Overgeneralization and Lang Transfer

THE USE OF OVERGENERALIZATION AND
TRANSFER LEARNING STRATEGIES BY

ELEMENTARY AND INTERMEDIATE
STUDENTS OF ESLl
Barry P. Taylor
San kancisco State University
An orally administered test requiring the written English

translation of eighty Spanish sentences was administered to
twenty native Spanish speaking students of English as a second
language at the elementary and intermediate levels.
A taxonomy of twenty error types was designed to analyze
the errors in the Auxiliary (Aux) and Verb Phrase (VP) of the
translations. The error types were categorized into errors of
overgeneralization, transfer, translation, indeterminate origin, and
errors not considered.
The results indicated that the errors made by the elementary
and intermediate students were not qualitatively different. How-
ever, the subjects’ reliance on the stategies of overgeneralization
and transfer was found to be qualitatively different. The ele-
mentary subjects’ reliance on the transfer strategy was found to be
significantly higher than that of the intermediate subjects; the
intermediate subjects’ reliance on the overgeneralization stategy
was found to be significantly higher than that of the elementary
subjects.
These findings appear to be consistent with a theory which
considers second language acquisition to be an actively creative
process dependent upon a student’s ability to assimilate and
subsume new information into already existing cognitive
structures. The overgeneralization and transfer learning strategies
appear to be two distinctly different linguistic manifestations of
one psychological process: reliance on prior learning to facilitate
new learning. The results also tend to confirm the weakness of a
transfer-based theory of errors and require an explanation which
takes into account not only interference from within the target
language inself, but also the learner’s cognitive characteristics and
his resulting learning strategies.
In a recent article, Taylor (1974b) discusses the possibility

that the strategies of syntactic overgeneralization (with resultant
grammatical simplification) and redundancy reduction can account
‘This is a revised and expanded version of a paper presented at the 9th

Annual TESOL Convention, Los Angeles, California, March 4-9, 1975, a paper
which appeared in On TESOL 1975.
73
74 LANGUAGE LEARNING VOL. 25, NO. 1
for many of the kinds of errors made by both first and second
language learners. He also indicates that a strategy involving a
partial reliance on native language structure might be able to
account for second language errors which appear to indicate a
process of transfer. This paper discusses a study which was
specifically designed t o investigate the relationship between the
strategies of overgeneralization and transfer and the degree to
which elementary and intermediate students of ESL rely on those
strategies while learning English.
The strategy of syntactic overgeneralization can be defined as
a process in which a language learner uses a syntactic rule of the
target language inappropriately when he attempts t o generate a
novel target language utterance. Errors which seem t o reflect an
overgeneralization strategy suggest three important facts about a
learner’s knowledge of the syntax of the target language:
1. The learner has mastered the mechanics of a particular
syntactic rule of the target language.
2. The learner does not know how to use the rule appropri-
ately; i.e., he has not learned the distribution of the rule or the
exceptional cases where the rule does not apply.
3. The learner is an active participant in the language
acquisition process and is exercising his already acquired knowledge
of the target language in a creative way; he is neither operating
under a repetition or imitation strategy, nor transferring native
language structures in his target language attempts.
This research was designed to investigate how adult native
speakers of Spanish use syntactic overgeneralization and native
language transfer in the acquisition of English as a second language,
and how errors attributable to those learning strategies are related.
Specifically, this study was conducted to provide evidence to
support the following claims:
1. Second language learners of English make errors which are
not attributable to the structure of their native language and
which, therefore, cannot be predicted by the Contrastive Analysis
Hypothesis.
2. Many errors which second language learners make can be
attributed to inherent difficulties and/or irregularities in English
itself and can be explained by a strategy of target language
syntactic overgeneralization.
3. Overgeneralization errors are neither random nor idio-
syncratic, and they can be enumerated within a limited taxonomy
of error types.
OVERGENERALIZATION AND TRANSFER 75
4. Errors which appear to indicate a reliance on native

language structures (native language transfer) are more common
among elementary speakers of English than they are among
intermediate speakers. Errors which appear to be attributable to
overgeneralization are more common among intermediate speakers
than they are among elementary speakers.
There are several fundamental assumptions behind these
claims. First, if language acquisition is a creative process which
involves the learner as an active participant, then the learner will
make errors which indicate that he is dealing with the target
language directly and without an extensive reliance on his native
language.
Second, if Ausubel (1967) is correct in assuming that a
learner relies on what he already knows when confronted with a
new learning situation, because the elementary target language
learner knows less of the target language than a more advanced
learner, the elementary learner will need to rely more heavily on
his native language. As he learns more about the target language,
his reliance on his native language will decrease, and errors
attributable to target language syntactic overgeneralization will
increase.
And third, if an interlanguage is a linguistic system which
differs systematically from both the native and target languages,
then the errors made by a second language learner will be
systematic and will reflect the systematic grammar of his inter-
language. A student of a second language will frequently com-
pensate for his lack of knowledge of a set of related target
language rules (such as rules for question formation) by using a
specific target language rule (such as inserting a do in all questions)
more often than that rule is used by native speakers of the target
language. This phenomenon can be viewed as a reduction of the
complexities of the target language syntactic system and can be
defined as “target language syntactic simplification.”
EXPERIMENTAL METHODOLOGY
Error analysis is not a new way to investigate the process of

language acquisition. Just as errors in child language acquisition
have been used to explain how the child approaches the task of
learning his native language, the recent concern of error analysis
has been t o try to infer, from second language errors, the processes
and strategies under which the learner operates. In examining and
analyzing errors, however, we are frequently unable t o compare the

proficiencies of various speakers when we have no control over
those speakers’ linguistic production. Because some speakers are
more “reflective” (Brown 1973) and reserved, using only what
they are sure that they know in their attempts at speaking the
target language, while others are more “impulsive” (Brown 1973),
we cannot easily compare the errors made by various speakers (also
see Corder 1972). The fact that some speakers may make more
errors than others may indicate only that they monitor their
speech less; we cannot, therefore, conclude that their interlanguage
is less like the target language grammar than that of speakers who
cleverly avoid specific syntactic structures and, therefore, make
fewer errors.
For this reason, controlling the elicitation of specific
grammatical constructions from foreign speakers of English is
extremely difficult. In order to remedy this methodological
problem of experimentation, Corder (1972) has suggested an
elicitation procedure which requires direct translation from the
native language to the target language. This method has proved
successful in studies conducted at Edinburgh and has the ad-
vantages of 1)forcing the experimental subject to attempt t o form
a desired target language structure, and 2) assuring that the subject
understands the semantics of the structure which he is required to
produce. Moreover, by forcing a subject to form a structure which
he has not completely mastered, the experimenter can gain insights
into how the subject understands the language to operate and how
he organizes new syntactic constructions in his interlanguage. It
follows that many imperfectly controlled structures are apt to
yield examples of both overgeneralizations of target language rules
which the subject has mastered, and negative transfer from the
native language.
In accord with Corder’s suggestions (1972), direct translation
was used to elicit attempted productions of semantic information
in English. It might be argued, however, that a translation test
“loads” a study in favor of transfer and interference. Nevertheless,
because the power of the overgeneralization strategy over the
transfer strategy seemed t o be so strong, a translation test was used
in spite of its potential shortcomings. Furthermore, a translation
test seems to be the most efficient way to elicit specific syntactic
structures from subjects.
Subjects
The subjects for this study were all students in the fifteen-
week Intensive Course in English at the English Language Institute
at The University of Michigan during the fall of 1973. Twenty
native speakers of Latin American Spanish were selected from
elementary and intermediate English classes at the ELI. These
students ranged in age from 17 t o 42 years, with a mean age of
26.
The subjects fell into two convenient groups. ELI placement
test scores and independent evaluations by each subject’s four ELI
instructors were used to divide the subjects into an “elementary
proficiency group” and an “intermediate proficiency group,” with
ten subjects in each category. The elementary subjects had studied
very little or no English prior to the ELI course; the intermediate
subjects had all had prior instruction in English.
Although the distinction between “elementary” and “inter-
mediate” proficiency is rather arbitrary because the terms are not
quantifiable on an absolute scale, this distinction was important to
this study because of its validity on a relative scale. Distinguishing
between two discrete levels of proficiency in English made it
possible to isolate two stages in the acquisition of English as a
second language. Because this study is concerned not only with
comparing the relationship between the strategies of overgeneraliza-
tion and transfer, but also with investigating whether a reliance on
those strategies is dependent upon a learner’s degree of proficiency
in English, this division of subjects was necessary.
Materials and Testing Procedure

