DOCUMENT REOUME

ED 026 639 AL 001 734

By-Nemser, William
Approximative Systems of Foreign Language Learners.
Pub Date 10 Mar 69
Note-13p.
EMS Price MF-$0.25 HC-$0.75
Descriptors-*Contrastive Linguistics, English, German, Hungarian, Interference (Language Learning),
Language Instruction, *Language Learning Levels, Linguistic Theory, Phonology, *Second Languae Learning,
Serbocroatian, Syntax
Identifiers-*Approximative Systems, Source Language, Target Language
Evidence suggests that the speech behavior of language learners may be
structurally organized and that the contact situation should therefore be described
not only by reference to the source (SL) and target (IL) languages of the learner,
but also by reference to a learner system (AL). Investigation of such learner systems
is crucial to the developthent of contrastive analysis theory and to its application to
language teaching. However, these systems also merit investigation in their own right
through their implications for general linguistic theory. Experimental and informal
observation of the contrastive approach in its present form reveal serious limitations,
in part because learner behavior cannot be exhaustively described without reference
to the AL. Theoretical and practical considerations therefore converge to suggest
the direct and systematic examination of such learner speech, viewed within the
general framework of the current theory. Such investigation would (I) provide
attested information, of immediate utility in teaching and course development, on
patterns of learning behavior for the principal structures of the target languages; (2)
permit further assessment of the current suppositions of contrastive analysis; and (3)
make possible a preliminary description of AL, thus progressing toward a
reformulated contrastive approach. (Author/DO)
 `U S, DEPARTMENT Of HEALTH, EDUCATION 1 WELFARE
OffICE Of EDUCATION

THIS DOCUMENT HAS DEEM REPRODUCED EXACTLY M RECEIVED FROM THE

PERSON OR ORGANIZATION ORIONIATIN IT. POINTS Of VIEW OR OPINIONS

STATED DO NOT NECESSARILY REPRESENT OFFICIAL OffICE Of EDUCATION

P9SITION OR POLICY,

AFFROXIMATIVE SYSTEMS OF FOREIGN LANGUAGE LEARNERS

William Nemser

CENTER FOR APPLIED LI GU STICS

ApproxiniaillEXELSEP in fordsalanamEISEADA
situation can be classified
The language systems represeneed in a contact

in accordance with their functions as follows:

1 The tarat language is that in which communication is being attempted;

in the case of a learner it is the languag he is learning; when he uses it.

2. The source language is that acting as a source of interference

(deviations from ehe norm the target languagel); it is normally the learner's

native language.

formulated with Francis Juhasz

This paper represents the development of ideas
while at Columbia University some years ago, John Lotz and WilliamCW. Gage
significantly to ehese
of the Center for Applied Linguistics have contributed
responsibility
ideas and to eheir presentation here without, however/ incurring
for defects in either.

1 Uriel Winreich, Langges New York; 1953; p.

A L 001 734
 3. An appro'imativc system is the deviant linguistic syst m actually

employed by the learner attempting to utilize the target language. Such approximative

variation is also
systems vary in character in accordance with proficiency level;

introduced by learning experience (including exposure to a target language script

system), communication function, personal learning characteristics, etc.

For the sake of brevity, the following symbols will be employed throughout

thi,s paper.
IT: Target Language
Ls: Source Language
La: An approximative system
La...
41:Indices referring
to systems at successive
stages of proficiency

In identifying a specific type of La the name of the L precedes that of the

Le thus "German-English" refers to an La typical of native speakers of German

communicating imperfectly in English.

Our assumpt.on is threefold:

1 Learner speech at a given time is the patterned product of a linguidtic

system, LA distinct from L and LT and internally structured.

