REPORT RESUMES

ED 018 786 AL 001 171

BECOMING A BILINGUAL.
BY- ERVIN- TRIPP, SUSAN
CALIFORNIA UNIV., BERKELEY, LANG....BEHAV. RES. LAB.
REPORT NUMBER WP -9 PUB DATE MAR 68
EDRS PRICE MP-90.25 HC -90.84 19P.

DESCRIPTORS- *BILINGUALISM, CHILD DEVELOPMENT, LANGUAGE

DEVELOPMENT, *SECOND LANGUAGE LEARNING, *SOCIOLINGUISTICS,

A BILINGUAL -IN- PROCESS COULD BE A CHILD GROWING UP IN A

BILINGUAL ADULT MILIEU, OR AN ADULT WHO HAS MOVED TO A
DIFFERENT LINGUISTIC MILIEU. THE LEARNING PROCESS MIGHT BE
CASUAL EXPOSURE OR SYSTEMATIC PEDAGOGY. THERE IS STRONG
EVIDENCE THAT FOR CHILDREN UNDER ELEVEN LANGUAGE IS SOUND AND
FOR ADULTS, SENSE. CHILDREN ATTEND MORE TO THE SURFACE, JUST
AS THEY ALSO CONNECT SPEECH MORE TO THE IMMEDIATE SITUATION
IN WHICH IT OCCURS. FOR ADULTS, LANGUAGE IS TRANSPARENT,
SINCE ADULTS RAPIDLY PENETRATE THE SURFACE OF AN UTTERANCE TO
'ITS MEANINGS, TO A NETWORK OF CONNECTED THOUGHTS. THE BASIS
FOR THIS DIFFERENCE BETWEEN CHILDREN AND ADULTS IS UNKNOWN.
IF THE DIFFERENCE IS NEUROLOGICAL, OR LIES IN THE LOSS OF AN
ABILITY, CHILDREN MUST BE EXPOSED TO DIFFERENT TEACHING
METHODS THAN ADULTS, SINCE THEIR ABILITIES DIFFER. IF THE
DIFFERENCE IN BEHAVIOR IS A CONSEQUENCE OF SHIFT OF SET OR
ATTENTION OR THE RESULT OF THE ADULT'S GREATER RICHNESS AND
. SKILL IN SEMANTIC ASSOCIATION, THE PEDAGOGICAL IMPLICATIONS
ARE QUITE DIFFERENT. THESE CONSIDERATIONS AND OTHERS
CONCERNING AGE OF LEARNING AND THE SOCIAL MILIEU ARE BROUGHT
TO BEAR IN THIS PAPER, TO SUGGEST NEW DIRECTIONS THAT
RESEARCH MIGHT TAKE IN THE STUDY OF BILINGUALISM.
(AUTHOR/AMM).
 U.S. DEPARTMENT OF HEALTH, EDUCATION & WELFARE

CO OFFICE OF EDUCATION

CO THIS DOCUMENT HAS BEEN REPRODUCED EXACTLY AS RECEIVED FROM THE

PERSON OR ORGANIZATION ORIGINATING IT. POINTS OF VIEW OR OPINIONS
STATED DO NOT NECESSARILY REPRESENT OFFICIAL OFFICE OF EDUCATION

P.
O POSITION OR POLICY.

BECOMING A BILINGUAL

Susan Ervin-Tripp

Working Paper No. 9

Language-Behavior Research Laboratory
March 1968

The Language-Behavior Research Laboratory

is funded by a grant from the Institute
of International Studies, University of
California, Berkeley, whose support is
gratefully acknowledged.
"PERMISSION TO REPRODUCE THIS
IMP. MATERIAL HAS BEEN GRANTED
BY

TO ERIC AND OROANIMONS OPERATING

THE U.S. OFFICE OF
UNDER AGREEMENTS WITH
EDUCATION. FURTHER REPRODUCTION OUTSIDE
OF
THE ERIC SYSTEM REQUIRES PERMISSION
THE Mai OWNER."

AL 001 171
 A bilingual-in-process could be a child growing up in a bilingual
adult milieu, or an adult who has moved to a different linguistic milieu.
The learning process might be casual exposure or systematic pedagogy.
In this paper, some of the considerations affecting age of learning and the
milieu are brought to bear to suggest new directions that research might
take. These observations arose from discussions at the UNESCO Bilin-
gualism conference, Moncton, Canada.