The test which was designed t o investigate syntactic over-
generalization and transfer phenomena in English consists of eighty
Spanish sentences recorded by a native speaker of Mexican
Spanish. The test was administered in two sittings, at least one day
but no more than four days apart. At the first sitting the subjects
heard fifty of the test sentences in Spanish, preceded by five
pre-test sentences provided t o offset any learning effect. At the
second sitting the remaining thirty sentences were heard, again
preceded by five pre-test sentences. Each sentence was heard twice,
followed by a thirty-second period of silence during which time
the subjects were asked t o write the translation of the sentence in
correct Enghsh. The use of immediate translation limited the time
that the subjects had to reflect on the problem and required an

immediate, first-impression response. If the subjects had not
mastered the syntax of the structure, it was assumed that their
responses would reflect either native language syntax or an over-
generalization of an English structure which they had mastered.
The subjects were instructed t o respond in writing rather than
orally because a written response is easier to evaluate than an oral
response which might be marked with numerous false starts,
hesitations, “backtracking,” and other performance variables. While
it might be argued that writing is a different kind of activity from
speaking, the test sentences were sufficiently straight-forward to
suggest that identical abilities would be tested regardless of the
modality used. The stimulus sentences contained only common
vocabulary which the subjects could be expected t o know; there-
fore, the test emphasized syntactic rather than lexical translation.
In addition, the subjects were told that both spelling and vocabu-
lary were unimportant, and translations of vocabulary items which
might prove difficult were provided on the answer sheet; main
verbs were given in their infinitive forms in Spanish and English,
and nouns were provided as they appeared in the sentences, with
the accompanying English translation. Each response was scored
and evaluated only on the basis of the one syntactic point being
investigated, rather than on the complete acceptability of the
response.
The eighty test sentences were written to test the subjects’
mastery of the Auxiliary (Aux) and Verb Phrase (VP) in eight
sentence types of English. Sentence types which might cause the
subjects either to tranfer Spanish syntax into their translations or
to overgeneralize because of the essential arbitrariness and/or
complexity of the distribution of the English structure were
selected. The following are the eight sentence types of English
which were tested:
1. Simple active declarative statements
2. Yes-no questions
3. Negative statements
4. Negative yes-no questions
5 . Subject-focus wh-questions
6. Object-focus wh-questions
7. Negative subject-focus wh-questions
8. Negative object-focus wh-questions
The translations of the Spanish sentences of the above types
required mastery of verbs (other than be) in the present and past
tenses, and with the present tense modals will and can. Both
singular and plural VP’s were tested: the subjects were either
common nouns (e.g., the student, the boys) or proper nouns (e.g.,
John, Mr. and Mrs, Miller, Pablo and Elsa); no subject pronouns
were used. The wh-questions required mastery of the English
wh-words who and what in both subject and object positions and
were used with verbs other than be in the present and past tenses,
and with the present tense modals will and can, all in both the
singular and plural. The only deviation from this strict paradigm of
structures which was established to ensure that every stimulus
sentence differed structurally from every other stimulus sentence
by at least one feature was the omission of subject who and what
in the plural. Because subject who and what d o not occur in the
plural for most speakers of English (e.g., Who studies here?, What
has four legs? vs. *Who study here?, *What have four legs?),those
structures had t o be omitted from this paradigm.
Table 1 illustrates the structures which were tested. The full
text of the test can be found in Appendix I. The test sentences
(represented in the Appendix in both English and Spanish) are in a
linguistic ordering that groups sentences together according to
syntactic properties. The eighty test sentences and the ten pre-test
sentences were administered in random order using a standard table
of random numbers. In addition, in order to offset the possibility
of a “fatigue effect” altering the results, some of the subjects
received the test in random order, and the other subjects, in reverse
random order. In the table, the number preceding each structure
represents the position of that sentence in the linguistic ordering.
In the Appendix, the numbers refer t o the random order.
The test was administered during the seventh week of the
fifteen-week course, so all of the structures contained in the test
had already been explained and drilled in the subjects’ ELI classes.
However, it was assumed that although a specific structure may
have been presented and practiced, the subjects had not, neces-
sarily, mastered it. As Corder (1967:165) says,
The simple fact of presenting a certain linguistic form t o a learner
in the classroom does not qualify it for the status of input, for the
reason that input is “what goes in” not what is available for going
in, and we may reasonably suppose that it iis the learner who
controls his input, or more properly his intake. This may well be
determined by the characteristics of his language acquisition
mechanism and not by those of the syllabus. After all, in the
mother-tongue learning situation the data available as input is
relatively vast, but it is the child who selects what shall be the
input.
00
TABLE 1 TABLE 1--Continued 0
Structures tested
sentence non-& pres V non-& past v nodal+V sentence non-& pres V non-& past V modal+V
types types
3 pl sb: 3 pl sb: 3 pl sb:

fsb-fs:& *sb-fsie *sb-fs:uh*will
*sb-fs:uho+can
*sb-fs:- *sb-fs:- fsb-fs:uhattwill
fsb-fs:uhat+can
Questions (9) 3 sg sb (11) 3 sg sb (13) cant3 "3 sb (49) ob-fs:- (51) ob-fs:* 53 ob fs:uhomtuill
14 -+3 s(: sb {5k] o b I f s : G G
(10) 3 pl sb (12) 3 pl sb 15 =+3 PI :b (50) ob-fs:& (52) ob-fs:@ (55) ob-fs:uhat+will
[16/ a + 3 pl sb (56) ob-fs:uhat+z
Negatives (17) 3 sg sb (19) 3 sg sb (21) 3 sg sbt& Negative 3 sg sb: 3 sg sbi 3 sg sb:

(22) 3 sg sbt& Wh-auestions
(18) 3 pl sb (20) 3 pl sb (23) 3 pl sM.& sb-fs:@ sb-fs:* sb-fs:uho+uon't
(24) 3 pl sbt& sb-fs:uhotcan't
sb-fs:w sb-fs:M sb-fs:uhat+uon't
sb-fs:what+can't
Negative (25) 3 sg sb (27) 3 sg sb ob-fs:- ob-fs:-
questions
(26) 3 pl sb (28) 3 pl sb ob-fs:m ob-fs :*
(72) ob-fs:uhattcan't
3 p l sb: 3 pl sb: 3 p l sb:

Wh-questions 3 sg sbi 3 sg sb: 3 sg sb:
*sb-fs:& ~~
*sb-fs:w& *sb-fs:uhotwon't
sb-fs:@ sb-fs:@ sb-fs:uhotuill fsb-fs:uho+can't
sb-fs:uhotcan *sb-fs:s *sb-fs:M *sb-fs:uhat+uon't
sb-fs:W sb-fs:S sb-fs:uhat+uill *sb-fs:what+can't
sb-fs:what+can (79) ob-fs:- (75) ob-fs:&
ob-fs:= ob-fs:- ob-fs:uhoni+will
ob-fs:uhomtcan (74) o b - f s : W (76) ob-fs:=
ob-fs:w ob-fs:m ob-fs:what+ulll ob-fs:uhattcan't
ob-fs:uhattcan
>third person sbsubject ob=object *not acceptable in Standard English

sg=singx~ar pl=plural fs=focus
THE ERROR ANALYSIS
The translations of the eighty Spanish sentences by the

twenty subjects yielded a corpus of 1600 English sentences.
However, because two of the eighty sentences (#18 and #59) had
to be omitted for technical reasons after the administration of the
test, the corpus was actually 1560 sentences. In the analysis of the
corpus only the Auxiliary (Aux) and the Verb Phrase (VP) were
examined for errors. An error was defined as any mistake from a
native speaker’s point of view (discounting spelling errors or
incorrect forms of the past tense of a main verb) generated by the
misapplication of one or more of the structural rules of English.
No other errors were considered. For example, the sentence What
guests can to bring to the party? contains two errors. The first
error is the lack of subject-modal inversion in a question, and the
second is the use of to after the modal auxiliary can. The omission
of the article the before the subject guests is an error to be
ignored.
A Taxonomy of Error Types
Five categories of errors were established: overgeneralization

errors, transfer errors, translation errrors, errors of indeterminate
origin, and errors not considered. Each of these five categories was
subdivided to form a total of twenty error types. Most of the
twenty error types have more than one way of being manifested in
the subjects’ sentences. Appendix I1 presents the full taxonomy of
error types, a discussion of how that taxonomy was used in the
error analysis, and the rationale for grouping the error types into
the five categories.
Results and Discussion

