already acquire<l.
Therefore, uur focus in the rest of this chapter wiU be on the signiJicance
of errors in lcarncrs' dcvcloping systcms, otherwise kno'\vn as error
analysis.

EllliOR f!INALYSIS
H.u1nan,\carnJ.ng is fu110atnentally a.proct::ss that involvcs the making of rnis,
tkes. ;\lbtakes, nli.sjuJgrncnts, nscalculations, and erroneous assun1ptions
fOrn1 an in1portant aspect of learning virtually any skill or acquiring information. You learn to swim by first jumping in to the water and flailing arn1s
and ;.,:gs ...:~1til y.""}u ;:H0covt:r that thcrc is a Lon1binat:'.~;1 of move:nent~-a
structured pattern-that succeeds in kceping you afloat and propclling
you lhrough thc water. The first n1istakes f learning to swin1 are giant
oncs, g.radually Jiminblng as you learn fron1 n1aking thosc mistakes.
Learning to SYVln, to play lcnnis, to type, orto read all involve a process in
\Vhich sut:cess cou1cs hy profiting froin rnistakes, by using mistakes tu
obtain tCcJback from the enYironrncnt, and vvh that feedback to make
new alternpts that successively approxi.rnale <lesire<l goals.
La11guage learning, in this sense, is like any other hurnan learning. We
have already seen in the second chapter that children learning their first
language l~lke countless "nlistakes" fro1n the point of vicw uf ault g:i:am1natical langu:otge. i>Iany of the:;c nlist:akes are logical in the limited lin- --

Mistakes and llfrorsc

'

In order to analyze learner_language in an appropria.(e perspective, it is crucial to n:ake a dj,$1it!CJ:io'n bet"v.eefl. ~J#takes and errors-; technically t>vo
very different pht:no1nena. A mistake refers to a performance error that is
either a ~ guess or a "slip:' in that it is a failure to utilize a knovn
syste1n correctly. Ali peopie make inistakes. in both nativc and .second language situations . . Native. speakt:rs are norn1ally capable of recognizing a!'1d.
correcting such "lapses"' or nlistakcs, which are not the result of a deficiency in con1petence but the"result uf sorre sort oftempor.ify breakdown
or in1perfection in the process of pro<lucing speech. 1'hcse hesitations.
slips of the tongue, random ungrammaticalities, and other perfonnance
lapses in native-speaker production also occur in second language speech.
l\'listakes, when attention is called to then1. can be self-corrected.
J.Vlistakes 1nust be carefully distinguished from errors of a second language l:::;:::T,e:-. :dic-~:-':1.:::-.1.si.::s ;" rhe langi..:age cf thc learner that are' clirccr
n1anil'estations of a systen1 ~'ithin which a learner is opcrating at thc tin1e.
An eiTOt~ ~t noticeahle deviation fron1 the a<lult granunar of a na ti ve speaker.
reflects the co1npetcnce of the learner. Learners of English who ask, "Does
John can sing?" are in all likelihood reflecting a competet1<.:e level in which
all verbs require a pre-poseJ du auxiliary for question formation. As such.
it is an error. most like!y not a rnistake, and an error that reveals a portion
of the learner's comperence in the target language.
Can you tell the difference between an error and a nlistake? Not
ahvays. An error cannot be self-corrccte<l, according to Jame~ (1998: S:;).
Vi.-'hile 1nistakes can be self-correcte<l Lf the deviation is pointed out to the
speaker. But the learner's cap~tciry f n self-correction ~s objectively observ-

CHAl'TER 8

Cross-Linguisric lnffuenu' and Learncr Language

able only if the lcarner actually self-corrects; therefore, if no su ch self..

correction occurs, V\'e are srill left with no n1eans to identify error vs. n1is
takc. So, can we turn to freqllency of a deviant fonn as a criterion?
Sometimes. lf, on one or t\VO occasions, an English learner say::; "John cans
sing," but on other occasions says "John can sing," it is difficult to deternne
'vhcther 'cans" is a 1nistakc oran error. lf, howcver, further exannarion of
the earner's speech consistently reveals such utterances as 'John vvills go,"
'John mays come," and so forth, w-ith very few instances of correct thin.1person singular usage of modal auxiliaries, you might safely conclude that
'cans," "1nays," and other such for1ns are errors indicating that the learner
has not distinguishetl modals fro1n other verbs. But it is possible, hecause
of the few correct instances of production of this form, that the learner is
on the verge of making the necessary differentiation between the two
types of verbs. You can thus appreciate the suhjectivity of deternning the
difference bet\veen a mistake and an error in learner speech. That undertaking alw;iys bears with it the chance of a faulty assun1ption on the part
of a teachcr or researcher.
The fact that learners do 1nake errors, and that these errors can be
observedj analyzed; and classified to reveal something ofthe-systetn operating within the learner, led to a surge of study f learners' errors 1 called
error analysis._ Error analysis became distinguished fro1n contrastive
analysis by its examination of errors attributable to all possible sources, not
just those resulting from negative transfer of the native language. Error
analysis easily superseded contrastive analysis, as we dbcovered that only
son1e of the errors a learner n1akcs are attributahle to the inother tongue;
that learners do not actually n1ake all the errors that contrastive analysis
predicted they should, and that learners from disparate language backgrounds tend to make snlar errors in learning one target language.
Errors-overt manifestations of learners' systems~arise from severa! possible general sources: interlingual errors of interference from the natlve Ianguage, intralingual _errors within the target language, the sociolinguistic
co1ext o.f commurcatiOT'!., psycholinguistic or cognitive srrategies,:ind no
doubt countless affective variables.

rors in Error Analysis

There is a danger in too 1nuch attention to learncrs' errors. \X!hile errors
indeed reveal a system at work, the classrooin language teacher can
becon1e so preoccupied with noticing errors that the correct utterances in
the second language go unnoticed. In our observation and analysis of
errors-for all that thcy do revea! about the learne1--wc n1ust be\.varc of

CHM'TER 8

CrossLinguistic lnfluence and Learner Language

219

placing too n1uch attention on errors and not lose sight of the value of positive rcinforcernent of clear, free cornmunication. \'< hlle the diminishing of
errors is an in1portant criterion for increasing language proficiency, the u!timate goal of secnnd language learning is the attainment of communicatiYe
fluency.
Another shortcon1ing in error analysis is_ an overemphasis on production data. ~1' and .listen in , -.;;vritln rrnd reaclin . The
cnmprehc'nsion of language is as important as productlon. lt so happens
that production !ends itself to analysis and _thus becomes the prey of
researchers, but comprehension data is equaHy important in developing an
understanding of the process of SI~.\.
()ver the years, n1any studics Qames 1998;Tarone 1981; Kleinmann
1977; SchJchter 1974) have shown that error ;:nalysis fails to accolint for
the strategy of 3Voidance. A learner 'l;:vho .tOr onc reason or another avoids
a particular sound, vvord, structure, or discourse caregory may be assun1ed
incorrecrly to have no diffit:v1ry therewith.- Sch1chtcr (lt)/1) found, for
exan1ple, that it was 1nfslcacling to draw conclusions ahout relative clause
errors among cert3in English learners; native J1par,ese speak,".'.rs were
largely avoiding that structure and thus nor manifesting nearly as many
errors as sorne native Persi:in speakers. The absence of error therefore 'Joes
not n,ecessaily reflect nativelike com_petence, because tearners n1ay be
:ivoiding-th~ very structures that pose difficulty for them.
Finally, error analysis can keep us too closely focused on specific languages rather than vie\\"ing universal aspects of langu:ige. Gass (1989) recon1mended that. researchers pay more attention to Hnguistic elements that
are com1non to ali languages. The language systen1s of learners n1ay ha.:e
elements that reflect neither the rarget l1nguage nor the native language.
but rather a universal feature of sorne kind. Such assertions are in keeping
\.Vith the bioprogramming theories referred to in Chapter 2. But there are
problen1s, of course, with the search for universal propcrties of learners
errors. 'It is not at all clear in any precise \vay v..Then the influence of the
universal \Vill appear in the intert1ngu::ige oflearners r:;ther than a violation
uf it based on influence from either the source or targer l:i.ngt1age" (Celcer.Iurcia & H.awkins 1985: 66).
\Ve do \Vell, therefore, in the analysis of learners' errors, to engage in
'performance analysis" or'interlanguage analysis" (Ce!ce-i\lurcia & Hawkin.s
1985: 64), a less restrictive concept that places a healthy invesgation of
errors ''-:itllin the larger perspective of the learner's total ianguage per~
forman ce. \Vh!le a significant portion of this chapter deals with error
analysis. lct us nevertheless rc1nember that production errors are only' a
sub:;(:t of the overall performance of the lean1er.
1