Pedagogy and Age

Phonology. There is strong evidence that for children under eleven

language is sound, for adults, sense. Children generalize more between
words alike in sound, give more clang associations, confuse the meanings
1
of similar-sounding words. In adults, similar behavior appears in feeble-
mindedness and under drugs. One might say that for adults language is
transparent, since adults rapidly penetrate the surface of an utterance to
2
its meanings, to a network of connected thoughts. Children attend more
to the surface; just as they also connect speech more to the immediate
situation in which it occurs.
The basis for this difference between children and adults is unknown.
If this difference is neurological, 3 or if it lies in the loss of an ability
(like the traditional notion of eidetic imagery) then there is a clear pedagog-
ical implication: children must be exposed to different teaching methods
than adults, since their abilities differ. If the difference in behavior is a
consequence of shift of set or attention (like the shift from color-sorting of
blocks to form-sorting), or if the difference is a result of the adult's
greater richness and skill in semantic association, then the implications
are quite different. First, one would have to find out what the age curve

6.1*,..Mirmankurlorma.
is, for specific items, to see if age difference in learning rate affects new
4 matches the
sounds where there is no negative transfer, and if the curve
simu-
generalization curve mentioned above. If so, then one might seek to
play with sounds that
late in adults the conditions of attention to sounds and
attention to sound
are common in child use of language. In one experiment,
semantic information (glosses) for a few
was increased by simply delaying
Phonological skill in this group was no better than in a control group
days.
5
with no delay.
in Osgood's
Lexicon. Children's lexicon is composed almost entirely,
terms, of signs rather than assigns.
6
New words are normally learned in
of the adult's vocabulary
the context of visual-motor activity, whereas much
verbal. Asher
is learned in a purely verbal context so that its meanings are
increases in learning rate and retention when adults
has claimed dramatic
like children with respect to learning context, i. e. , when they
were treated
they performed and
were taught to recognize words referring to actions
7
objects they handled.
grammatical
Grammar. Differences between adults and children in
capacity"
capacity may arise from limitations in memory and "programming
rules they can
rather than limitations in the character of the grammatical
quantitative than qualitative.
process. That is, the differences may be more
to be some limits in the grammatical rules used by
At six, there do appear
of the English
English-speaking children. There are some specific details
verbs, pronominali-
system to be worked out, such as nominalizations of
complex structures
zation, participial verb complements, and semantically 8
"so" clauses, and perfect aspects. The children do not know
like "if" and
styles. But it
the rules involving rare structures, or those used in various
how early
is impressive to see in a variety of studies in different languages 9
are acquired.
most grammatical patterns and sociolinguistic variations
 -3

In order to know whether control of a grammatical pattern in one

language will facilitate learning an analogous pattern in another language,
one needs an underlying theory of the logical structure of grammatical
rules.
The results from studies of grammatical development have so far not been
stated in a sufficiently abstract form, transcending the specific structures
of each langue, and even of Iangage. The emphasis of general cognitive
research on children has been on development before two and after five, so
we know little about the cognitive operations children develop during this
age period, which is most critical for language. But judging from the child
language diaries it appears that there must be, by school age, an extraordi-
nary capacity for grammatical learning. There is no evidence of basic intel-
lectual barriers to learning new language structures quite early, provided
(a) the semantic distinctions are not difficult ones, such as the conditional,
and (b) the training input is not too complex quantitatively, in terms of the
amount of imbedding, or the co-occurrence of new meanings with new gram-
matical structures. It may be that even these limitations are sufficiently
inconvenient so that from the standpoint of learning grammatical patterns,
unlike the learning of sounds, early teaching is no distinct advantage.
appear
Imitation. Studies of input-output relationships in imitations
Here
to be a fruitful way to characterize the linguistic system as it changes.
I shall draw on some pioneering work of Charles Welsh, who has been deve-
loping a process model of utterance imitation for a two year-old child. This
model can predict the output for any input. While processing models have
10
been offered before, the convenience of imitation is that both input and
output are fully specifiable.
The Welsh model contains first a phonological analyzer. Both
11