2. L Is at successive stages of learning form an evolving series La

a 1...n
the earliest occurring when a learner first attempts to use I, the most

advanced at ehe closest approach of La to Lt (merger, ehe achievement of perfect

proficiency, is rare for adult learners)

3. In a given contact situation, the La s of learners at the same stage of

proficiency rough* coincide, with major variations ascribable to differences in

learning experience.
 The speech of a learner, according to the assumption is structurally

organized manifesting the order and cohesiveness of a system; although one

frequently changing with atypical rapidity and subject to radical reorganization

;ming p-nceeds, As such,

through the massive intruaion of new elements as 2
Ls and At but in its
learner speech should be studied not only by reference to

own terms as well. From the point of view of the history of lit, Weinreich was

undoubtedly correct in assigning interference in the speech of bilinguals,

whidh be likens to sand carried by a stream", to the mos., of It along with other

community of It speakers
accidental and transient phenomena unincorporated by the
1
within their communal language system. However, from the point of view of the

contact situation proper, to regard these same features, as the te:1 interference
4

implies, exclusively as intrusive Ls elements interrupting the normal flow of

following
- a kind of hiccough view of language contact 7 is less rewarding,

the hypothesis, than viewing them first in terms of the learner system to which

they pertain

u e U4 e
II. kijoce
Eamatliaa
of these La's
In presenting evidence for the reality and structural autonomy
students, while
or approximative systems, it should be pointed out that language

of special interest here, represent a minority among La users. Moreover learner

lysteg, are by definition transient, while effective language teaching implies

preventing,or postponing as long as possible, the formation of permanent intermediate

systems and subsystems (deviant phonological and grammatical structures). Never

theless, it is clear that evidence for La is abundantly present in the patterning of

1. Weinreich, op.cit., p. 11.

errors in the perc ption and production of a given target language by learne4s

sharing the same native language. This regularity, in fact, forms a principal

basis for the belief that a comparison of Ls and LT provides information essential

to pedagogic strategy. Such characteristics constitute the "foreign accent"

typical of learners as well as of other bilinguals sharing the same mother tongue,

i.e. speakers of the same Ls attempting to communicate in a given LT

Stable varieties of L are found in

a
Lini21 that is, the speech

of long-time useru of LT who, often having attained considerable fluency in this

1
language, have yet obviously reached.a plateau in their learning. Attested examples

of such speech include the regular rendition by many veteran German-English speakers

of the English initial /64 cluster as (tv], and of the velarized variant of

English /1/ in post-vocalic position as a non-velarized phone (English Essall] 'swell'

German-English Plve ]). Similarly many speakers of Hungarian-English regularly

omit the plural marker in enumerative phrases (tjmel,...2a) but overtly indicate

contingency in both the apodosis and the protasis of conditional sentences (Il

would have one 1 would have seen him).

Moreover the speech of members of the same immigrant groups communicating

in their own native language (here German or Hungarian) often reveals Ole

systematic and widespread intrusion of elements of the dominant language of

the area (here English), with the interchange of the roles of Ls and LT and
2
the creation of new L systems (English-German, English-Hungarian).

.1.10401101.=111.mmOrPOMMI

Such systems have not been extensively studied. For same comments on rwedish-
English, Yiddish-English and several others, see H. L. Mencken, The American
Langusge, Fourth Edition, New York, 1949, pp. 212-222.

2. See Einar Haugen's monumental study, The Norwegian Lgium_ia.America, 2 vols.,

Philadelphia, 1953; see also, for example, James Macris, Arkagnis of Emagi
Loanwords in New York CiAxjayssh, Ph.D. dissertation, Columbia University, 1955,
and Leo Pap, zg2D._.__2_:icanSeechPortuuese-Ax, New York, 1949. Mencken, op. cit.,
pp. 616-697, briefly exemplifies English interference in some twenty-eight
languages as used by native speakers resident in the United States.
 Another subgroup of stable La ideolects is f rmed by utilay..Aystems,

such specialized "little" languages of limited semantic function; and requiring

limited grammars and lexicons, as the systems often used by taxi-drivers, hotel- 1

reservation clerks; bartenders and other groups with frequent but circumscribed
1
requirements to communicate with foreigners.

The term learner pidgin can be applied to systems of a related type often

employed by language students who have attained fluency in the target language

without mastery of its fundamentals; but have arrived at a stage in instruction

where attention has largely shifted from form to content. Not only do teachers

often concur in the use of this system; but even participate as users (the

following exchanges were observed in a language classroom: Arabic-English

speaker: Same? (i.e. Are the two wsturarlounstdinthe same wayij; Teacher:

Same.; Teacher: Short answer. [Le. Use the short answer form.]. In converse ion

Im_sagElL). This variety oi La is also frequently used by other LT speakers

when communicating with non-natives and; apparently; even sometimes with other

native speakers:

"Another brandy," he said, pointing to his glass.