220

UW'TER 8

Cms.dinguistic lnflucncc und leamcc Language

Identifying and Describing Errors

N

()ne of the comrnon_~i1Tis:_t__tJS, in....~L1l9.S.f~-t~_i?:fgllg_ !!:!_~_J!!l&~:~!A~""~,}~~-~,f,I-l,~ of

both first~;;L~.S.'2Ils"Ll~Db!:Bg~J&1!.r!1~ _is thc fact that sus!:_ _zyste1ns si:.nn_q~~

so

l?:t:_dire_ctly obst.:rv:ctJ. They must be infCr~ed_ by n1eans of a1~~_lly_zi9g__ pro. .YEJ}S.~E}!!J.~Lt;;S~~-H2!S:J~S::_l_l!9_1_1__4'!-_t~!!..y;1-i1at n1akes the task cvcn thornier is the
instability of lean1crs' .':>ystems. Systems are in a constant state of flux as
new infOFn1ation fl!.J\V::. in an<l, thruugh the process of subsun1ption, causes
existing structures to be revised. Repeate<l observations of a learner will
often reveal apparcnlly unpre<l_ictabk: or even contra<lctory data. In undertaking the task of pcrfunnance analysis, the teacher and researcher are
callcd upon to it1fer ordtr and logic in this uns_table and variable system.
1'he first ~~ in thc process -oJ--~nalysis,: i~ the "identfic;ttio_n'. ai1<l
<lt:scription of-errors: Cord~r (1971) proYidcd a model f9r i,de11ti1:'ying-er~o-

,_

"'
~

1'
o
'o!
e

1i;~ll~_?r _i("i~yti~~a_ti_t:. . ~~t~_~rance_~ _i_1! _ a___s~co_t~d :]'~~~g~1aiie. ~1~'i~~~t,. ~~<l~i- is

sC1J.Cff1JtiZeJ

-3~1: A~-;;rJ1g t~ Corder's -n1~~('C1, '-~~~I-""~ie_~!e_n~~-

_Q

t1tte_r_ed by- the learne-r an~ _ su ~-~cq~_eny _ -~!:J:!1:~5~EL~-~-9:..~~'l~~---l?,:~_:_:,!~na~~$..f!2F

ifsy;ncra.sies. _\ 11Jaj0:r_d:1tinCiiofii;, n1ade- at the oulsct bertvccn oy(!_Q:-(.lnd