segments and phonotactic patterns are analyzed according to the child's

rules. For example, the child may consistently convert "banana" to
 through a general
['mama], and "gramma" to Poamaj and ''gun" to Nan):
but preserves
analyzer which perceives all nasals as in initial position
12
other features of the initial consonant.
with category
The second component in the model is a dictionary
within certain
markers, which assimilates what is heard to familiar words,
crying" became "Cynthia and Tasha
limits. Thus "Chomsky and Veritas are
" If the sen-
cry" but "cui Bono is the quarter" became "cui bona a quarter.
enter the child's diction-
tence is less than five words long, new items could
ary, receiving the category marker inferred from its position.
The third component is an auditory storage device for holding
13
In one model this analysis
material while further analysis occurs.
rather like a Markov chain. In Welsh's current
consists of predictions,
form of category
thinking, there is a set of pre-analyzed templates in the
in terms of category
sequences. The surface structure of the sentence,
selected, which
markers, is scanned and the appropriate template is
what to do when standard order
includes "encounter-operate rules" for
English S-V-O) is violated. It is these templates which result in
(e. g. ,
the chair was
the return of "The boy the chair hit was dirty" as "Boy hit
fast" as "I saw the man
dirty, " and "The man who T saw yesterday runs
here" as "Some
and he run fast, " and "The pencil and some paper are
pencil here and some paper here. "
language
An analysis of such rules at various stages of second
14 One can expect that there
learning would prove highly enlightening.
imitation as new templates
might be sharp changes in comprehension and15 .
It is i
important to note
or new enounter-operate rules are acquired.
not logical models of the rules of a language, such
that these models are
of a linguist, but an attempt to characterize the processing algo-
as those
different
rithms of real speakers. They will therefore contain quite
components and types of rules.
 -5

Imitation is often used as a pedagogical device, and it is frequently

considered both a necessary and sufficient account of language learning.
Recent evidence suggests that it is neither, at least in terms of structural
learning. 16 Spontaneous imitations of two-year-olds, whose linguistic
systems are undergoing rapid change, are as simple or simpler syntactically
than their free speech. Many adults and some children learn languages
without any overt imitations, as well as without correction, to a degree
beyond that required for intelligibility. Thus we do not in fact know how
to account for the fact that the linguistic system changes very rapidly,
except to refer to changes in the system of comprehension. For example,
children may say "otherbody' before they say "somebody, " "tomorning"
before they say "tonight, " and "do-ed" before any regular past tense. This
evidence suggests that children's structural analysis of what they hear,
rather than any rote imitation, is the key to systematic change.
Elicited imitation in the classroom probably has two values: motor
drill and the manipulation of attention. The first, of course, refers to a
peripheral skill in articulating sequences. The second is more interesting.
It may be that elicited imitation is like disconfirmation in logical or cognitive
development. Disconfirmation can draw attention to features hitherto
ignored as noisy or irrelevant. Short simple sequences might be repeated
to a point which violates former processing rules, thus forcing the rule
system to change. The imitation of versus, songs, and dialogues, advocated
by Jones, thus has value only if there is evidence that the learner compre-
hends the components and produces imitations that are phonologically or
grammatically superior to free speech. Even this kind of practice may not
succeed in altering the structures for sentence production, of course. If the
imitations used in the classroom are consistently filtered through the exist-
ing processing device of the pupil without any effect on that device, then they
are not pegagogically useful for learning the linguistic system, though they
may have other uses.
 -6-