The waiter who was in a hurry came over. "Finished,"
he said, speaking with that omission of syntax stupid
people employ when talking to crunken people or foreigners.
"No more tonight. Close now."

OIMAIW.WWW,OWWWW184.1~11m001.01.1.11.WOOMMOMMOMMOMMOSIOMMO.000.04111111.1101ftalWalio.

However; for two

1. Little information has been collected on these systems.
apparently successful attempts to construct such La systems (English-Russian;
English-Mandarin) for specific; circumscribed communication requirements,
see Eugene H. Rocklyn, Self-Instructional Tactical Lamm:I...Course in_Russian;
Washington, D.C.; George Washtngton University; Human Resources Research
Office, 1965; and Catherine Garvey and Eugene H. Rocklyn, DeveloEmInt
and Evaluation of a Tactical Mandarin Chinese Lan ua e Course, Washington;
D.C.; George Waahington University, Human Resources Research Office,
1965.

2. "A Clean Well-Lighted Place," The Shore Stories

of Ernest Heminam, New
York, 19531 p. 381,
 Moreover these learner pidgins are apparently preserved in the language-

types customarily designated as pidgins and creoles historically La systems

usually incorperating Ls grammatical elements and lit lexical elements.

An argument for the structural independence of an La from the source and target

systems is the frequent and systematic occurrence in non-native speech of

elements not directly attributable to either Ls or lit In the phonology,

intermediate phones are common (Hungarian subjects in an experimental study

often rendered English /0/1 for example,as [fel or All]). Similarly, "internal"

interference resulting from the extension of the productive processes of LI (such

formations as Ap-ed are common in learner speech, as they are in child language), and

pattern confusion (observed in a language classroom: Serbo-Croatian English

What does Pat doin necurs frequently in ehe grammar.

More theoretically it can be argued that the demands of communication

force the establishment of phonological, grammatical and lexical categories,

and that the demands of economy force the imposition of the balance and order

of a language system.
2
Finally, there has been at least one attempt to study an L variety directly.
i
Customary descriptive procedures were employed to characterize, in sui generis

terms,the phonology of a native speaker of Hungarian, at an early stage in her

learning of English, attempting to communicate in the latter language. Some

fluctuation between categories attested to the transient nature of the organization

411.00.1NOM....4...10.111.111..111..00

1. William J. Nemser, Agglxiat_lhaptic Experiments} American Council of

Learned Societies Research and Studies in Uralic and Altaic Languages,
Project No. 32, 1961, pp. 56-84.

2. William J. Nemser and Francis S. juhasz, A Contrastive Anal sisofLEIgalriEra

and Illnaliftitallalual American Council of Learned Societies Research and
Studies in Uralic and Altaic Languages, Project No. 70, 1964, pp. 163-216.
(for example,two high front vowels, [i] and [J. :], wore sporadically distinguished,

on the basis of length). However such fluctuation is, of course, also typical of

categories in the process of change in normal language systems, and the analysis

revealed a system exhibiting true internal coherence, with distincLive components

from both English and Hungarian recombining to form phonological structures

differing from those of either language (a mid-central vowel, for example, was

opposed to a low-front vowel, representing a merger of E/e/ and AV, on the basis

of rounding). Observation of the same speaker's Hungarian-English gramthar disclosed

an analogous tendency toward autonomous organization.

Moreover, there is some evidence that the various evolutionary stages of

L differ not only in amount but in Sims of interference (using "interference" to

a
cover both external and internal types). Earlier stages are apparently characterized

by the extensive underdifferentiation (syncretism) of Lt phonological, grammatical

and lexical categories, with the learner extonding the distribution and (in the
-2

ease of the grammar and lexicon) semantic domains er the limited number of formal

elements he has acquired. Later stages are characterized by ehe addition, as

interference types, of rsiiInteux (Serbo-Croatian English speakers elle edly

often aspirate English tense stops in all positions, apparently phonemicizing

a feature non-distinctive in English), huarsosuplion (Spanidh-English speakers

often substitute [ 1, which does not have Phonemic status in Spanish, for Arq:
1
English /szn/ 'sun' ) Spanish-English (sao] ) and analogy (so-ed).