111at_i_c~tl _-: _t_he:_5::nt~11~~-te~et :<~overtly -__e_r_~()ncous -~itterJ nces aJ?_e:;-g~~6:t_tllati~

~~~lifif1:~~~-Bi~i-~~~~~~.-J~Y~I~i~~-~E$Jl2--fu~~Et?X~!~~~thin the::c2Ji!sd'tt-:Q:f :.'.CQill~,fi!!liJfl.:-'C<:;i_'.t.'.~E~.,.~-~~~t~, in otl1er \Yord.s, are,,.~o~,- ~~-~!-~Y,,
covert at all if you attcnd tu surrqu1u.ling discourse (befare _or afte+ the
uttcranc_e): _, ,___ fill~:;-a;~Ulk }~ 1'. _iS~ifitl~1tiCU11y~COffeC~t~a~_!i.~~=sei~~li~~-;,
ie'\;~f:?iy as-:i respOiiSC"tQ'(\\Vh.-o' ar~ );)ll?--;,it ,5on~i0us1y_a11 err_or. ;A simpler

E
e
..2

-in -1-:gu:~

e
Q

__Sove~-_-etr?r~~ _Q:\~rtly .erro_~-~-S'.-~-~!-?,,,,-~_1_~~-~~-~a_1!_c:es_ 1EfP'<~L~!ru:!~ll;?nl~ly~~dm~_

e
"'

"

oc
e

E
u
e
e

a,1J,ll1Z)re;;;:~:agr;~for~~:;~~zr;~~s, then: would-be:5~D.-t~nc~ level" anti

s;
e

"discourse lcvel" errors.

Corder's modcl in Figure 8.1 indicates that, in thc case of both overt
an<l-covert errors. if a p!au&ibk: interpretation can be n1:-1de of thes-eritcnce,
{j~~n o-ii-C"'SE'O-id fonJ1 a re;construction of the sentence in the t_~Eg~_t _ tan:
guag(::_,___con1r_~re the reC<)_~~~tT_l:l_~!i_Jn_ --~-~~~1- __t_!?-~-- origif?.al i<liosyncrati~ sentcnce, and then describe the differences. lf the native language of the
-~j~:\fef- fS-f~~n;--tl"ie--n1--)JCf}fi-;;CJies lL~ng translation :.ts a possible indicator of native lJ.nguotge fntcrference as the source of error. In sorne cases,
of course, no plausible intcrpretation is possible at aH, anti the re~earcher is
lel't: with no analy:>is of the error (()U1'3).
Consi<ler the follo'i-ving exarnples of idiosyncratic utterances of
learners, and let us allov.r them to be fed through Corder's procedure for
error ana!ysis:

"

..;!

--+-

o
o

.,,,

o
e
"'
k
eo

g'

,5

:o

ilo
V

"-

oI

di
2

o
:

""

221

_22

CHAPTER 8

Cross-Linguistic

lnfluence and Lrarner

WAPTER [!

Langudge'

1.; Does John can sing?

A.

NO
C. YES
l). Can John sing?
E. ()riginal sentcnce contained pre-posed do auxiliary applicable
to n1ost verbs, hut not to verbs vvith n1o<lal auxilia.des. OU'f2

2. I sa\v their department.

A. YES
B. NO ( context vv:t<; in a conversation ahour living quarters in
i\lexico)
C. NO
F YES, Spani::;h.
G. yo vi su cleparta1nento.YES
IL I saw their apartment.
E. Deparunento \vas translatecl to false cognate dejJart1nent.
OUTz

3. The different city is another one in the another rwo.

A. NO
C. NO

F. YES, Spanish.
G. No plausible transL"ltion or interpretation.
I. No analysis. ()lJT 3
It can be secn that thc model is not con1plicated and represents a proceUure that teachers and researchers n1ight intuitively follo\V. Of course, 9.n.s:s;
!lIJ~_.tfJl!:i~Lhl_~J,,-Uitit:s.L.JJA.~,,mJi;;t~~~C-!2"'.~<k--'~J:ibe__j_t_~9Sl1lii~', something
the aboye proc<lure has only begun to accon1plish.
A numher of different categories for description of errors have been
identified in research on learner language (for an overvie\\r, see Lennon
199 l).
l. The n1ost generalized breakdown car1-he 1nade by identifying
crrors of{idditioij)otnission, s,~1_L"?.~~!Y_i!on, and orderin~, .foll'.)W!ng
standard mathe1natical categories. In English a do aux1l1ary m1ght
be added (Does cau be sng?), a definire article 01nitted (J ruent
to 1novie), an ite111 substituted (! lost 111)-' road), ora word order
confused (I to t!Je store iuent). But such categories are clcarly

very generalized.
2. \Vithin each category, feuels of language can be considered:
phono!qgy or orthography, lexicon, gran1n1ar, and disconr.se.

Ccno>~Lingui;tic lnf/uence 'nd teamu; L.o.ngu.oge

223

Often. of course, iL is difflcult to disting11ish different levels of

errors. A word with a faulty pronunciation, for exan1ple, might
bide a syntactic or !exical error. A French learner \.Vho says
"[zhey} suis all a l'cole" might be mispronouncing the grammatically correcr "je;' or corrcctly pronouncing a grammatically incorrect "j'a."
3. Eri:()r.s may also be vie\ved as either gl(fl'!-P] or fr.>Cttl "(Burt &
Kiparsky 1972). G10bal errors hindei','"ori1fnttnlcatio11; rhey prevent the hearer fron1 cmprehen<li:ng sorne aspect of the n1essage. for exarnple, "Well, it's a great hUrrr J.ronnd," in whatever
context, n1ay be difficult or impo~sible to interpret. Local errors
do not prevent the message fro1n being_ heard 1 u~nally because
there is only a minor violnt!on of one segrn~ent of a sentence,
alJov::t-ing the hearer/reader_to make an accurate guess about tbe
inte.11de<l 111eaning. '"A scissors," for eran1ple, is a local error. The
global-local distinction is discussed in the vignette at the end of
this chapter.
4. Finally, I.(~nnon_ (1991,)_ s1,1_g,gests that two related dimen.sions of
error, dornain
:1od e,xteut
:shonld-be cor1$idered in anv' error
--"'>C.,'"w'
...:,;_.;,
analysis-, EJ_On?_ain,is the_ rank: of:linguistic-unit (frnm phoneme to
discoursc}tht must__ be -1:-;tken as contcxt ln order for the error to
become apparent, and extent is the rank of_linguistic unit that
\vould have to be deietecL replaced, suppHecl, or reordered in
order to repair the sentence. Lennon's C:1tegories help to opera-

tionalize Corder's overt-covert distinction discussed abovc. So, in

the ex:imple just cited above, "a ~cissors,'" the don1;iin is the
phrase, and the extent is tbe inclefinite article.

Sotlrces of Error
H~ving exan1ined procedures of error analysi.s used to identify errors in
second language !.~~trner production data, tJur filal step in the a.nal:ysis of
erron.eous learner speech is that of deterrnlning the source of error. Wli}',
are c'rtain e1:ro1\S rnde? \\l1at cognitve strategies and styles or even Personality variables underlie certain errors? \'\'hile the answers to these q_uestion.;; are son1e'.vhat speculariYe in that sources 1nust be inferred from
avai!ahle data, in such questions lies the ultimare value of learner language
:111alysb in general. By trying to i<lentify sources '\ve can take another step
tO"'.-V<trd understanding ho\v the learner's cognitive and affective processes
relate to the linguistc system and to trmulatc an lnregratcd ur:.derstanding
of the process of second language ;H..:t1ubitlon.

224

O-IAPTfR

CrossLinguistic lnfluence and Learner Language

lttterllngual 1'ransfer
As we ha ve alreaJy ~een, interlingual transfer is a signlcant soun.:e of error
for all learners. 'fhe beginning stages of learning a second languagc are
especially vulncrabie to interlingual transfer fron1 the native language, or
interference. In these early stages, before the-systen1 of the second-1anguage
is fan-Ullar, the native language is the only previous linguistic systen1 upoh
\vbich the lt:arner can draw. We have aH heanl English l~arners say "sheep'~
for 'ship,'' or "the hook f Jak" instead of "Jack's Oook"; French learners
may say"Je sais Jean" for"je connais Jean,'' an<l so forth. AH these errors are
attrlbutuble to negative inter!ingual tr-an:Sfr. \\._.hile it is not ahvays c!ear
that an error is the result of transfer from the native language, many such
errors are detectable in learncr speech. Fluent kno"\vledgc or even familiarity with a leanH.:r's native language of course ai<ls the teacher in
detecting and analyzing such errors.
The -learning of a thi:rd lnguage (and subsequent languages) provi<les an interesting context for research. Depending upon a number of factors, including the lnguistic an<l cultural relatedness of the languages and
the context of learning, there are varying degrees of interlingual interference frorn both the first anU second Janguage to the third language, especially if the second and third languages are cluscly rclatcd or the learner is
atteinpting a third language shortly after beginning a second language.

Intralitigu_al Transj'er
Onc of the n1ajor contrlbutions of learner language research has been its

recognition of sources of error that extend beyond interlingual errors in

learning a second language. It is now clear that intrilingual transfer (Withn
the target ianguage itsel) is a n1ajor factor in second language learning.- In
Chapter 4 we di,-,cusse<l overgeneralization, which is the negalive coun~er
part of intralingual transfer. Researchers (see )aszczolt 1995;Taylor 1975)
have found that the early stages of language learning are characterizccl by
a pi"edoruinance of interf"."rence (intf'.'.r!ngu;:i 1 tra!1sfer). lT!.lt r,r;.c~ i.~n.rners
have begun to acquire parts of the nev.- system, more and more intralingual
transfer-genera.lization within the target language-is n1anifeste<l.1'his of
coursc foilows logically fi-om the tenets of Jearning theory. As learners
progress in the sccond language, ther previous experience and their