Social Milieu Of Learning

The above discussion pertains to school teaching of a second lan-

guage or of the mother tongue. In almost all respects other circumstances
of bilingual acquisition are dissimilar; in social support of the two languages,
values, norms of correct usage, and sociolinguistic rules for speech. I
shall touch on each of these points briefly.
By social support of bilingualism, I mean that the learner hears
speech in several languages outside the classroom, either because he moves
between two monolingual communities or because there are consistent rules
governing alternations in a bilingual community.
Social support appears to be of greater importance to children than
to adults. It is a common complaint of sojourners abroad that their children
both learn and forget languages too readily, whenever the linguistic milieu
is changed. It could be that when the milieu is reinstated there would be
marked savings on re-learning, so that there is not so much "forgetting"
as lowered availability. On this point we sorely need systematic research.
Perhaps children's selection of linguistic variety is more dependent on the
social milieu and less dependent on private motives than the adult's. Adults
can sometimes alter the language used to a given interlocutor at will. In
addition, their rich inner speech and their access to reading may provide
a form of support for linguistic forms which children lack. An adult may
retain a language for forty years as its sole speaker, merely with the aid of
inner speech. On the other hand, if children are dependent on social support,
elementary school foreign language programs may have serious problems
in the event that there is a continuous exposure to a language neither in the
school nor outside.
 Values play an important role in determining whether a given con-
dition of social support will produce or sustain learning. At a gross level,
beliefs about the ease or appropriateness of becoming bilingual may affect
the probability of child or adult learning. In India it is assumed that
children will readily become multilingual; in the United States bilingualism
is taken as a matter of course only where the second language is English.
Speech markers of social identity carry a strong value which may promote
or retard learning. Labov, for example, has noted that the speech features
of the women teachers in New York may not be learned readily by working
class boys. Both the teachers and pupils may share a belief that non-stan-
dard English emplies toughness.
In addition to altering the effects of a fixed social milieu, the learner's
values may lead him to alter the milieuto increase or decrease exposure
to the second language. Thus Japanese women married to Americans
learned fluency as a simple function of years in the United States, but
beyond the needs of rudimentary communication there were vast differences
in the degree of learning of phonology and grammar and even of American
18
ideas, related largely to their values and education.
Values should enter predictions at two points. If circumstances
do not guarantee exposure, values may lead to seeking out conditions for
listening and inner. speech. If the social milieu provides support, then the
social meaning of linguistic markers will determine how far second language
learning progresses beyond lexical alternations and the basic syntax neces-
sary for intelligibility.

Primary Language Data

Any full analysis of the process of learning must contain realistic

specification of the actual input system, or in this case the "primary
 -8-

language data, " including the stable and variable features, the
social
Gumperz
meaning of each variable, and the co-occurrence rules. As John
community
has most fully demonstrated, it is the norms of the face-to-face
19
which influence bilingual speech.
L2 in a
In school learning, for example, the pupil may never use
monolingual setting, nor learn the sociolinguistic rules of that
setting. Even
nearby, there
in social milieux where two monolingual communities are
usually is at least a bilingual belt between, and only interpreters and
with resultant
travelers would have occasion to frequent both communities,
constraints on their linguistic behavior.
they
Probably most bilinguals live among others like themselves;
The bilingual
may have contact with only one or no monolingual community.
for
is likely to be exposed to a single set of semantic and phonetic ranges
many linguistic categories. An American Indian
child in the Southwestern
and pre- vocalic
United States hears about him a form of English with inter-
Canadian franco-
glottal stops and simplified final consonant clusters. The
lexicon in both speech varieties, so the
phone hears considerable common
"man" are not.
"sink, " "hotel, t and "table" are shared, but "homme" and
maximal co-occur-
One is likely to find maximal separation of varieties and
carefully monitored
rence restrictions only in the highly self-conscious,
20
formal and written registers.
Even in bilingual communities maintaining considerable linguistic
separation, there may be sociolinguistic convergence. American Nisei
Japan;
have not learned Japanese speech etiquette, and appear rude in
appropriate to
American Lebanese may lack classical Arabic allusions
by the
formal situations; the familiarity and status distinctions carried
be
second person pronoun or inflection of the verb in many languages may
presumptuous. Thus
lost by American bilinguals so that the speaker sounds
be a social boor.
even if the classical "true bilingual" existed, he might
 -9-

Interference

In all studies of language learning, there must be some way to

characterize the linguistic system of the learner. Traditionally, this
analysis has consisted of noting from tapes or writing the deviations of
the learner's output from some ideal norm. When these deviations can
be attributed to structures in another language, they are called interferences.
There are at least three general classes of phenomena which have
been included in this term. These are features in the systematic norms of
the bilingual community, or its language and sociolinguistic rules; system-
atic features of the learner's language at a particular point in time; and
performance errors.