The data collected and analyzed for this study did not
provide any clear support for the claim that either overgenera-
lization or transfer errors would be characteristically different for
the elementary versus the intermediate subjects. Some trends were
apparent, however. For example, only elementary subjects used be
in an “all-purpose auxiliary function” to replace models or d o , and
they had a distinct preference for unmarked verb forms (using
unmarked verbs to fill an “all-purpose verb function”). Only
intermediate subjects exhibited a consistent effort t o hypercorrect
verbal endings (notably their use of -s on plural verbs), and their

efforts to hypercorrect question-formation (by inserting a
redundant do when a modal was present) -were more notable than
the elementary subjects’ less frequent attempts t o do so.
However, the large number of similarities in the error types
which were most common for the elementary and intermediate
subjects tends t o indicate that increased proficiency in English does
not qualitatively affect the kinds of errors which a learner makes.
While the intermediate subjects made fewer errors in almost every
error type, their most frequent errors were usually also the errors
which the elementary subjects made most frequently. Errors of
subject placement in questions and the insertion of to after can
were among the most common transfer errors for both elementary
and intermediate subjects. Similarly, the lack of subject-modal
inversion in questions was almost as common for the two groups
of subjects, with that error being the most common elementary
overgeneralization error, and the second most common inter-
mediate overgeneralization error.
Although the results obtained in this study provide a good
deal of counter-evidence to the claim that intermediate subjects
overgeneralize or employ transfer strategies in characteristically
different ways, there is considerable evidence in the data to
support the claim that increased proficiency in English results in
the use of these two learning strategies to different degrees. While
overgeneralization and transfer errors may not be qualitatively
different for elementary and intermediate language learners, they
were found to be quantitatively different.
Table 2 lists the number of errors made in each sentence type
by the elementary and intermediate subjects, and it divides those
errors into those attributable t o overgeneralization and transfer.
For example, of the 29 overgeneralization and transfer errors made
in statements by elementary subjects, 18 were attributable to
overgeneralization and 11 to transfer.
Table 3 presents the figures from Table 2 as proportions. It
compares the ratios of the total number of overgeneralization and
transfer errors made by the elementary and intermediate subjects
in the eight sentence types with the distribution of those errors
between overgeneralization and transfer. For example, .62 of all
overgeneralization and transfer errors made in statements by
elementary subjects could be attributed to overgeneralization.
An examination of the overgeneralization columns in Table 3
indicates that for every sentence type intermediate subjects made a
higher proportion of errors attributable to overgeneralization than
TABLE 2
Number of errors made b y elementary ( E ) and intermediate (I)
subjects in the eight sentence types in
overgeneralization and transfer
overgeneral- trans-
ization fer total
sentence type E I E I E I
Statements 18 9 11 0 29 9
Q u e s ti o n s 32 9 35 4 67 13
Negatives 36 3 19 1 55 4
Negative q u e s t i o n s 51 13 34 5 85 18
S u b j e c t Wh-questions 52 20 5 0 57 20
Wb j e c t Wh-ques t i o n s 60 37 132 32 192 69