e,""'{isting subsun1ers begin to iriclude structures within the target language
itsclf.
Ncgative intralingual transfer, or overgenerJ.lization, has already been
illustrate<l in such utterances as "Does john can si11g?" Other exan1ples
abound-utterances like "He gocd,""I don't know what tin1e is it," and "Il a
ton1b." Once again, the teacher or researcher cannot ahvays he certain of

Cl-IAl'TER 8

Cross-Linguistic Jnfluence and Le;Jrner Language

225

the source of an apparent intralingual error, but repeated systematic obser~

vations of a learner's speech data \Viil often rcmove the ambiguity of a
single observation of an error.
1~he analysis of intralingual errors in a corpus of production elata can
bccome quite complex. For examp!e, in BarryTaylor's (1975: 95) analysis of
English sentences produced by ESL learners, just the class of errors in producing the n1ain verb follo\ving an auxiliary yielcled nine different types of
error:

Tab!e B.. Typic:i! English intra!ingual error;; in the use of articles (rom Richards
1971' 187)

1. Omission of THE
(a) befare unique nauns

(b} befare nouns of natonality

(e) before nouns made partclar in context

(d) befare a noun madified by a participle

(e) befare super!atives
(f) befare a noun modified by an of-phrase

2. THE l'Jsed lnstead of 0

(al before proper names
(b) befare abstract nouns
(e) befare nouns behaving like
abstract nouns
(d) befare plural nouns
(e} before sorne

3. A Used lnstead of THE

(a) before superlatives
(b1 befare unique nouns

Sun is very hot

Himalayas are .
Spaniards and A.rabs .
At the conclusion o article
She goes to bazaar every day
She is mother of that boy
Solution given in this article
Richest p~rson
lnstitute of Nuclear Physics

TI 1e Shakespear!f, the Sunda

The friendship, r:-ie nature,
the science
After the school, after the breakfast
The complex structures are still
developing
The sorne knovv!edge

a \VOrst, a best hoy in the class

a sun becomes red

4. A lnstead of 0
(a) befare a plural noun qualied by
an adjective
(bi befare uncountables
(e\ befare an adjective

5. Omission of A
beore class nouns defined by adjectives

a holy places, a human beings,

abad nevs
a gold, a 1vork
... taken as a definite

he was good boy

he \Vas brave mtln

'!6

C11APTER 8

C!--IAPTER 8

Cross"Unguistic lntfuence and Lcarner Language

1. Past-tense forn1 of verb following a modal

2. Present-tense -son a vcrh follov.-ing a tnodal
3. -ing on a verb follo'\ving a n1oda!

4. are (for be) following zvill

5. Past-tense forn1 of verb following do
6. Present-tense -s on a verb follo'>ving dn
7. -ing on a verb following do
8. Past-tense form of a verb folio'>ving be (inserted to replace a
modal or do)

Cross-Linguistic

rr:+'f::rr-ce

and LParnt'r L>nguage

227

ali e:x:perienced oreign learners whose"bnnkish"langn:ige gives the1n away

as classroom Ianguage learners.
The sociolinguistic context of natural, untutored Lingu:ige acquisition
can give rise to certain dialect acquisition that may itsel.f be a source of
error. Corder's rerm "idiosyncratic <lialect" applies especially well here. For
exarnple, a japanese in1migrant who livect in a predn1ninantly ;v1exicanAn1erican area of a U.S. city produced a learner Linguage that Vias an interesting blend of I\ilexican-1\n1erican English and the staodard English to
vvhich he '\vas exposed in the university, colored by his Japanese accent.

9. Present-tense -son a verb following be (insertecl to replace a

modal or do)
And of course these are linted to the particular data that Taylor was analyzing and are therefore not exhaustive within a gran1matical category.
Moreover, they pertain only to errors of overgeneralization, excluding
another long list of categories of errors that he found attributahle to interlingual transfer. Similarly,Jack C. Richards (1971: 185-187) provided a list of
typical English intralingual errors in the use of articles (see Table 8.1 on
page 125). These are not exhaustive either, but are examples of so me of the
errors coinmonly encountered in English learners from disparate native
language backgrounds. Both 'l"'aylor's and Richards s lists are restricted to
English, but clearly ther counterparts exist in other languages.

Co-,nrnuncation Strategies
In (:hapter 5, communication strategies v,rere ctefined and related to
learning styles. Learners obviously use procluction st:~..ttegies in order to
enhance getting their messages across, but at tin1es these techniques can
themselves becon1e a source of error. ()ncc an ESL Iearner said, "Let us
work for the well done of our country." Wlle it exhibited a nice little mist
of hunHJ_r, the sentence had an incorrect approximation of the word 1cel
/are. Llke\vise, word coinage, circum.!ocution, false cognates (fro1n Tarone
1981), and prefabricated patterns can all be sources of error.

STAGES OF l.FARNER lAiNGUAGE DEVELOPMENT

A third major source of _error, although it overlaps both rypes of transfer, is

the context of learning. "Context" refers, for example, to the classroon1
witl~ its teacher anc;i its materials in the case of school learning or the social
situation in the case of untutored second language learning. In a cla~sroom
context the teacher or the textbook can lead the learner to make faulty
hypotheses about the language, what Richards (1971) called "f<:~~~----~-9.1}
cep1s" ant..i what St..:'.'.nson (197 4) termed "induced errors." Students often
errors because of a misleading explanation from the teacher, t1ulty
presentation of a structure or word in a textbook, or even because of a pattern that was rotely men1orized in a drill but improperly contextualize<l.
Two vocabulary items presented contiguously-for t'"xample,point at and
point out-might in later recali be confused sin1ply because of the contiguity of presentation. Ora teacher may provide incorrect inforn1ation-not
an uncommon occurrence-by \vay of a misleading definition, word, or
gramn1atical generalization. Another manifestation of language learned in
classrooin contexts is the occasional tendency on the part of learners to
givc uncontrJ.cted and inappropriately formal forn1s of !anguage. We have

n;;ke-

There are n1any different ways to describe the progression of learners' lin
guistic development as rheir attempts at production successively approxi1nate the target language system. Incleed, learners are so variable in their
acquisition of a second language that stages of developn1ent defy description. Borrowing soine insights from an earlier model proposed by Corder
(1973), I have found it useful to think in terms of four stages, based on
observations ofwhat the learner <loes in tenns of errors alone.
,~t~ge of r_3:~_do1n errors, a srage that Corder called
"presystematic," in \Vhich the learner is only vaguely 'vare that
thCre is son1e systematic order ro a particular class of iterns, The
\Vritten utterance "The different ciry is another one in the
another two'' surely comes out of a random error srage in which
the Iearner is making r.-nher wild guesses at what to '\vrite.
Inconsistencies-~fike"jhn cans sing,""John can to sing; :lnd "John
can singing," all said by the same learner within a shnrt period of
time, might indicare a stage of experitnentation and inaccurare
guessing.
2. The second, or emergent, stage of learner language finds the

l. The first is a

228

CHAPffW. 8

Cross-Linguistic lnflucnu_' Jnd Lea111er Lmguage

lcarner growing in consistency in linguistic production. The

learner has bcgun to discern a systc1n anJ to internalizc certain
rules. These rules may not be correct by target language standar<ls, but thcy are nevertheless legitimate in thc nnd of the
learner. This stage is characterzed by sorne "backsliding,' in
v.hich the learner seems to havc grasped a rule or principie and
then regresses to sorne previous stage. In general the learner is
still, at thi~ stage, unable to correct errors \vhen they are pointed
out by sorneone else. Avoidance of structures and tapies is typicaL Considcr thc folkrv..-ing conversation bctween a lcarncr (L)
an<l a native spt:aker (NS) of English:
L:
1 go New York.
NS: You're going to NewYork?
l.:
[<loesn't un<lcrstand] \Vhat?
NS: You will go to NewYork?
L:
Yes.
NS: When?
L:
1972.
NS: Oh, you went to Ne\i.rYork n 1972_
L:
Yes.lgo 1972.

Su ch a conversation is reminiscent of those mentioned in Chapter 2 wherc

children in first Ianguage situations could not discern any error in their
speech.

3. A thir<l stage is a truly systematic stage in which the lcarner is

now able to n1anifest more consistency in producing the second
language. \\'hile those rules that are stored in the learncrs brain