Compouns norms. In the language of a bilingual community there

may be fixed or compound features shared by both linguistic varieties. This
is especially likely to be the case with semantic and phonetic features. In
the example given above, sink is a lexical item common to both the French
and English linguistic milieux for representing the same semantic category.
Second, there may be systematic alternations between the two varie-
ties, which are part of the sociolinguistic norms of the community and carry
social meaning which the members can identify. Blom and Gumperz have
found that even when speakers can recognize the social meaning of switching,
they may not be able to control switching consciously when they talk among
themselves.
21
They refer to situational switching for the case when the
variety is predictable from the interlocutors, setting, or topic. Metaphor-
ical switching occurs within a given situation for connotative purposes.
 -10-

Negative transfer and simplification. A newcomer, whether child

or adult, to a new linguistic milieu must master a new system. If the milieu
is bilingual, he must master as well the rules for alternation between the
two varieties. These rules can be characterized by either a linguistic model
or a performance model. He must learn general grammatical categories,
rules of arrangement of those categories, phonetic and semantic distinctions,
and particular morphemes which represent semantic and grammatical
categories. It frequently is the case that in lieu of learning all of the new
features, he continues to employ the same distinctions, the same grammat-
ical categories, the same rules of arrangement, and even may import
morphemes into the new variety. In the process of learning he may over-
generalize newly learned features and alter the initial system accordingly.
For example, a Frenchman speaking English may regularly use "who" as
the subject of relative clauses, as in "That's the book who is on the table. "
He has a common syntactic rule in both varieties and merely alternates
II
qui II and who
/
as diamorphs.
1/
22
In such cases, whether it be L 1 or L2
which is affected, we speak of interference because features are used in
common in both languages which are not shared in the speech community
from which the norms derive.
However, it also happens that learners employ patterns common to
neither language. When this happens, we may find something analogous to
the interesting idiolectual rules in child language development. A frequent
occurrence is the omission or overgeneralization of morphemes in the new
variety, even where the appropriate semantic or syntactic category exists
in the primary language. We might call such instances simplifications. By
using a reduced set of distinctions, by omitting inflectional morphemes, the
learner cuts down the task in sentence production. Possibly the morpholog-
ical and syntactic simplifications of second-language learners correspond
 language.
to some simplifications common among children learning the same
learners, learner
Where these become stereotyped modes of addressing new
23
and teacher may develop a pidgin.
recognize
Performance errors. While the speaker may control and
his output the rules which
a norm for speech, he does not always realize in
learners, of bilin-
he knows. This is true of practiced speakers as well as
inconsistent, and
guals as well as monolinguals. Performance errors are
long, gram-
tend to occur in fatigue or under stress, or when sentences are
from overtaxing
matically complex, or contain novel lexicon. They arise
speech system
the "programming capacity" of the speaker. The bilingual's
rules, both linguistic and sociolinguistic, than the
contains more complex
violate co-occur-
monolingual's, and therefore his performance errors may
producing intrusions.
rence restrictions socially or linguistically,
there
It would be of great interest to psycholinguists to know whether
example, it
is a non-random distribution of performance errors. For
frequent following
appears in English texts of Frenchmen that loanshifts are
change, there may be oscillation
cognates. In a system undergoing constant
between rules from two adjacent stages of development
in the learner's
characteristic feature of performance errors that
dialect. It might be a
simplification typical of an
they include forms of negative transfer or
supplement
earlier stage of learning. For this reason, it is of value to
imitation measures
textual data with tests in the form of comprehension of
richer criteria of those regularities which occur under all
which provide
24
conditions of performance.
interference are correct,
If the distinctions between different types of
system, is
then the second kind of analysis, the analysis of the learner's
central to an understanding of the process of bilingual learning.
 -12-
the
A series of studies in which the social conditions of learning and
data are specified should predict outcomes in terms of the
primary language
learner's idiolect, or the language of a group of learners. For example,
and Italian
a child of an isolated Italian immigrant couple hears English
He is
morphemes both realized with many Italian phonological features.
to use a common phonological system with lexical
likely, like his parents,
of his peers,
alternation. But he may adopt the English phonological system
it is not uniquely
interpreting his parent's phonology as idiosyncratic, since
joined to Italian lexicon by co-occurrence restrictions.
specific learning
The rate of acquisition of different features under
is
conditions would be of great interest. In my data, semantic compounding
affecting both Li and L2. But among native speakers of
very common,
marker
French in the United States, the lexicon seems to be the conscious
borrowing occurs when
of the language being spoken, so little morpheme
which new syntactic
language is controlled by instructions. The rate at
varies considerably. Sequences which affect the "basic
rules are acquired
"25 verb - object,
grammatical relations : modifier -head, subject - predicate,
coordinate rules for
are learned very fast and learners rapidly acquire
almost always main-
representing these relations. Thus French bilinguals
English and French, and
tained a difference in noun-adjective sequence for
rapidly to use S -V -O as the
Japanese newcomers to English learn very
other hand, they
normal order in English and S-O-V in Japanese. On the
placement,
have great difficulty in maintaining separate rules for adverb
to objects
and in learning the sub-categorization of English verbs according
"he them put. "
and complements, so that they say "he puts " but never
learned
Transformations reflecting basic grammatical relations may be
secondary
faster and be more resistant to change than those reflecting
relations or subcategorizations.
s