Perception," American
1. See Albert H. Marckwardt, "Phonemic Structure and Amral
S eech 21 (1946), 106-111.
 In addition to the atypical rapidity with which they often undergo structural

changes, La's differ from normal languages in that La speakers do not usually form

speech communities Adult members of such communities normally model their speech

on that of other members of the same group; children, viewed vs groups of speakers

of child language at various stages of evolution, model their speech on both

internal standards the speech behavior of their peers - and external standards -

the speech behavior of older children and adults (for such child learners of LT

of course, there is no LO; La speakers, however, primarily select external

standards, the speech of native speakers of I. Nevertheless, it is likely that 1.4

speakers frequently provide reinforcement for the speech behavior of each other (even

resulting in the creation, in some instances, of such dialects or languages as the

2
varieties of English used by numerous speakers in India ), and it in observable that they

frequently communicate with each other wore easily than with lit speakers. Mbreover,

La features o:e sometimes disseminated among learners under special conditions (as

was a trilled r substitute for the French uvular phoneme, reportedly, among one group of

English-speaking learners at Middlebury College), are sometimes conventionalized

in L tinglish /0/ is regularly merged with /s/ in the instruction at certain

a
schools in Germany), are sometimes transhitted between generations (the children

of native speakers of Yiddish in New York, while native speakers of English, appear to

frequently adopt certain interference patterns from their parents' speach), and

even become conventionalized in Lit (during one era in Hungary, some native speakers

of Hungarian snobbishly replaced the Hungarian trilled r with the uvular variety of

French to suggest prior knowledge of that language and hence higher social status;

c.f also widely posited instances of substratum, intrusion).

1. Other than earlier phases of their systems: see below, pp. 10-11.

2. See, for example) Ashok R. Kelkar, " 'Marathi English': a Study in Foreign
Accent," Word 13 (1957), 268-282.
III. Reasons for studying La
101

1. Direct and systematic examination of learner speech has been largely

neglected. Classroom teachers, while ayrare of general patterns in learner

behavior and often taking them into account in their teaching, have rarely
within a linguistic
attempted comprehensive studies of these regularities

framework. Contrastive analysis specialists, on the other hand, often primarily

have
concerned with techn4.ques for establishing inter-systemic correspondences,
from im
been content for the most part to derive empirical support for their formulations

pressionistic observation and intuition. Investigation of La data would: therefore,

yield as its first result new concrete information on learner behavior of high utility

to the classroom teacher in the planning of pedagogic strategy.

2. Such investigation is also a prerequisite for the validation of both

1
the strong and weak claims of the contrastive approach:

a. The strong claim states that learner behavior is predictable on the basis
arising include: (i)
of a comparison of Ls and I. However problems immediately

different analyses yield different predictionsl(ii) predictions are often ambiguousland

(iii) the various levels pf linguistic structure are interdependent, with the

result that predictions of phonic interference, for example, must take into account

not only the phonologic systems of Ls and LT but their morphophonemic, grammatical

and lexical levels as well. Serious attempts to validate the theory have not been

numerous, with investigations not infrequently presenting ex zest facto reconstructions

1. Representative studies are found in John H. Hammer and Frank At Rice, eds.,
hjigglIcamby_o_Lamtrastive LinAuistical Center for Applied Linguistics,
Washington, D.C., 1965.
 - 10 -

as ev donee of predictive power. The few serious validation studies raise doubts

about the tenability of the strong claim,

that of accou t r learner
b. The weak claim of contrastive analysis is
For
behavior, Even this limited claim proves difficult to support, however.