are still not atl well~formed, they are n1ore internally_self.-consistent and, of course, they more closely approxin1ate the target language systcm. The most salient difference between the second
an<l thir<l stt:gc is the ability of icarrh.::ls tu ,.;011ecr thcir errors
when they are pointed out-evcn very subtly-to thetn. Consider
the English learnt.'r "rho described a popuLtr fishing-resort area.

L:

\lany fish are in the lake. These fish are serving in

the rt:;laur.ants near the lake.
NS: [laughingJ The fisb are serving?
L:
(laughing] Oh, no, the fish are serued in the
restaurants!
4. A final stage, which I will call the stabilization stage. in the

CHAP7ER 8

Cross-Linguistic lnfluence and /.earner Language

229

development of learner language systems is akin to what Corder

(1973) called a "postsystematic" stage. Here the learner has relatively fe\V crrors and has mastered the system to the point that
fluency and intended n1eanings are not problematic. This fourth
stage ts characterized by the learner's ability to self-correct. Thc
syste111 is complete enough that attention can be paid to those
few errors that occur an<l corrections be n1ade without \Vaiting
for feedback from someone clse. At this point learners can stabilize too fast, allo\ving minor errors to slip by undetected, and thus
manifcst fossilization of their language, a concept that will be
tlefined and discussed later in this chapter (see Selinker and
Lamendel1a 1979).
lt should be made clear that the four stages of systernaticity outlined
abovc do not describe a learner's total second language systeni. We "\Vould
find it hard to assert, for exa1npie, that a learner is in an e1nergent stage,
globally, for ali of the linguistic subsystems of language. One might be in a
second stagc "\Vith respect to, say, the perfect tense systcm, and in the third
or furth stage when it comes to simple present and past tenses. Nor do
these stages, which are bascd on error analysis, adequately account for sociolinguistic, functional, pragmatic (see Kasper 1998), or nonverbal strategies, ali of which are important in assessing the total competence of the
second language learner. Finally, '\V'C need to remember that production
errors alone are inadequare measures of overall competence. They happen
to be salient features of second language learners' interlanguage and
present us "\Vith grist for error-analysis nlills, but corrcct utterances \varrant
our attention and, especially in the teaching-learning process, deserve positive reinforcetnent.

VARIABIDTY Il'l LEARNER LANGUAGE

Lest you be tempted to assume that all learner language is orderly and systematic, a caT.-eat is in ordcr. A great deal of ::i.ttcntior:. !1J..S bccn giYen to th-::
variablity of interlanguage development (Baylcy & Preston 1996: James
1990; Tarone 1988; Ellis 1987; Littlewood 1981).Just as native speakers of
a language vacillate betv:een e..xpressions like "It has to be you" and "It must
be you," learners also exhibit variation, sometimes within the parameters of
acceptable norms, somerimes not. Sorne variability in learner language can
be explained by what Gatbonton (1983) described as the "gradual diffusion" of incorrcct forms of language in en1ergent and systematic stages of
development. First, incorrect forms coexist \Vith correct; then, the in corre et
are expunged. Context has also been identified as a source of variation. In
classrooms, the type of task can affect variation (I'arone & Parrish 1988).

n-1APTFR 8

Cross"Linguistic lnl!ucnce and l.carner Languae

CHAl'TER 8

And variability can be affected, in both tutored and untutored learning, by the
exposure that a learner gets to norn1s.
\Vhile one snply must expect a good proportion o ltarner language
data to fall beyond our capacity for systen1atic categ.orization, one of the
inore controversial current debates in SLA theory centers on the exrent to
which variability can indeed be systematically explained. The essence of
the problem is that learners can and do exhibit a tren1endous degree of
variation in the way they speak (and write) secon<l languages. Is that variation predictable? Can \VC explain it? ()r do we dismiss t aU as 'free variation"?
Notable an1ong mo<lels ofvariability are ElajneTarone's (1988) capability continuu1n paradigm and Rod Ellis"s (1994, 1986) variable competence modcl, both of \Vhich havc inspire<l others to carry out research
on the issue (see Foster & Skehan 1996; Bayley & Preston 1996; Preston
1996; Crookes 1989; Adamson 1988; Young 1988; for <Xample).
Tarone (1988) granted that non-systematic free variation and individual variation do indeed exist, but chose to focus her research on contextual variability, that is, the extent to which both linguistic and situational
contexts may help to systematically describe \Vhat 1uight otherwise appear
simply as unexplained variation. 'farone suggested four categories of variation:

231

Rod Ellis (1994b, 1986) has drawn a n1nre "interna!" picture of the
learner in his variahle competence model. f)r:l\Ving on Bialystok"s (1978)
earlier work, Ellis hypothesized a storehouse of "":iriahle interlanguage
rules" (p. 269) depending on ho~ aurom:i.tic and how analyzed the rules
are. He drew a sharp distinction bet~!een planned and unpbnned discourse in order to examine variation. The fonner impHes leo;s auton1acity,
and therefore requires the learner to cal! upon a certatn. c~itegory of learner
Janguage niles, while the latter, more autom:nic production, predisposes
the learner to dip into another set of rules.
Both models garoered criticism. Gregg (1990) quarreled with bot.h
Tarone's and Ellis's rejection of Chomsky's "hc~mogeneous competence par
a<lign1" (see the discussion in Chapter 2 of this book :ibont competen ce and
performance). "\X.'hy should the fact that a learner's competence changes
over time lead us to reject the standard concept of co1upetencc?" argue<l
Gregg (1990: 367). lt V.-'ould appear fron1 Ellis's argun1ents that Clion1sh.y's
"performance variables" n1ay be better thought of as part of one's 'variable
con1petence" and therefore not attributable to mere "slips" in perfonnance.
Such arguments and counter-argumenrs (see responses to Gregg by Ellis
1990a andTarone 1990) will conti.nue, but one lesson we are learning in all
this is apparent: even the tiniest of the bits and pieces of learner language.
howeVer random or"variable" they may appear to be at first blnsh, cuuld be
quite "systetnatic" if we only keep on looking. It is often te111pting as a
teacher oras a researcher to dismiss a good deal of learners' production as a
tnystery beyond our capacity to explain. Short of engaging in an absurcl gan1e
of straining at gnats, we must guard against yielding to that temptation.

l. variation according to linguistlc context

2. variation according to psychological processing factors
3. variation according to social context
4. variation according to language function

The en1phasis on context led us to look carefully at the conditions

under which certain linguistic forn1s vary. For example, suppose a learner
at one point in time says (a) "He must paid for the insurance," and at another
tin1e says (b) "I-le must pay the parking fee." An exan1ination of the linguistic (and conceptual) context (the first of Tarone's c:Hegories) nlight
explain the variation. In this case, sentence (a) was uttered in the context
of describing an event in t~1e past, and sentencc (b) rcferr::d to the prese::ir
1non1ent. Thus the apparent free variation of the main verb t()rn1 in a inodal
auxiliary context is explained.
One of the moSL fruir.ful arcas of learner language research has focused
on the variation that arises from the disparity b(~CTveen classroon-z contexts
and natural situations outsi<le language classes._\S re.'.->earchers have cxamined instructed second language acquisition (Ellis 1990b, 1997; f)oughty
1991; Buczowska & Weist 1991), it has become apparent not only thar
instruction 1uakes a dil11:rence in learners' success rates but also that the
classroorn context itself explains a great deal of variability in lean1ers'
outpul.

Cmss~Linguistic lnf!uence and Leamec Language

,f

D5

::i \

a.x..

'

'

ze.1-....) 011

'
FOSSILIZATION;
It is quite corrunon to encounter in a learners language various erroneous
features that persist despite what is othenvise a reasonably f1uent con1mand of the b.ng11age. This phenomenon is most salientlv n1aslifested
phonOlogically in "foreign accenrs in the_ speech of n1any of those \Vho
have _learned a second language after puberry, as v..-'e saw in Chapter .3. We