-13-

Differences in the rate of acquisition of new rules, and the perme-

entirely
ability of old rules to convergence with the new, cannot be predicted
order rules
on the basis of contrastive analysis. The facility with which the
their
for the basic grammatical relations are learned arises either from
strategies
fundamental importance for intelligibility, or from their role in
for listening to the speech of others. In this respect, as in many others,
bilingual are very
the problems in the analysis of the process of becoming
similar to those in the study of monolingual child language acquisition.
 Footnotes

1. Susan M. Ervin-Tripp, "Language Development, " In Review of

Child Development Research, Lois and Martin Hoffman, Eds. , Vol. 2,
Russell Sage Foundation, 1967, pp. 62-63.
2. Jacqueline Strunk Sachs, "Recognition memory for
syntactic and
semantic aspects of connected discourse. " Ph. D. Dissertation, University
of California, 1965. Surface phonological features may be registered by
hearers also to classify speakers according to a system of stereotypes.
3. Eric H. Lenneberg, Biological foundations of Language, John
Wiley, 1967.
4. In second-language learning, either positive or negative transfer
may occur, or prior training may be simply irrelevant. Unfortunately,
most emphasis has been placed on negative transfer. For sophisticated
application of these psychological concepts to second language learning,
see Eugene Briere, "An experimentally defined hierarchy of difficulties
of learning phonological categories, " Language, 1966, 42, 768-796.
5. Japanese was taught in taped lessons to American
students with
structure drills based on a contrastive analysis. For a third of the sessions
between the
no gloss was given. There was no difference in pronounciation
students who first learned the Japanese sequences without the gloss, and
those who learned meanings along with the sequences. Jesse Sawyer, et al.,
''The utility of translation and written symbols during the first thirty hours
of language study, " Internat. Rev. Applied Linguistics in Language
Teachini,
1963, 1, 157-192.
6. Charles Osgood, George Suci, and Percy Tannenbaum, The mea-
surement of meaning, Urbana, Univ. Illinois Press, 1958, p. 8.
 An
7. J. J. Asher, "The strategy of the total physical response:
application to learning Russian, " Internat. Rev. Applied Linguistics in
Language Teaching, 1965, 3, 291-300.
8. Paula Menyuk, "Syntactic structures in the
language of children, "
imbedding
Child Development, 1963, 34, 407-422. In addition, sentence
increases with age, suggesting that children's "programming capacity"
of clauses
increases quantitatively. See, for example, the increasing use
Univ. of Minn.
reported by Mildred Templin in Language Skills in Children,
Inst. Child Welfare. Monogr. , No. 26, 1957, p. 94.
9. Ervin-Tripp, op. cit. , Lenneberg, op. cit. , Slobin, D. I. "The
acquisition of Russian as a native language, " In F. Smith and
G. A. Miller
(Eds. ) The genesis of language: a_psycholir.c approach. Cambridge,
M.I. T. Press, 1966.
10. Presidential address of Charles Osgood in
1963 before the
American Psycholgical Association, "On understanding and creating sen-
does
tences, " Amer. Psychol., 18, 735-751. This is a general model which
not yield as specific predictions as an input-output model.
11. Charles Welsh is a graduate student in psychology at the Univer-
of his model
sity of California, Berkeley, and has presented an outline
informally at the Institute of Human Learning. His dissertation will contain
a more fully-developed version.
12. A nasal anticipation rule is probably common in child language
Susan M. Ervin
in the second year. For another example of such a rule see
and Wick R. Miller, "Language development, " 1963 Yearbook of the National
Society for the Study of Education, p. 115.
13. James P. Thorne, H. McL. Dewar, H. Whitfield and P. Bratley,
"A model for the perception of syntactic structure, " Eng. Lang. Res. Inst. ,
 given
Univ. of Edinburgh. In this computer program, English sentences are
and
rapid syntactic interpretations, using only a dictionary of functions
syntactic category sequence predictions. This might be a hypothetical pro-
cessing model for actual perception of sentences, as the title implies.
14. For example, the dictionary might be changed first, by the
addition of diamorphs employing similar category-markers in the
translation
'equivalents.
15. The template change conveniently accounts for the rapid learning
mentioned below, such as
of certain high-frequency phrase structure rules,
S -V -O order in English. In a phenomenological analysis of learning to
comprehend Hebrew during a year in Israel, Robert Epstein, in a term
paper, reports bursts and plateaus in comprehension though vocabulary
increased at a more constant rate. Epstein suggests these bursts involve
(e. g.
shifts in "listening technique, at first involving selective attention
attention to first and last words) and later "methods of ordering the syntax
they may
of sequences. " If such sudden shifts can be objectively confirmed,
correspond empirically to the development of templates or enounter-operate
rules.
16. Susan M. Ervin, 'Imitation and structural change in children's
language, " New Directions in the Study of Language, ed. by Eric Lenneberg,
M.I. T. Press, 1964, 163-190.
17. Labov, William, "Stages in the acquisition of standard English. "
In R. Shuy, Social dialects and language learning, Natl. Council of Teachers
of English, 1964, 77-103.
18. Susan M. Ervin-Tripp, "An Issei Learns English, " Journal of
Social Issues, 1967, 23, No. 2, 78-90. On the learning of semantic shifts,
in relation to attitudes, see also Margaret J. Earle, "Bilingual semantic
 V.11,1,17/1 -