Neg s u b j Wh-questions 41 12 11 1 52 13
*Neg o b j Wh-questions lo5 33 101 26 206 59

~~~
*There were eight stimulus sentences in each sentence-type category except the
obj Wh-questions (affirmative and negative) in both of which there were sixteen
stimulus sentences. For a more realistic comparison with the other sentence
types, the numbers in those two types should be divided by two.
did the elementary subjects. A comparison of the means of these

proportions also supports this result. These findings confirm the
hypothesis discussed earlier that intermediate subjects rely more
heavily on an overgeneralization strategy than do elementary
subjects. Indeed, when this hypothesis was tested for significance
by means of a t-test of difference between proportions, the
differences were found to be significant (t=1.776, p<.05 one
tailed).
The transfer columns in Table 3 indicate that for all sentence
types the proportion of elementary errors attributable t o transfer
from Spanish exceeds the proportion of intermediate transfer
errors. When the t-test was applied to these results, it was found
that the transfer strategy was used significantly more often by the
elementary subjects than by the intermediate subjects (t=l.776,
TABLE 3
Proportion of elementary ( E ) and intermediate ( I ) instances of
errors in the eight sentence types t o the total number of
elementary and intermediate errors per sentence
t y p e in overgeneralization and transfer and
results of t-tests
overgeneralization transfer
sentence type E I E I
Statements .62 1.00 9 38 0.00

Questi o n s .48 69 @ 52 9 31
Negatives 9 65 9 75 35 25
Negative q u e s t i o n s .60 72 .40 .28
S u b j e c t Wh-questions 91 1.00 * 09 9.00
Object Wh-questions 31 54 .69 .46
Neg sub j Wh-questions 79 92 .21 .08
Neg o b j Wh-questions 51 56 49 .44
mean .61 77 39 23
t 1.776 * 1.776"
*p<.05 one tailed
H . 0 5 one tailed). These results also confirm the hypothesis

discussed earlier that transfer is a more prevalent strategy among
elementary students than among intermediate students. The major
findings of this study are, then, that reliance on overgeneralization
is directly proportional t o proficiency in the target language, and
reliance on transfer is inversely proportional. That is, as a learner's
proficiency increases he will rely less frequently on his native
language and on the transfer strategy, and more frequently on
what he already knows about the target language and on the
overgeneralization strategy. Figure 1 illustrates the relationship
between the proportion of elementary and intermediate errors in
both overgeneralization and transfer.
A word of warning on the significance of the proportions of
overgeneralization versus transfer errors listed in Table 3 is
necessary. While it is possible to compare the elementary and
intermediate overgeneralization errors, and the elementary and
.80
75
70
65
.60
* 55
rn
8 -50
k
k
Q, .45
(H
0
UY
.40
c
0
+ *35
.d
k
0
-
fi
0
.30
k
fi
-
(H 925
0
; .20 overgeneralization
cd
.15 transfer
.10
05
.00
Elementary Intermediate
subjects subjects
Figure 1 . The means of the proportions of overgeneralization

and transfer errors made b y the elementary and intermediate subjects.
intermediate transfer errors, we should be careful not to compare

across columns and attempt to compare overgeneralization and
transfer. Since the figures represented in the overgeneralization and
transfer columns represent different kinds of errors, comparing one
to the other is similar to comparing dissimilar phenomena. As G.
Richard !Tucker has pointed out (personal communication), a test
designed to allow subjects a free choice between making an
overgeneralization or transfer-type error in their responses would
have needed to be conducted before we could say that over-
generalization errors were more common than transfer errors. While
that result certainly appears to be the case, it is not a statistically
supportable conclusion to be drawn from the data which is
presented in this study.
CONCLUSIONS
Several important conclusions can be drawn from this in-

vestigation into the relationship between the learning strategies of
syntactic overgeneralization and transfer, and the reliance on those
strategies by elementary and intermediate language students.
The results reported here tend to confirm the weakness of a
transfer-based theory of errors. The findings indicate that a large
number of errors which second language learners make can be
explained only within a framework which takes into account
interference from within the target language itself. The errors
which stem from that source have been referred to here as
overgeneralization errors, and these errors have been explained as
attempts by the learner to simplify and regularize the linguistic
complexities peculiar to the target language.
One of the major characteristics of the overgeneralization
strategy is that it results in a simplification of the syntactic system
of the target language. For example, when a learner generates the
sentence He study there every night he demonstrates that his rule
for present tense formation involves using a zero-morpheme to
mark number for all persons. When he produces the sentence Did
they studied last night? he indicates that his rule says that because
the sentence is past, all verbs are in the past. When he says Does
she can cook well? he tells us that his rule for question-formation
requires that every question contain a d o .
These three sentences illustrate three types of overgeneraliza-
tions which all result in a simplified grammar of the target
language. The target language rule which says that all present tense
verbs for all persons except third person singular require a zero-
morpheme, but third singular requires -s, has been reduced to a
rule which requires a zero-morpheme for all persons. The target
language rule which requires that a verb in the past tense be in the
simple form when it follows a modal or d o in a question has been
simplified to one which requires that a verb in the past tense
always carry the tense marker. The target language rule which
requires that d o be inserted in questions only if there is no other
auxiliary in the sentence has been reduced to one which requires
that every question contain a d o .
The learner’s interlanguage rules which produce unacceptable
target language utterances seem t o stem from his cognitive
characteristics and the resulting psychological learning strategies
which he uses to acquire the target language. It has been suggested
throughout this study that the principal motivation behind the
learning strategies which the learner brings t o language acquisition
is the desire to reduce his learning burden. Both the strategies of
native language transfer and of overgeneralization will make the
learner’s task easier: when he relies on his native language he
avoids learning the target language rule; when he overgeneralizes he
relies on a target language rule of great generality and which he
already knows and avoids learning the appropriate rule.
Within this framework, overgeneralization and transfer learn-
ing strategies appear to be two distinctly different linguistic
manifestations of one psychological process. That process is one
involving reliance on prior learning t o facilitate new learning.
Whether transfer or overgeneralization will be the dominant
strategy for a given learner will depend on his degree of pro-
ficiency in the target language.
Ausubel (1967) has claimed that successful long-term learning
proceeds by a reliance on previous cognitive experience (i.e., prior
learning). If we assume that, linguistically, the only “previous
cognitive experience” which the elementary subjects had was their
knowledge of Spanish, then it is easy to understand why the
elementary subjects relied so heavily on the transfer strategy. The
intermediate subjects, however, by virtue of having learned a
considerable amount of English, had, linguistically, a broader
previous cognitive experience and could rely on their already
acquired knowledge of English in attempting to generate acceptable
English utterances. It is easy t o see, then, why their reliance on the
overgeneralization strategy was so much greater than that of the
elementary subjects.
From the point of view of language learning, these results

appear to indicate that learners of a second or foreign language
begin relying on their ability to analogize, systematize, and
regularize the target language data to which they are exposed
immediately upon beginning to learn the new language. Because of
their lack of familiarity with the new linguistic system, however,
they also rely extensively on their native languages for support.
With increased proficiency in the target language, they rely pro-
portionately less frequently on their native language grammar, and
rely more frequently on their ever-increasing knowledge of the
target language, coping directly with it and overgeneralizing its
rules.
REFERENCES
Ausubel, David P. 1967. Learning theory and classroom practice. Toronto:
The Ontario Institute for Studies in Education.
Brown, H. Douglas. 1972. Cognitive pruning and second language acquisition.
Modern Language Journal 56.21 8-222.
Corder, S. P. 1967. The significance of learner’s errors. International Review
of Applied Linguistics 5.161-170.
Corder, S. P. 1972. The elicitation of interlanguage, unpublished manuscript,
University of Edinburgh.
Dulay, Heidi C. and Marina K. Burt. 1972. Goofing: an indicator of children’s
second language learning strategies. Language Learning 22.235-252.
Jain, M. P. 1 9 6 9 . Error analysis of an Indian English corpus, unpublished
manuscript, University of Edinburgh.
Newmark, Leonard and David A. Reibel. 1968. Necessity and sufficiency in
language learning. International Review of Applied Linguistics
6.145-164.
Richards, Jack C. 1971a. Error analysis and second language strategies.
Language Sciences 17.12-22.
Richards, Jack C. 1971b. A non-contrastive approach t o error analysis. English
Language Teaching 25.204-219.
Selinker, Larry. 1 9 7 2 . Interlanguage. International Review cf Applied
Linguistics. 10.209-231.
Taylor, Barry P. 1974a. Overgeneralization and transfer as learning strategies
in second language acquisition, unpublished doctoral dissertation, The
University of Michigan.
Taylor, Barry P. 1974b. Toward a theory of language acquisition. Language
Learning 24.23-35.
Wardhaugh, Ronald. 1970. The contrastive analysis hypothesis. TESOL
Quarterly 4.123-130.
Wolfe, David L. 1967. Some theoretical aspects of language learning and
language teaching. Language Learning 17.173-188.
APPENDIX I
THE TEST
The following is the entire text of the test which was presented t o the
twenty subjects. The test was played t o t h e subjects on a tape recorder and
was entirely in Spanish.
Directions
This is a test to determine how well you know English. You are going
t o hear * sentences in Spanish. Translate each sentence into English in
the spaces provided o n your answer sheets. You wit1 have only thirty seconds
to translate each sentence. Therefore, you must work quickly. Be sure t o