Hungarian-English speakers of the final cluster

example, the frequent interpretation by
of
of English Awl as (ks] can be explained as resulting from one, some, or all

/gz/ in final position,

the following factors: a) the non-occurrence in Hungarian of
in
b) orthography and the Hungarian phgneMic rule requiring clusters uniform
final /z/, usually
voicing, c) the total unvoicing by some English speakers of

described as distinctively lax in English but voiced in Hungarianland the uniform

voicing rule, and d) Hungarian morphophonemic rules making the voicing of a

final stem obstruent dependent on that of initial suffix obstruents other than
speakers, for one of
/v/ (/hilz-bon/ fas-t61/)11 with the Hungarian-English

the reasons mentioned above, iaving selected the English allomorph is/ to

represent the plural.

is required as well
3. It can be shown that the direct examination of La

The approach is based

by the suppositions of the contrastive approach itself.
subsequent
on a general view of learning according to which prior learning affects

learning, posic,-4e1y where the new skill coincides with one already mastered,

negatively where they are opposed (positive and negative transfer). However,

language structures, viewed as compendia of verbal skills, are not comparable

counterpart in the
in their entireties. PhonologicAl tlements sometimes have no

See John Lotz, "Contrastive Study of the Morphophonemics of Obstruent Clusters

in English and Hungarian," in 'Miscellanea di studi dedicati a Emerico Virad
Modena, 1966, pp. 3-7.
 11-

opposed system (the clicks of certain Bantu languages cannot be related to English

phonemic categories), cultural differences clearly often make this true of lexical

elements as well, and grammatical categories,toolare often incommensurate. More

significantly, however, application of the theory for purposes of predicting and

elucidating learner behavior often depends on what can be called the "blinding-

flash" fallacy - the supposition that Ls and LT come into total contact - so far as

overlap permits - from the outset of learning, with Ls categories fusing with

their lit counterparts throughout the systems. Actually, of course, the learner's

exposure to Lt is necessarily gradual. This fact entails a dilemma for contrastive

analysis which can only be resolved by reference to La At post-initial stages of

:Iguage learning, the prior learning which conditions st.Assatlaat learning includes

not only the learner's knowledge of Ls but his own recent experience in language

acquisition - his knowledge of La - as well. He is no longer the pristine speaker

of L assumed by dialinguistic analysis, but als the user of a more recently

s

acquired system. Thus the precepts of contrastive analysis itself force the

inclusion of reference to L data in the prediction and elucidation of his

a
subsequent learning behavior./

4. Finally La merit examination in their own right, having interest

for general linguistic theory comparable on the one hand to child language and

on the other to the language of victims of certain types of speech disorder, as de

pendent systems forming evaluative gradations toward specific languages but

falling outside the normal dialectical and stylistic scope of these languages.

0.00.1010.411.1..,

1 See J. A. Upshur, "Language Proficiency Testing and the Contrastive Analysis

Dilemma," komlatjAmaira 12 (1962), pp, 123-127.
 12-

§mmmarzsiaLessausio
Evidence suggests that the speech behavior of language learn rs may bu

structura4y organized, and that the contact situation should therefore be

described not only by reference to the native and target languages of the

learner (L and LT), but by reference to a learner system al ) as well.

Investigation of such learner systems is crucial to the development of

contrastive analysis theory and to its applications to language teaching.

However, these systems also merit investigation in their own right through

their implications for general linguistic theory.

In its present form, the contrastive approach seeks (a) to predict and account

for learner behavior by reference to similarities and differences between

Ls and l and in terms of these systems, and by this means (b) to inslicate

a strategy for language pedagogy. However) experimental and informal observation

reveal serious limitations in the approach, in part because learner behavior

cannot be exhaustively described without reference to L

Theoretical and practical considerations therefore converge to suggest the

direct and systematic examination of such learner speech, viewed within the

general framework of the current theory. Such investigation would (a) provide

attested information, of immediate utility in teaching and course development,

on patterns of learning behavior for the principal structures of the target

languages,(b) permit further assessment of the current suppositions of contrastive

analysis, and(c) make possible a preliminary description of La, thus progressing

toward a reformulated contrastive approach of greater sophistication.

 13 -

An ultimate goal might be the reformulation of Ls and LT descriptions in

terms permitting ehe accurate projection of La throughout its successive stages
in each contact situation. This goal remains in the distant future, however,
and no present alterilative exists to empirical investigation.

March 10, 1969