-iv-

merging and an aspect of acculturation. " J. Personality and social Psychol. ,

1967, 6, 304-312.
19. Blom, Jan-Petter and John Gumperz, "Some social determin-
ants of verbal behavior, " to be published in John J. Gumperz and Dell Hymes
Eds. , Directions in Sociolinguistics, Holt, Rinehart and Winston, 1968.
John Gumperz, "Linguistic and social interaction in two communities. " In
J. J. Gumperz and Dell Hymes, Eds. The ethnography of communication.
Amer. Anthropol. 1964, 66, Number 6, Part 2, pp. 137-153.
20. John Gumperz, "On the linguistic markers of bilingual com-
munication. " J. Soc. Issues, 1967, 23, No. 2, 48-57.
21. Blom and Gumperz, op. cit.
"If two morphemes have phonemic shape or semantic function
22.
in common, they will often be identified by bilingual speakers.... Such
semantic and morphological overlapping has been described as producing a
'compound sign'; in pursuance of my suggestion for the phonemic identifi-
cation, I shall refer to this as a diamorph. " Einar Haugen, "Problems of
bilingual description, " General Linguistics, 1955, 1, 1-9.
23. For some examples in an inflectional language, see Slobin, oR
cit. Ferguson has recently suggested that copula deletion may be a feature
shared by child speakers, baby talk to children, stereotyped speech to
foreigners, and pidgins. Charles Ferguson, "Absence of copula in normal
speech, baby talk, and pidgins. " (mimeo. )
24. Examples of such tests can be found in D. I. Slobin (Ed. ) Field
Manual for the Cross Cultural Study of the Acquisition of Communicative
Competence, ASUC Bookstore, University of California, Berkeley, 1967.
This is a draft manual to coordinate studies of first-language acquisition
and language socialization in various societies.
 ^

V.

-v

25. It has been argued that these relations apply to the deep struc-
ture of sentences, and are universal constraints on grammars. "They
supposedly describe an aspect of children's capacity for language....
Evidence exists that the basic grammatical relations are honored in chil-
dren's earliest patterned speech, if not before. " This evidence is presented
by David McNeill in "Developmental Psycholinguistics" in Frank Smith and
G. A. Miller, The genesis of language, Cambridge, M.I. T. Press, 1966,
pp. 15-84.