also frequently-observe syntactic and ,Jexical errors persisting in the speech
of tho.se who have learned a \anguage quite welL The reiarJvc1y pern1ane11t
inc_orpration finc9rrect lingui_stic fpr111s Jnto_a ,person's secnd 1ngage
compete11ce hus been referred to as:fsslli.zad.0-n. fos.sili1;1tion is a normal
and natnr:il stage for many learners, and shon!d not be vie;,ved as son1e sort
of terminal illness. in spite of the forbidding nletaphor that suggests an
unchange:-ible situ:tt1on etched in s1one. _\ better n1et;1phor rnight be sotnething like 'cryogenation" -the process of frcezing 1natter ar very lo\Y tc-n1-

232

( idi'Ti'R

Cros5-Lnguistic lnfluence and Lcarne1 Langu.-ige

peratures; we woul<l then have a picture of a situation that could be

reversed (given sorne '\\\lnnth, of coursel).
llow do itcms bccun1c fo.ssilizcd? Fossili'.l.ation can be seen as consistcnt with principles of hun1an learning already discussed in thiS book: conditioning, renforce1nent, need, rnotivation, SCif-deternnation, and othcrs.
Vii?H and OUer \ 1976) pro\-lJed a for111al account of 08silization as a factot
of poSitiVe an<l negative affectJve and cognitive feedback. They noted that
there are nvo kinds of infbrination tt'J.11sn1ittetl benveen sourccs (learners)
and audiences (in this case, native speakers): infr1nation about the ajfectiue relationship between source and ~iuJiece, an<l cognitiue infor1nation-facts, suppositions, beliefs. Affective inforn1ation is prirarily
encoded in tcrn1s of kincsic n1echan.isms such as gestures, tone of VoiCe,
and facal expre~:sions, \Vllile cognitive -irtfrmation is usually co-nveyed by
n11,;'ans of linguistic <levices- (sounds, phrases, structures, discourse). The
feedback learners get frorn thcir audience can be ether positivc, neutral,
soinewhere in between, or negutiVe. The two types and levels of feedback
are charted belovv:

Aff(!.ctive Feedback:
Positive:
Neutral:
Negative:

Keep talking; I'm listening.

I'n1 not sure I \Vant to n1aintain this conversation.
This convcrsation is over.

Coghitive.Fecdback:
Positive:
Neutral:
Negative:

I understand your messag~; t's clear.

I' m not sure if I correctly understand you or not.
I don't understand what you are saying; it's not c!ear.

CHAPTER 8

Cco5'~Lingui;Nc lnf/uence and Leamec Language

233

positive affective feedback, encourage learners to try again, to restate, to

reformulate, or to draw a different hypothesis about a rule. Positive feedback in the cognitive dimension will potentially result in reinforcen1ent of
the forms used and a conclusion on the pa~t of learners that their speech
is well-forn1ed. FossHized item_s, acco.rding tii>'-this model, are those deviant
Uems _in, the :.s_pe~~. Qf a leat~e_r th.lt first gain positiVe -affectivc feedback

("Keeptalktilg")then.positiye cognitive feedback ("l understand'),. rein

forcfng-~n ittcurrt:ct.fortl1 of language.
It is interesting that this internalization of incorrect forms takes place
by means of the sarne processes as the internalization of correct forms. \Ve
refer to the latter. of course, as "learning," but the same elements of input,
interaction, and feedback are present. \Vhen correct forms are pfoduced,
feedback that says "I understand you perlectly" reinforces those forms.
Having di~cussed Vigil and OHer's model in sorne detail, we need to
exercise caution in its interpretation. \VhUe it is n1ost helpful, for cxainplc,
in understanding models of error correction, as we shall see in the next section, there are flaws in attributing such importance to feedback alone.
Selinker ..1nd Lamcndella (1979)-noted thatVigil an<l OUer'S 1nodel relied on
the notion of extrinst'c feedback, and that other factors internal to the
learner affect fossilization. Learners are not n1erely pawns at the merey of
bigger pieces in the chess game of language learning. Successful language
learners ten to-take charge of their own attainment, proactivelyseeking
means fDr-acqisition. So, fossilzation coukl be the' result of the--presence
or absence of interna! 1notivating factors, of seeking interaction \Vith other
people, of conscious1y JOcusi.ng on forms; and uf one's strategic in\-estment
n the learning process. As teachers, we may, and rightly, attach great ilnport~mce to the feedback we give to students, but we must recogrtize that there
are other forces at work in the process of internalizing a second language.

Various co1nbinati0ns of the nvo 111ajor types of feedback are possible,

For exa1nplc, a pcrson can indica te positive atie-c:tivt~ feedback ("l affirrn
you and vatue what you are trying to conu:nunicate") but give neutral or
negatLvc cognitive feedback to indicare that the n1essa:ge itself is unclear.
1-\cgative aff~ctive feeUback, h.Jwever, r.:::gardless ofrhe degree of cognirive
feedback, will likcly result in the abortion of the con11ntnlication. This is, of
course, consislent v1dth the overriding affCctve nature of hun1an interaction: if people are not at least affinne<l in their anc1_npts to comn1unicate,
there is litt!e reasou for continuing. So, one of the first requirements for
n1canngful con11nunication, as has been pointed out n earlier chapters, is
an affcclive a1firn1allon by the other person.
Vigil and Oller":c; rnuJcl thus holJs that a positive affective response is
in1pe-r'dtive to the learner's desire to continue attempts to con1mu11icate.
CognitiYe feedback then deternlines the degree of internalization. Negative
or neutral teedback in the cognitivc din1cnsion will, with the prcrequi.site

FOR:t"l-FOCUSED!NSTRUCfION
1\.s die focus of ...:lassruv1n instrc~:~ion has shifted over the past fevv deca<le:--

fron1 an emphasis on language forms to attention to functional language

within con1municative contexts, the question of the place of \\'hat has
come to be caUed "form-focused instruction"'(FF'I) has become n1ore and
more important. \Vhat <lo we n1ean, exactly, by FFI? A number of varying
definitions have emerged (Doughty & Williams 1998), but f?r th~ sake of
simplifying a complex pedagogical issue, lct us rely on Spa_da>'s nicely
worded definition: ~any pedagogfal effort whch is used t' <lraw the'
le-arners' attention to 'language forrn either irriplicitlY or explicitly"' (1997:
73). In1plied in the definition is a range of approaches to form. On one side
of a long continuum are explicit, discrete-point metalnguistic explanations

r /-IAPTEI?

Cross--l_inguist1c lnf/uence and Learner Language

and discussions of rules an<l exceptions, or curricula governed and

sequenced by gra111n1atical or phonological categories. On the other end of
the continuun1 are (a) implicit, peripheral references to form; (b) noticing
(Ellis 1997: 119), that is, the learner's paying attention to specific linguistic
features in input; and (e) the incorporation of forms into con1n1unicative
tasks, or what Ellis (1997) calls gramn1ar consciousness raising.
The research on this issue (l)oughty 8{_ \X''illian1s 1998; Long &
Robinson 1998; Spada 1997; El!s 1997; Lightbown & Spada 1990; Long
1988, to cite only a few sources) addresses a number of questions that must
be answered before one can conclude whether or not FFI is beneficia!:
1. Are son1e types of FFI more beneficial than others?
2. Is there an optimal tin1e to provi<le FFI?
3. Are particular linguistic fearures more affected by FFI?
4. Do particular srudents benefit n1ore fron1 FFI?
It is difficult to generalize the diverse findings on FFI over the years, but it
may be reasonable to conclude the following:

la. Most of the research suggests that FFI can indeed increase
learners' levels of attain1nent, but that the "Neanderthal" (Long
1988: 136) practices (grammatical explanations, discussion of
rules, rote practice) of bygone years is clearly not justified. Error
treatment and focus on language forros appear to be most effective when incorporated into a communicative. learner-centered
curriculun1, and least effective when error correction is a dominant pedagogical feature, occupying the focal attention of st11dents in the classroom.
2a. Very few research studies have been able to identify particular
stages in \Vhich learners are rnore ready than others to internalize FFI. r\ 1nore nportant question (Spada 1997: 80) is perhaps "y;rhether there are more propitious peda,cogical moments
to draw learners' attention to language form." Should a teacher
intcrrupt learners in the middle of an attempt ro comn1l1nicate?
One study (lighd)O\Vn & Spada 1990) suggested the ans\ver to
this question is "no." Should FFI come before or after con1municative practice?'fo111asello and IIerron (1989) found evidence