translate the sentences into correct English and be sure n o t t o simply give a
word-for-word translation. If the sentence is a question, be sure t o translate it
as such, and d o not try to answer it.
In t h e margins of your answer sheets you will find translations of some
vocabulary which you might find difficult. In all cases, verbs are,given in the
infinitive form. When you write your sentences, be sure t o use the correct
verb form for the English sentence, and do not simply copy the translation of
the infinitive form.
This is not a test of vocabulary or spelling. If there is a word that you
cannot translate, raise your hand. If there is a word that you cannot spell,
guess.
Please write clearly.
T h e Sentences
he-test items
The following ten sentences are in random order. The first five were
presented o n the test tape in Spanish for the subjects to translate at the first
sitting of this test before the fifty test items for that sitting were presented.
The second five were presented before the thirty items for the second sitting.
All sentences were recorded in Spanish and the subjects were n o t aware that
they were n o t to be scored or evaluated.
1. ‘Estudiarh Carlos la leccion?
Will Carlos study the lesson?
2. Los estudiantes n o tienen dinero.

The students don’t have any money.
3. Carmela no preguntb al professor.

Carmela didn’t ask the professor.
4. LA q u i h puede Elsa invitar?
Who(m) can Elsa invite?
*first sitting-50 sentences and 5 pre-test sentences

second sitting-30 sentences and 5 pre-test sentences
5. iQuiPn n o estudi6 10s verbos?

Who didn’t study t h e verbs?
6. iVive Alfredo en Venezuela?

Does Alfredo live in Venezuela?
7. El senor Garci’a trabaja muy duro.

Mr. Garcia works very hard.
8. El senor Martihez vivi6 en Caracas.

Mr. Martinez lives in Caracas.
9. i N o escribio Rafael la carta?

Didn’t Rafael write t h e letter?
10. ‘Compraron tus hijos la casa?

Did your sons buy t h e house?
Test sentences
The following sentences, here presented in English and Spanish, were
recorded in Spanish by a native speaker. Each sentence was read twice, and
the subjects were asked t o translate each o n e into Spanish during the
thirty-second interval between items.
The sentences are presented here in a linguistic ordering. The number to
the left of each sentence is t h e position of t h at sentence after the linguistic
order was randomized. At t h e first sitting for this test, half of t h e subjects
heard t h e five pre-test sentences (1-5) and numbers 1-50 of the following
sentences. At t h e second sitting, they heard pre-test items 6-10 and test items
51-80. T h e other half of t h e subjects heard t h e first five (1-5) pre-test items
and numbers 8 0 - 3 1 a t t h e first sitting, and pre-test items 6-10 and test items
30-1 a t t h e second sitting. Half of the subjects, then, received the test in
random order, and t h e other half, in reverse random order.
41. El senor Martinez habla inglPs muy rapido.
Mr. Martinez speaks English very quickly.
13. Los estudiantes estudian e n la biblioteca cada noche.

T h e students study in t h e library each night.
67. El senor Garcia escribi6 la carta anoche.

Mr. Garcia wrote t h e letter last night.
46. JosP y Carlos hablaron conmigo aver.

Jose and Carlos talked with me yesterday.
5. Pablo saldrd a las o ch o manana.

Pablo will leave a t 8 o’clock tomorrow.
29. Juan puede nadar bien.

J o h n can swim well.
22. Los hombres estardn ah i a las ocho.

The men will be there a t 8 o’clock.
17. Los ninos pueden venir ahora.
The children can co me now.
28. ‘Estudia Margarita e n el ELI?

Does Margarita study a t t h e ELI?
62. ‘Cornen Enrique y Rafael a las seis todos 10s dias?

Do Enrique and Rafael eat a t 6 o’clock every day?
75. iEstudi6 Javier la lecci6n anoche?

Did Javier study t h e lesson last night?
24. ‘Prepararon las mujeres el desayuno?

Did the women make ( t h e ) breakfast?
6. ‘Puede el director hablar conmigo ahora?

Can the director talk with m e now
47. ‘Sabra Alejandro la respuesta manana?

Will Alejandro know t h e answer tomorrow?
38. ‘Pueden 10s estudiantes hablar inglds?

Can the students speak English?
80. ‘Vendran Margarita y Carlos a la fiesta?

Will Margarita and Carlos come to the party?
21. Roberto no camina a clase todos 10s dias

Robert doesn’t walk t o class every day.
76. (omitted for technical reasons)
70. Ricardo no tenia las entradas.

Ricardo didn’t have the tickets.
42. Los chicos no durmieron bien anoche.

The boys didn’t sleep well last night.
56. Enrique no puede comprar la guitarra.

Enrique can’t buy t h e guitar.
15. Juan no ira a1 cine contigo.

J o h n won’t go to t h e movies with you.
36. Roberta y Elsa n o pueden cantar bien.

Roberta and Elsa can’t sing well.
32. Los muchachos n o verdn la televisi6n esta noche.

The boys won’t watch television tonight.
60. ‘No habla Gilberto franc& bien?

Doesn’t Gilberto speak French well?
51. ‘No caminan las chicas a clase todos 10s dias?

Don’t t h e girls walk to class every day?
18. ‘No habl6 Alicia con el professor?

Didn’t Alicia speak with t h e professor?
44. i N o fueron 10s abogados a la fiesta?

Didn’t the lawyers go to the party?
54. i N o puede Juan venir pronto?

Can’t John come soon?
74. i N o estarl el senor Martinez aqui temprano?

Won’t Mr. Martinez be here early?
30. i N o pueden 10s estudiantes terminar la lecci6n hoy?

Can’t the students finish the lesson today?
14. ‘No leerln Luis y Nelly el peribdico?

Won’t Luis and Nelly read the newspaper?
8. iQui6n conoce a este hombre?
Who knows this man?
45. iQuC tiene tres ojos?

What has three eyes?
77. ‘A quiCn ve el senor Briceno cada dia?
Who(m) does Mr. Briceno see every day?
79. iQuB quiere decir “circular?”

What does “circular” mean?
35. iQuidn escribi6 este libro?

Who wrote this book?
69. iQuC se termin6 anoche?

What ended last night?
39. i A quiCn visit6 el senor Martinez?

Who(m) did Mr. Martinez visit?
23. iQuC pregunt6 el professor?

What did the professor ask?
59. iQuiCn vendrri tarde?

Who will come late?
34. LQuiCn puede hablar franc&?

Who can speak French?
52. iQu6 rompera?

What will break?
37. iQue puede pasar ahora?

What can happen now?
63. ‘A quien invitara Jacinto?
Who( m) will Jacinto invite?
71. ‘A quiCn puede Angela ver?
Who(m) can Angela see?
72. iQu4 dira Antonia despuCs de la fiesta?

What will Antonia say after the party?
43. iQuC puede Erlando leer en ingles?
What can Erlando read in English?
65. i A quien imitan 10s estudiantes?

Who(m) do the students imitate?
7. iQu4 leen Pablo y Jacinto cada manana?

What do Pablo and Jacinto read every morning?
20. i A q u i h persuadieron Maria y Elsa?

Who(m) did Mary and Elsa persuade?
66. iQuB comprendieron 10s nitios?

What did the children understand?
27. i A quidn preguntaran 10s abogados?

Who(m) will the lawyers ask?
53. i A quiCn pueden 10s medicos visitar?

Who(m) can the doctors visit?
61. iQu8 comeran el senor y la senora Veldsquez?

What will Mr. and Mrs. Velasquez eat?
78. iQuC pueden 10s hudspedes traer para la cena?

What can the guests bring for dinner?
73. iQui6n no sabe la respuesta?

Who doesn’t know the answer?
1. iQu6 no funciona aqui?
What doesn’t work here?
64. (omitted for technical reasons)
48. iQuC no necesita Rafael?

What doesn’t Rafael need?
16. ‘Quien no estudi6 la lecci6n?

Who didn’t study the lesson?
68. iQu6 no llegb en el correo hoy?

What didn’t arrive in the mail today?
26. i A q u i h no invit6 tu hermano?

Who(m) didn’t your brother invite?
40. iQuC no ley6 Ernesto?

What didn’t Ernesto read?
9. iQui4n n o estarri aqut a las diez?
Who won’t be here at 10 o’clock?
57. iQui6n no puede nadar bien?

Who can’t swim well?
4. iQuP no cambiari?
What won’t change?
10. iQu6 no puede volar?

What can’t fly?
19. i A quiCn no veri Ra61 manana?
Who(m) won’t Raul see tomorrow?
58. i A q u i h no puede el senor Martinez visitar hoy?

Who(m) can’t Mr. Martinez visit today?
11. iQuP no comeri el nino?

What won’t the child eat?
25. iQud no puede Laura comprender?

What can’t Laura understand?
33. i A quiPn no conocen Beatriz y Margarita?

Who(m) don’t Beatrice and Margarita know?
49. iQuP no tienen el senor y la senora Williams?

What don’t Mr. and Mrs. Williams have?
2. i A quiPn no preguntaron 10s estudiantes?

Who(m) didn’t the students ask?
3. iQuP no estudiaron las ninas?

What didn’t the children study?
50. ‘A quiPn no invitarin 10s profesores?

Who(m) won’t the professors invite?
1 2 . i A q u i h no pueden Maria y Nelly oir?

Who(m) can’t Mary and Nelly hear?
31. ‘QuP no practicarh 10s jbvenes?

What won’t the boys practice?
55. iQuP no pueden 10s estudiantes pronunciar?
What can’t the students pronounce?
APPENDIX I1
THE TAXONOMY OF ER R O R TYPES
Overgeneralization Errors
I. Redundant au x insertions
a. Insertion of d o in a sentence containing a modal
b. T w o do’s in a negative question; 1st: pres tense conjugated
2nd: simple + not
c. Two do’s in a negative question; 1st: pres tense conjugated
2nd: past tense + not
d. Insertion of d o in a subject-focus Wh-question
e. Insertion of be in a subject-focus Wh-question (with the verb usually
in t h e past participle f o r m)
f. Insertion of be in statements and subject-focus Wh-questions (with
t h e main verb in t h e simple form)
11. Aux substitutions

a. Insertion of be instead of d o
b. Insertion of be instead of will
c. Insertion of be instead of can
d. Use of d o + g o t o + V instead of will
will
e. Use of {would $
’ go to + V instead of will
f. Use of will going to + V instead of will
111. Incorrect placement of negative marker

a. not placed after a tensed verb when n o d o has been inserted
b. not attached to do (inserted a t t h e beginning of a question) in a
question containing a modal t o which t h e not should be attached
IV. Incorrect form of main verb following an auxiliary
a. Past tense form of verb following a modal
b. Present tense -s o n a verb following a modal
c. -ing o n a verb following a modal
d. are (for b e ) following will
e. Past tense form of verb following d o
f. Present tense -s o n a verb following d o
g. -ing o n a verb following d o
h. Past tense form of verb following be (inserted t o replace a modal or
do)
i. Present tense -s o n a verb following be (inserted to replace a modal
or d o )
V. Errors in t h e use of t h e infinitive marker to

a. Use of t o + V after will
b. Use of to + V after d o
c. Use of to + V after be (inserted to replace a modal or d o )
VI. Question-inversion errors
a. Lack of subject-modal inversion
b. Lack of subject-do inversion
c. Lack of subject-modal inversion in a question in which d o has been

inserted at the beginning
d. Lack of subject-modal inversion in a question in which the modal
has been left out and the modal + V has been replaced by to + V
e. Lack of subject-modal inversion when the modal has been left out
f. Lack of subject-do inversion when d o has been left out
g. Lack of subject-second d o inversion when a do has been inserted at
the beginning of a question, and a tensed and negated d o follows
the subject and precedes the verb
h. Lack of subject-be inversion when be has been inserted to replace
do
VII. Verb tense errors

on Aux:
a. do, present instead of past, when the sentence contains an un-
ambiguous adverbial time reference
on Main Verb:
b. Main verb, present instead of past, in statements and subject-focus
Wh-questions, when the sentence contains an unambiguous adverbial
time reference
c. V + ing (without any be in the sentence) instead of conjugated
present tense
VIII. Verb number errors

on Aux:
a. d o in singular instead of plural (unnecessary - s )
b. d o in plural instead of singular (no - s )
c. d o in singular instead of plural (unnecessary - s ) when it is in the
present tense instead of the past tense
d. d o in plural instead of singular (no =s) when it is in the present
tense instead of the past tense
e. do, in simple form + not, inserted before a verb, when the subject is
singular and there is a correctly conjugated do inserted a t the
beginning of the question
f. be in singular instead of plural when it has been inserted t o replace
do
on Main Verb:
g. Main verb, plural instead of singular (no - s ) , in statements and
subject-focus Wh-questions
h. Main verb, singular instead of plural (unnecessary s),in statements
and subject-focus Wh-questions
i. Main verb, plural instead of singular (no -s), when the modal is
omitted from the sentence and the main verb must carry tense and
number (only in affirmative statements-not in negatives (except if
the main verb is be) or in questions)
j. Main verb, plural instead of singular (no -s), in statements and
subject-focus Wh-questions when the verb is present instead of past
Transfer Errors
IX. Do-insertion errors
a. Lack of insertion of d o in negatives and questions, with the tense
(-s,-ed) left on the main verb
b. Lack of insertion of do in negatives and questions, with the main
verb in the simple form
c. Lack of insertion of d o in negatives and questions, with the wrong

tense (-s, - e d ) on the verb
X. Errors in the use of the infinitive marker to

a. Use of to + V after can
XI. Incorrect placement of the negative marker

a. not placed before the simple form of the verb in a question, and
not on the d o that is at the beginning
b. not placed before will
XII. Word-order errors

a. Placement of subject after the c.ain verb instead of after d o in
questions
b. Placement of subject after the main verb instead of after the modal
in questions
c. Placement of subject after the main verb instead of after d o (which
was not inserted) in questions
d. Placement of subject after the main verb instead of after the modal
(which was not inserted) in questions
e. Placement of subject after the main verb rather than after be (which
was not inserted to replace d o ) in questions
f. Placement of subject after the main verb instead of at the beginning
in statements
Translation Errors
XIII. Aux substitutions
a. Use of d o instead of will or can
b. Use of a modal (will o r can) instead of do
c. Substitution of one modal for another (among these are the use of
will for can, would for will, might for can, can for will, may for
will, and should for can)
XIV. Negative substitutions and omissions

a. Lack of not-insertion in a desired negative sentence
b. Insertion of not in a desired affirmative sentence
XV. Main verb omissions

a. Omission of a main verb following d o
b. Omission of a main verb in a sentence containing can, and with a do
inserted at the beginning
c. Omission of a main verb following a modal
XVI. Modal/tense substitutions

a. Main verb, in the simple present tense, instead of modal + V
b. Main verb, in the past tense, instead of modal + V
c. Main verb in the simple form instead of modal + V (simply an
omission of the modal)
XVII. Tense substitutions

on Aux:
a. d o , present instead of past, when the sentence contains either no or
an ambiguous time reference
b. d o , past instead of present, when the sentence contains either no or
an ambiguous time reference
on Main Verb:
c. Main verb, present instead of past, in statements and subject-focus
Wh-question, when the sentence contains either no or an ambiguous
time reference
Errors of Indeterminate Origin

XVIII. Incorrect use of the infinitive marker to
a. Use of to + V in a sentence in which do has not (but should have)
been inserted
b. Use of to + V instead of a conjugated verb
c. Use of to + V instead of modal + V
XIX. Incorrect placement of the negative marker

a. Insertion of not after the main verb instead of after the modal in
the sentence
Errors Not Considered

XX. Among these are the following:
Incorrect translation of vocabulary
Incorrect past tense forms
Misplacement/misuse of adverbs
Incorrect use of prepositions
Errors in verb forms not being investigated
Omission of subjects
Change in subject number from singular to plural or from plural to
singular
Discussion of the Taxonomy of Error Types

Throughout the analysis i t was assumed that the subjects had attempted
to generate acceptable English sentences, and their productions were con-
sequently evaluated as such. If, for example, a subject changed the subject of
a sentence from singular t o plural (XX-error type X X s e e above), that
sentence was analyzed as an attempted plural sentence, although it was not a
correct translation of the singular Spanish stimulus sentence. To ensure
reliability in the assignment of specific error types t o individual sentences in
the corpus, the entire corpus was analyzed two separate and independent
times. To achieve uniformity in this analysis and to account for all Aux and
W errors, a number of arbitrary decisions occasionally had to be made. The
following principles were used to guide the analysis in such circumstances:
1. If a modal was omitted from a statement, that omission was noted as
an error (XVI), and the main verb was expected to carry the correct tense and
number (if not, then VII and/or VIII or XVII).
2. If a do or be was inserted into a subject-focus Wh-question, that
insertion was noted as an error (I). The do or be was, in addition, expected to
carry the correct tense and number (if not, then VII and/or VIII o r XVII),
and the verb was expected to be in the simple form (if not, then IV).
(Although the insertion of a do may still result in an acceptable sentence with
a verbal emphasis, it was assumed that the subjects were not aware of this
transformation. Furthermore, none of the Spanish sentences had verbal
emphasis. )
3. If a do was inserted into a question or negative containing a modal,
that insertion was noted as an error (I). In addition, the do was expected to
carry the correct tense and number (if not, then VII and/or VIII or XVII).
Further, if the sentence was negative, the modal and not the d o was expected
to carry the not (if not, then 111). Last, if the sentence was a question, the
subject and the modal were expected to be inverted (if not, then VI).
4 . If a be was inserted to replace a d o or a modal, that substitution was
noted as an error (11) and the be was expected to carry the correct
number (if not, then VIII). Further, the main verb after be was expected to
be in the simple form (if not, then IV).
5. If a d o was inserted to replace a modal, that substitution was noted
as an error (XIII), and the d o was expected to carry the correct tense and
number (if not, then VII and/or VIII or XVII). Further, the main verb after
d o was expected to be in the simple form (if not, then IV).
6 . If a be was inserted to replace d o in a question, the subject and the
be were expected to be inverted (if not, then VI).
7. If d o or a modal was omitted from a question, thereby making the
sentence indistinguishable from an attempted statement, the lack of a question
marker in the form of a subject-absent d o or subject-absent modal inversion
was counted as an error (VI).
8. If, in a negative question in which two do’s were used, the not
followed the second d o instead of the d o at the beginning of the question, the
lack of subject-negative d o inversion was noted as an error (VI). Furthermore,
the second d o (immediately preceding the verb, and negated) was expected to
carry the correct number (if not, then VIII).
9. If, in a question, no d o was inserted or a modal was omitted, and the
subject was placed after the verb instead of after the absent Aux and before
the verb, the inversion of subject and main verb to replace subject-Aux was
noted as an error (XII).
10. If modals were substituted for each other or for d o , they were
noted as syntactically equivalent in their verbal auxiliary functions (albeit
semantically different) and as simple errors of translation (XIII). Modals or d o
replaced by be were noted as overgeneralization errors (11).
11. If a verb was translated as present tense when it should have been
in the past tense, that substitution was noted as an error (VII or XVII), and
the verb was expected to carry the correct number to agree with the subject
(if not, then VIII).
The above principles were adopted to determine how the taxonomy was
to be used to analyze sentence errors. They reflect many somewhat arbitrary
decisions. In analyzing sentences with numerous syntactic errors, an error
analyst is forced to make decisions which will guarantee that all errors are
accounted for. For example, if we examine the sentence Do John can swim
well? it is not difficult to determine that d o has been unnecessarily inserted.
It is not so easy, however, to decide how to analyze the number on d o (which
is, after ail, unnecessary), or the lack of subject-modal inversion. For these
reasons, the principles outlined above were adopted to ensure that the analysis
would reflect that the d o should have been marked for singularity (because
the subject was singular) and that the correct formation of a question with a
modal requires placing the modal before the subject.
Similarly, given the sentence Does Mr. Miller don’t study English?, it
seemed necessary to consider the second d o as misplaced and misnumbered,
although it probably would have been just as easy to consider the second d o
as redundant and the not misplaced, given that the first d o (actually d o e s ) is
correctly marked for both tense and number. The decision t o adopt the first
analysis was not entirely unmotivated, however, given that sentences of the
types Does John won’t go tomorrow? and Does Mary didn’t leave? also
appeared in the data. These two sentence types seem to require an analysis
which calls for the second auxiliary (won’t, didn’t) to be noted as misplaced.
Given the need for systematicity and consistency in data analysis, i t was
decided that in cases of sentences with two do’s or with a modal and an
unnecessary d o , the second auxiliary (the one immediately preceding the verb)
would be noted as misplaced, and the first auxiliary (at the beginning of the
question) would be noted as redundant.
In questions and negatives in which a be was used in place of d o or a
modal, it again seemed necessary to adopt a mode of analysis which would
reflect the subjects’ use of be in an auxiliary function in negatives, questions,
and unnecessarily emphasized subject-focus Wh-questions. Sentences of the
type Harry isn’t study yesterday (isn’t for didn’t), Is Teresa speak English? (is