to support giving corrective feedback after a conm1unicative task.
3a. 1'he possible number of linguistic featun:s in a Ianguage and fhe
many potential contexts of iearning make this question in1possible to an~wer. One tantalizingly suggestion, ho\vever, was supported in I)eKeyser's (1995) finding that explicit instruction -~vas
n1ore appropriate for c-asily stated gr;in1n1ar rules and in1plicit

CHAPTE/\ 8

Cro5s-Linguistic lnfluence and Learner Lzrngua;c

235

instruction was n1ore successful for more complex rules.

4a. The \Vide-ranging research on learner characterstcs, styles, and
strategies supports the conclusion that certain learners clearly
benefit more than others from FFI. .A_f1alyric, ficld-independent.
left-brain~oriented learners internalize explicit FFI better than
relational, field-dept:ndent, right-brain-oriented learners
Gamieson 1992). Visual input will favor visual learners (Reid
1987). Students wbo are "js" and "1~s" on the l'v1yers-Briggs scale
will n1ore readily be able tu focus on form (Ehrn1an 1989).

ERROR TR.EAT1\1ENT
One of the n1ajor issues involved in carrying out FFI is the manner in \Vhich
teachers deal with student errors. Should errors be treated? Iiow should
they be treated? \"'V'hen? For a tentative ansv..er to these questions, as rhey
apply ro spoken (not written) errors, let us first look again at the feedback
n1ode! offered by Vigil and Oller (1976). Figure 8.2 metaphorically depicts
what happens in that modeL
The "green light" of the affective feedback 1nod.e allows the sender to
continue attempting to get a message across; a "red light" causes the sender
to abort such attempts. (fhe metaphorical nature of such a chart is e\ident
in the fact that affective feedback doe.s not precede cognitiYe feedback, as
this chart may lead you to belif-ve; both n1odes can take place sin1ultaneously.) The traffic signal of cognitive feedback is the point ar \Vhich error
correction enters. A green light here symbolizes noncorrective feedback
that says "I understand your message." A red light sy1nbolizes correcrive
feedback that takes on a n1yriad of possible forms (outlined helo\v) and
causes the learner to make sorne kind of alteration in production.To push
the metaphor further, a yellow light could represent those Yarious shades
of color tha.t a1-e interpreted by the !earner as fa.lling sc111e\\-here in
between a cornplete green light and a red light, causing the learner to
adjust, to alter, to recycle, to try again in so111e way. Note that fossilization
may be the result of too n1any green lights vvhen rherc should have been
sorne yellow or red lights.
'The most useful implication ofVigil and Olier's inodel for a theory of
error treatn1ent is that cognitive feedback rnust be optima! in order to be
effective. Too much negative cugnitive feedback-a barrage of interruptions, corrections, and overt attention to n1alforn1ations-often leads
learners to shu1 off thcir anempts at con1n1unication.They perceive that so
n1uch is \Vrong with their production that there is litt!e hope to get anything right. C)n the other hand, too n1uch positive cognitive feedback-\villingness of the teacher-hearer 1.0 !et errors go uncorrec1t"<.L to indic:ite

236

CHAl'TER B

ClfAPTER 8

Cross-Linguistic lnfluence and Learner Language

understanding when understanding may not have occurred-serves to

reinforce the errors of the speaker-lcarner. The result is the persistence,
and perhaps the eventual fossilization, of such errors. The task of the
teacher is to discern the optimal tension hetween positive and negative
cognitive feedback: providing enough green lights to encourage continued
comn1unication, but not ;<;o many that crucial errors go unnoticed, and providing enough red lights to call attention to those crucial errors, but not so
n1any that the learner is discouraged fron1 attempting to speak at ali.
We do well to recall at this pont the application of Skinner's opera.nt
conditioning n1odel of lcarning <liscussed in Chapter 4. The affective an<l
cognitive 1nodes of feedback are reinforcers to speakers' responses. As
speakers perceive 'positive" reinforcement, or the "green lights" of Figure
8.2, thcy will be leJ to internalize certain speech patterns. Correctivc tee<l~
hack can still be "positivc" in thc Skinnerian scnse, as Vv'e shall see below.
H\Yever, ignoring erroncous behavior has the effect of a positive reinforcer; thercforc teachers tnust be very careful to discern the possible reinforcing conscquences of neutral feedback. What we 1nust avoid at all costs
is the administration of punitive reinforcement, or correction that is viewed
by learners asan affective red light-devaluing, dehuinanizing, or insulting
thcn1.
Against thts theoretical backdrop \VC can evaluate son1e possibilities of
when and how to treat errors in the language classroon1. Long (1977: 288)
suggested that the question of when to treat an error (that is, which errors
to pro,ide sorne sort of feedback on) has no simple answer.
Having noticed an error, the first (an<l, I "\Vould argue, crucial)
decision the teacher makes is whether or not to treat it at all. In
or<ler to n1akc the decision the tcacher may ha ve recourse to fac-

Message

=V>

Abort
Red

Yel!ow

(O)}

Green

(+)

Affective
Feedback

Figure 8.1. Affective and cognitive feedback

X
Contnue

Y>

o }
0
Cognitive
Feedback

Recyde
Contlnue

Cross-Lmguistic lnf/uence and Learner Language

237

tors with iinn1ediate, ten1porary bearing, such as the importance

of the error to the currcnt pedagogical focus of the lesson, the
teachcr's perception of the chance uf eliciting correct perforn1ance from the stu<lent if negative feedback is given, and so on.
Consideration of these ephemeral factors n1ay be preempted,
however, by the teacher's beliefs ( conscious or unconscious) as to
what a language is and ho\v a new oPe is learned. These beliefs
n1ay have been formed years bcfore the lesson in question.
In a very practica! article on error treatn1ent, 1-Ien<lrickson ( 1980)
advisetl teachcrs to try to <liscern the difference between global and local
errotS, already <lescribed earlier n this chapter. Once, a learner of English
was describing a quaint ol<l hotel in Europe and said, "There is a French
widow in evcry bedroom." The local error is clearly, and hun1orously, recognized. Ilendrickson recomn1ended that local errors usually need not be
correcte<l since the message is clear and correction 1night interrupt a
learner in the flow of productive con1munication. Glohal errors need to be
treated in son1e way since the message may otherv,rise remain garbled.
"The different cty is another one in the another two" is a sentence that
would certainly nee<l treatment because it is incomprehensible as is. ~1any
utterances are not clearly global or iocal, and it is <lifficult to discern the
necessity for corrective feedback. A learner once wrote, "The granunar is
the basement of every langu.ige:' \Vbile this witty little proclan1ation may
indeed sound more like Chon1sky than Chomslry does, it bchooves the
teacher to ascertain just \vhat the learner meant here (no doubt "basis
rather than "basement"), and to provide sorne feedback to clarify the difference between the rv.-o.The bottom Une is that v.'e simply must not stifle
our students' attempts at production by sn1othering them with correcti,e
feedback.
The matter of hoiu to correct errors is exceedingly con1plex. Research
on error correction methods is not at all conclusive about the n1ost effective method or technique for error correction. lt seen1s quitf'" clear that students in ~he classroom generally \vant and expect errors to be corrected
(Cathcart & Olsen 1976). Nevertheless, sorne methods recornmend no
direct treatment of error at ali (Krashen &Terrell 1983). In "natural.' untutored environn1ents, nonnative speakers are usually corrcctecl by native
spcakers on only a small percentage of errors that they make (Chun et al.
1982); nativc speakers will attend basically onJy to global errors and then
usually not in the form of interruptions but at t1ansition points in conversations (Day et al. 1984). Balancing these various perspectives, 1 think \\7e
can safely conclude that a sensitive and perceptive language teacher should
make the language classroom a happy optin1um bet\.veen so1ne of the o,er

._.o

CH."'.F'TER 8
o-JAPTER

Cros.1-Linguistic lnf!uence and Learner L;inguage

239

Cross"Linguistic lnfluenn_ and /.. earner Language

politeness of the real world and the cxpectations that learners bring "\Vith
then1 to the classroom.
Error treatinent options can be classified in a nu1nher of possible ways