for can), and Who is work yesterday? (for Who worked yesterday?) seem to
indicate that in the subjects’ interlanguage, be can fill this auxiliary function.
The principles outlined above and the error-type taxonomy, therefore, reflect
this usage and require that the be and the subject be inverted in questions,
and that the main verb following be be in the simple form.
Five Categories of Errors

It was noted earlier that the twenty error types which comprise this
taxonomy could be grouped into give major categories of errors: ouer-
generalization errors, transfer errors, translation errors, errors o f indeterminate
origin, and errors not considered. This categorization of error types into larger
groups was necessary in order to compare the roles of the overgeneralization
and transfer learning strategies in the acquisition of English by the twenty
subjects.
Overgeneralization Errors
An overgeneralization error is defined here as any error which can be
attributed to the application of a rule of English in an inappropriate situation.
Error types I-VIII in the taxonomy have been classified as overgeneralization
errors.
I. Redundant aux insertions. Whenever a subject inserted an unnecessary
auxiliary verb into his translated sentence, but that auxiliary would have been
appropriate in other syntactic conditions in English, that error was considered
to be an example of both overgeneralization and a redundant aux insertion.
The insertion of a d o into a sentence which already contains a modal, as in
What does Laura can’t understand? appears to be an overgeneralization of the
do-support transformation, which inserts a d o in negatives and questions when
there is no other auxiliary in the sentence. Apparently, the common use of d o
in negatives and questions in English has interfered with the subjects’ ability
to discriminate the conditions under which this transformation is unnecessary.
The same kind of confusion seems to be responsible for sentences in
which two do’s are used, as in Does Gilbert don’t speak French? and Who d o
the students didn’t ask? Because this kind of error appeared only in negative
questions, it seems to be due to some confusion, on the subject’s part,
regarding the ability of the same d o to serve both interrogative and negative
functions.
Overgeneralization from standard question formation appears to be
responsible for errors involving the redundant insertion of d o in a subject-
focus Wh-question, as in Who did write this book? and What does have three
eyes? As mentioned earlier, the Spanish stimulus sentences did not have a
verbal emphasis, nor could it be assumed that the subjects were aware of this
function of d o in English. It seems, rather, that this error stems from
confusion regarding what type of Wh-questions require d o . What did he break?
requires one because what refers to the object of the matrix sentence, so this
sentence is an object-focus Wh-question. What broke?, however, does not
require a d o because what replaces the matrix subject. The greater frequency
in English of object-focus Wh-questions over subject-focus Wh-questions and
the necessity to use d o in most yes-no questions seem to be responsible for
this confusion.
The last kind of redundant aux insertion error which was manifested in
the data was the redundant and incorrect insertion of be in subject-focus
Wh-questions. Why the subjects might have felt that an auxiliary would be
required in subject-focus Wh-questions was discussed above. The use of be to
fill this function, as in What was end last night?, What is ended last night?,
and Who was wrote this book?, however, is peculiar. It seems that the subjects
were confused regarding the use of be in English. These errors, and some
which will be discussed under “Aux substitutions,” appear to be attributable
to the subjects’ assigning an “all-purpose auxiliary” function to be. Because be
is quite common both as a copula and as an auxiliary in English, appearing in
progressives and passives, it is not difficult to see how be could take on this
additional function in the subjects’ minds.
11. Aux substitutions. Auxiliary substitutions are errors attributable to
overgeneralization in which auxiliaries are replaced either by other English
auxiliaries, or by combinations of other English auxiliaries. These substitutions
all result in unacceptable English sentences. What is not work well?, John isn’t
goes t o the movies with y o u , and What isn’t fly? are instances of be filling the
functions of d o , will, and can, respectively. As mentioned under “Redundant
aux insertions,” these errors seem to support the explanation that be is filing
an “all-purpose auxiliary” function, and these errors can be considered to be
overgeneralizations of the use of be in English.
The auxiliary errors found in the following four sentences represent a
somewhat different situation. They don’t go t o read the newspaper, Won’t
they go to read it?, What wouldn’t go to change?, and Who will going to
invite Jacinto? all contain examples of English auxiliaries being combined in
an effort to denote futurity. In these four examples the modal will has been
replaced by d o + go to, by /t:uld
1 go to, and by will going t o , respectively.
The first example appears to reflect an attempt to use be + going t o to
indicate future, but d o has assumed the function of be. It should be noted,
however, that the simple form of the verb go has been correctly retained after
the d o rather than the -ing form, which would have been necessary after be.
The second and third examples also indicate confusion, Chis time between the
use of the modal will (also confused with would) and the phrase be + going
t o . The simple form of the verb is, again, correctly used after the modal. The
fourth example also illustrates confusion between the use of will and be +
going to. These four errors, then, appear to be examples of overgeneralizations
resulting from confusion in the uses of the modal will and the phrase be +
going t o .
111. Incorrect placement of the negative marker. The placement of the
negative marker not in positions which are incorrect, but which would be
correct if the structures were somewhat changed, results in overgeneralization
errors of negative placement. The sentences Ricardo had not the tickets and
What Mr. and Mrs. Williams haven’t? contain examples of not placed after a
main verb, and not after an auxiliary. While the main verb have may be
negated in this way in some varieties of English (notably British English), it is
more likely that this error in the data results from a confusion of the main
verb and the auxiliary functions of have in English, where usually only the
latter can carry negation. Furthermore, the pressure from forms such as
doesn’t, don’t, didn’t, isn’t, aren’t, wasn’t, and weren’t (the last four as both
auxiliaries and as main verbs) may have contributed to this overgeneralization.
The sentence Who don’t will be here at 10:00? results from another
kind of overgeneralization (the use of do in a sentence containing a modal),
which was discussed earlier. In this case, the not has been attached to the
redundantly inserted do, and not to the modal. This error apparently stems
from the pressure to negate the do whenever it appears in negative sentences.
IV. Incorrect form of main verb following an auxiliary. When the main
verb in a negative, a question, or a statement containing an auxiliary retains
the ending which it would have if it were in a statement without an auxiliary,
that error can be considered to be an overgeneralization of statement-form
verbal endings. The sentences Who can Angela sees?, The men will are a t 8
o’clock there, Who doesn’t asked the students?, and What does “circular”
means? all contain examples of this kind of overgeneralization of tense
endings on the main verb. What these errors seem to indicate is that the
subjects realize that an auxiliary (do or a modal) is necessary in the
questionlnegative formation, but they have not learned that that auxiliary
alone, and not the main verb, carries the tense and number. This misunder-
standing is especially apparent in Who can’t Mr. Martinez visited today?,
where the verbal ending is not only syntactically inappropriate, but also
semantically incorrect.
The sentences Can the director talking with me now? and What do they
eating? illustrate a somewhat different type of overgeneralization. In these
sentences the subjects demonstrated confusion regarding the form of the main
verb to be used after dolmodal and be. Was (for Did) Javier studied the lesson
last night? and John isn’t (for won’t) goes to the movies with you are similar
in that the use of be is again confused with the uses of d o or a modal. In
both cases, redundant endings are retained on the main verb following the
auxiliary.
V. Errors in the use of the infinitive marker to. When subjects use the
infinitive marker to immediately before a simple form of the verb, and that
use of the infinitive is neither correct nor attributable to the structure of
Spanish, that error can be considered to be an overgeneralization of the use of
the infinitive where it is appropriate in English. The sentences Who will to
come late? and The boys didn’t to sleep last night illustrate an overgenerali-
zation of the common rule of English to insert a to between two verbs. The
pressure from such forms as plan to go, intend to go, hope to go, want to go,
asked to go, etc. might be responsible for this kind of error. What the subjects
have not completely mastered, as illustrated by their errors, is that d o and
modals do not follow this general rule. ( A discussion of the use of to after the
modal can is found under “Transfer Errors.”)
The sentence What isn’t to need Rafael? (for What doesn’t Rafael need?)
again illustrates an overuse of the infinitive. In this case it is overgeneralized
to follow a be which has been used in place of do. This is another example of
the “all-purpose auxiliary” use of be.
VI. Question-inversion errors. Whenever a question is written using
statement word order, that error can be attributed to overgeneralization. This
kind of error is not attributable to Spanish, where question-inversion would be
necessary. The sentences The students can’t to finish the lesson today? and
What the children don’t study? contain this error.
The sentences Does Mr. Martinez will be here early? and Does Gilbert
don’t speak French? illustrate a similar lack of subject-auxiliary inversion, this
time, however, following the redundant insertion of do. This lack of question-
inversion can also be found in sentences in which be has replaced d o as in The
lawyers weren’t went to the party?
When a modal or do has been omitted from a question, the lack of a
subject-modalldo inversion makes that question indistinguishable from the
corresponding statement, as in Enrique and Rafael eat at 6:00? (doomitted)
and They come to the party? (will omitted). This lack of inversion when the
auxiliary is absent is also found in Wh-questions such as What the teacher
asks? and What they read every morning? Finally, the lack of subject-auxiliary