(see G::es 1983; Long 1977), but one useful taxono1ny was recomn1ended
bv Bailey (1985), who drew from the work of Allwright (1975).Seven "'basic
olJttons" are complemented by eight "possihle features" 'vithin each option
(Bailcy 1985: 111).

<legree in the n1ind of the learners, albeit idiosyncratic. Learners are processing language on the basis of knowledge of their o\Vn interlanguage,
\Vhich, as a system lying betv.reen t\VO languages, ought not to ha\e the
value judgn1ents of either language placed upon it.Thc tcacher's task is to
value learners, prize their attemprs to comn1unlcate, and then provide
optilnal feedback for thc syste1n to evolve in successive srages until
learners are co1nn1unicating n1eaningfully and un-an1biguously in the
seconJ language.

Ba._,,;;ic Options:
1.
2.
3.
4.
5.
6.
7.

To treat or to ignore
To treat irnmediately or to delay
To transfer treatment [to, say, other learners) or not
To transfer to another individual, a subgroup, or the whole class
To return, or not, to original error maker after treattnent
To permit otl1er learners to initiate treatment
To test for the efficacy of the treatn1ent

Possible Features:
1. Fact of error indicated
2. Location indicated
3. Opportunity for new atten1pt given
4. Model provided
5. Error rype indicated
6. Re1nedy indcated
7. In1provement indicated
8. Praise indicated
All of the basic options and features within each option are conceivably viable modes of error correction in the classroon1. l'he teacher needs
to develop the intuition, through experience and solid eclectic the~retic~l
foundati0ns, for a<>certaining which option or con1bination of opuons is
appropriate at a given moment. Principies of optimal arfe.ctiv_e anti cognitive feedb:1ck, of reinforcement theory, and of con11nun1cauve L1nguage
teaching all combine to form those theoretical foundations.
,
At !east one general conclusion that can be drawn fron1 the study of
errors in the linguistic syste:ms of learners is that learners are indeed creatively operatng on a sccond language-constructing, ei~her conscious_ly
or subconsciously, a systcm for understanding and produc1ng utterances 111
the Ianguage. That systen1 should not necessarily be treated a~ an in1perfect
system; it is such only insofar as nativc speakers compare the1r own knowledge of Lhe language to th;:it of the learners. lt should rather be looked
upon as a variable, dynanlic, approximative systen1, reasonable to a great

In the Classroom: A Model for Error Treatment

In these enC-of-chapter vignettes, an attempt has been made to
provide sorne pedagogical inforrnatlon of historica! or impl!cat!onal
interest. This chapter has focused strongiy on the concept of error
in the deve!cping learner !anguage of students of second languages,
and the last sections above honed n on error treatment in formfocused instruction. Therefore, one more step will be taken here: to
offer a conceptual rnodel of error treatrnent that lncorporates sorne
of what has been covered in the chapter.
Figure 8.3 i!lustrates what I would c!aim are the split-second
series of decisions that a teacher rnakes when a student has uttered
sorne deviant form of the foreign language in queSt!on. In those feV>t
nanoseconds, inforrnation is accessed, processed, and evaluated,
with a decision forthcomng on what the teacher is going to do about
the deviant form. Imagine that you are the teacher and let me walk
you through the flow chart.
Sorne ~ort of deviant utterance is rnade by a student. Instantly,
you run th1s speech event through a nurnber of nearly sirnultaneous
screens: (1) You identify the type of deviation (lexical, pho_nological,
etc.), and (2) often, but not always, you identify its source, the
latter of which will be useful in determining how you rnight treat the
deviation. (3) Next, the complexity of the deviation rnay determine
not.o_nly whether to treat or ignore, but how to treat if that ls your
dec1s1on. In sorne cases a deviatlon may require so rnuch exp!anation, or so rnuch interruption of the task at ha0ci, that it isn't worth
treating. (4) Your most crucal and possibly the very first decision
among these ten factors is to qulckly decide whether the utterance
is interpretable (local) or not (g!oba!). Local errors can sometimeS
be ignored far the sake of rnaintaining a flow of cornmunication.
Global errors by definition very ohen cal! for sorne sort of treatrnent
even !f only in the forrn of a clarification request. Then, from you~
previous know!edge of this student, (5) you make a guess at
whether it is a performance slip (rnistake) or a competence error.
This Is not always easy to do, but you rnay be surprised to know that
a teacher's intuition on this factor vvill often be correct. Mlstakes

CHAPTER 8

DEV!ANT UTTERANCE

1. Type
lexical, phonological, grammatical,
discourse, ragmatic, sociocultural

- ---

3. Linguislic coruplexity
lntricate & invo\ved or

casy to explain/deal with

7. learner 1s !ingulslic stage

cmergent, presystcrnatic,

2. Source
L1, L2, tec1cher-induced,
other Ss, outside l.2 input,
A/V/print/e!ectronic media

~ 4. local or Glob"al

s. Mistake or Error f--

!-+--

6. Learner's affective state

!anguage ego fragility, anxiety,
confidence, receptiveness
8. Pedagogical focus

f.-----+

systematc, postsy-,te1natic

..,--

immecliate task goals,

lesson objectives,

cCJurse gud!.s/urposes
9. Co1n1nunicative context
convers<1tio11Jl flow facturs,
individual, group, or who!e~dass work,

i--

style
..,-- 1 O. Teacher
director indirect,

interventionist, !aissez-falre

S-S or S-T exchange

/~
JTREAT I

!IGNORE

end of utterance

WHEN?

in1mediate!y

WHO?

HOWl
a. input to S

fact
indicated

b. manner

i nd irc;tt,' un 1ntrusiv2

c. S's output

non e

d. follow-up

non e

"okay"

non e

acknow!edge

another S
!ocatlon
indicated

correctlon
modeled

OUT

rnuch !ater

whole class

self

type/source
indicated

meta!inguistic
explanation
dlrect/intrusivc
rephrasc utterance

"good"

[gushl

affective
,. cognitive

verbalize

Figure 8.3. A modd for classroon1 treatment (A speech errors

240

further clariication

CrossLmguistic lnfluence and Learner Language

241

rarely call for treatment, while errors more frequently demand sorne
sort of teacher response.
Ali the above information is quickly stored as you perhaps simultaneously run through the next five possible considerations. (6)
From your knowledge about this learner, you make a series of
instant judgments about the learner's language ego fragi!ity, anxiety
leve!, confidence, and willingness to accept correctlon. If, far
example, the learner rarely says anything at ali, shows high anxiety
and lo1iv confidence 1ivhen attempting to speak, you may, on this
count alone, decide to ignore the deviant utterance. (7) Then, the
!earner's lingust!c stage of development, which you must discern
within thls litt!e microsecond, will tell you something about how to
treat the deviation. (8) Your own pedagogical focus at the moment
(Is this a form-focused task to begin with? Does this lesson focus on
the form that was deviant? What are the overa!! objectives of the
lesson or task?) wil! help you to decide whether ar not to treat. (9)
The communicative context of the deviation (Was the student in the
middle of a productive f!ow of !anguage? How easily could you interrupt?) is also considered. (10) Somewhere in this rapid-fire processing, your own style as a teacher comes into play: Are you

generally an interventionist? laissez-faire? Jf, far example, you tend

as a rule to make very few error treatments, a treatment now on a
minor devlation wou!d be out of character, and possibly interpreted
by the student as a response to a grievous shortcomlng.
You are now ready to decide whether to treat or ignore the deviation! If you decide to do nothlng 1 then you simply move on. But if
you decide to do somethirig in the way of treatment, you have a
number of treatment options, as discussed earlier. You have to
decide when to treat, who will treat, and how to treat, and each of
those decisions offers a range of possibilities as indicated in the
chart. Notice that you, the teacher, do not always have to be the
person who provldes the treatment. Manner of treatment varies
according to the input to the student, the directness of the treatment, the student's output, and your follow-up.
After one very quick deviant utterance by a student, you have
made an amazing number of observations and evaluations that go
into the process of error treatment. New teachers \.\'ll find such a
prospect daunting, perhaps, but with experience, many of these
considerations wlll become automatic.