inversion when the Aux is absent shows up in sentences in which the modal +
V has been replaced by to + V, as in What the guests t o bring for dinner?
(can omitted).
VII. Verb tense errors. When a sentence contains an unambiguous
adverbial time reference, and the tense of the verb does not agree with that
time reference, the verb is said to be in the wrong tense. Does Jauier study
last night?, What finish the last night? (for What ended?), and Jose and Carlos
speak with me yesterday are all examples of the present tense and/or simple
form of the verb being overgeneralized to cover a past tense function. This
preference for the simple form and/or present tense may stem from the
students’ relying on that form as the “basic” verb form, with the past tense
ending being redundant given that the tense is specified by the adverb. There
seems to be a distinct parallel here between the use of an endingless verb
form and the use of be to fill an “all-purpose auxiliary” function. Namely,
just as be is frequently overgeneralized, the simple and/or present tense of the
verb often seems to fill an “all-purpose verb” function.
The sentence The students studying in the library every night illustrates
a somewhat different kind of overgeneralization. In this sentence the pro-
gressive form of the verb has been overgeneralized to fill a present tense
function. This error seems to stem from some confusion regarding the form
and use of the simple present and present progressive tenses.
VIII. Verb number errors. A verb number error is one in which a verb
does not agree with its subject in number. Because the third person singular
present tense form of regular verbs is the only verb form in English which
marks number, the learner is under a certain degree of pressure to over-
generalize the unmarked forms to that verbal form. There is, however, also a
tendency to overgeneralize the -s to forms which should otherwise be
unmarked. These two conflicting modes of overgeneralization are illustrated
by the following examples:
Doesn’t the girls walk every day?
Roberto don’t walk to class every day.
What doesn’t to study the girls? (The girls is the subject, and the
sentence should have been in the past tense.)
Don’t she speak with her teacher? (The sentence should have been in
the past tense.)
The Mr. Martinez speak English very quickly.
The students studies on the library every night.
What break? (will omitted)
Who write this book? (The sentence should have been in the past.)
The sentences Does Gilbert don’t speak French? and Who is imitate the
students every day? (for Who d o the students imitate?) represent similar kinds
of overgeneralization. In the first, the second d o is unmarked for third person,
presumably because the redundant d o at the beginning carries the ending. In
the second, be substituted for d o does not agree with the subject.
Transfer Errors
The kinds of errors so far discussed are caused by confusion with and/or
interference from other similar structures of English. In all cases the some-
times arbitrary peculiarities of English structure seem to be responsible for the
overgeneralizations. Transfer errors, however, are entirely different in origin. A
transfer error is here defined as any error in English which can be attributed
to the structure of Spanish. Error types IX-XII in the taxonomy have been
classified as transfer errors.
IX. Do-insertion errors. The particular way in which English uses the
auxiliary d o in questions and negatives when there is no other auxiliary
present in the sentences seems to be unique. It is not surprising, then, to find

that Spanish speakers have difficulty in remembering how to use it, if they
remember to use it at all. The sentences To whom visited Mr. Martinez? (for
Whom did Mr. Martinez visit?), What understand the children? (for What d o
the children understand?), and Richard has not the tickets last night (for
Richard didn’t have the tickets) all contain errors attributable to the absence
of the do-support transformation. The first sentence is a direct translation of
Spanish, where questions are formed by the inversion of the subject and of
the verb which carries the tense marker. The second sentence might be
analyzed in the same way. Alternatively, it could be interpreted as an instance
of the lack of insertion of d o , with the verb in the simple form. Because the
simple form of regular verbs is indistinguishable from the plural form, either
interpretation seems reasonable. In sentences of this type, in which the verb
could be analyzed as either marked or not marked for the plural, the latter
alternative was arbitrarily adopted. This decision gives the subjects the benefit
of the doubt by maintaining that the d o was omitted, but the verb form
following that absent d o was, correctly, in the simple form. The third
sentence is similar to the first in that d o was not inserted and the tense and
number were retained on the verb, but, unlike the verb in the first sentence,
the verb in the third sentence is marked for the wrong tense, present rather
than past.
X. Errors in the use of the infinitive marker t o . Modals in English are
never followed by a t o . In Spanish, however, poder, which is typically
translated as can/be able to, functions as a full verb and is always followed by
an infinitive. When can is followed by t o + V in a sentence, that error can be
attributed to transfer from Spanish. Can the director to speak with m e now?
and She can’t t o understand are examples of this kind of error.
XI. Incorrect placement of the negative marker. Negation in English is
formed by inserting a not after a model, be, or have occurring before a verb.
If n o modal, be, or haue is present in the sentence, a d o is inserted to carry
the negation. In Spanish, however, negation is expressed by placing a no
before the main verb if there is only one verb form in the sentence, or before
the first verb if there is more than one. The errors in negative placement in
the sentences Who did he brother no invite? and No will t o read Luis and
Nelly the newspaper? can, then, be attributed to transfer from Spanish. In the
first sentence the negative was placed before the main verb, as it would have
been in Spanish. The Spanish sentence would not have contained a d o ; this
error appears to indicate that the do’s function as a negative carrier has not
been learned. (the possible English sentence Who did his brother not invite?
was rejected as a potential model for this sentence since it is doubtful that the
learner was aware of that structure.)
The placement of the no before the modal will in the second sentence is
consistent with Spanish as well. (Like English, Spanish signals futurity in a
number of ways. None of these ways, however, is structurally parallel to the
English use of modal will + V.)
XII. Word-order errors. Question-formation in English involves the in-
version of the subject and the auxiliary, with the subject, accordingly,
preceding the main verb. In Spanish the word order is not so restricted. In
questions containing only a subject and a main verb, the subject and the verb
are inverted. In sentences containing a subject, a full verb filling the English
model function, and a main verb, the word order can be either: modal-like
verb/subject/main verb, like English, or modal-like verb/main verb/subject. In
sentences in which the subject follows the main verb, whether or not an
auxiliary is present, it can be assumed that that word order is attributable to
transfer from Spanish. The following sentences illustrate this error:
Doesn’t speak Alicia with the teacher?
What will to eat the Mr. and Mrs. Velasquez?

What understood the boys?
Whom invite Jacinto? (for Whom will Jacinto invite?)
Who is imitate the students every day? (for Whom d o the students
imitate every d a y ? )
The statement Will not to watch T.V. the boys tonight also reflects
Spanish word order. In Spanish it is possible to place the subject after the
main verb in statements, although it is less common to find it placed after the
object as well.
Translation Errors
For the purposes of this taxonomy, a translation error is defined as any
error which changes the desired response in an insignificant way. Most involve
simple substitutions of one syntactically correct structure for another equally
syntactically correct, albeit semantically incorrect, alternative. Most can be
attributed to simple forgetting or lapses of attention and are, in this respect,
comparable to Chomsky’s “performance errors” (1965).Error types XIII-XVII
in the taxonomy have been classified as translation errors.
XIII. Aux substitutions. When a modal is substituted for d o , d o for a
modal, or one modal for another, that substitution changes the desired
response only trivially. While this kind of substitution frequently alters the
meaning of the sentence (and occasionally produces a meaningless sentence),
the syntactic properties of the sentence remain essentially unchanged. In this
analysis all modals and do are considered to be syntactically equivalent.
Because this study is concerned more with the acquisition of syntactic form
than with the acquisition of syntacticsemantic relationships in English, there
has been no attempt to analyze the semantic differences caused by these
substitutions.
The following sentences illustrate these Aux substitution errors:
What Laura doesn’t to understand? (doesn’t for can’t)
Who will not to study the lesson? (will for d i d )
What will to happen now? (will for can)
Might he speak with me now? (might for can)
He must not buy the guitar. (must for can)
Whom the teachers cannot to invite? (can for will)
XIV. Negative substitutions and omissions. Lapses of attention and/or
memory seem to be responsible for the errors contained in sentences which
are affirmative when they should have been negative, and negative when they
should have been affirmative. Who does will see Raul tomorrow? (for Who
won’t Raul see tomorrow?) and Who can’t to speak French? (for Who can
speak French?) illustrate this type of translation error.
XV. Main verb omissions. When a sentence contains an auxiliary verb,
but the main verb has been left out of the translation, that error should
probably best be considered a simple error of performance resulting from
carelessness or lack of attention. The following sentences contain examples of
this performance error:
Robert doesn’t to class every day. (for Robert doesn’t walk to class
every d a y )
What does Erlando can in English? (for What can Erlando read in
English?)
John won’t to the movies with you. (for John won’t g o to the movies
with y o u )
XVI. Modalltense substitutions. When a modal is omitted from the
translation of a Spanish stimulus sentence, that omission and the subsequent
substitution of a present, past, or simple form of the verb for the modal + V
results in a modal/tense substitution error. Like other translation errors, this
substitution error appears to be attributable to translationfperformance
factors. The sentences The men are there at 8 o’clock (for will be), Who
speaks French? (for can speak), and Who come later? (for will come) all
contain errors of this type. (The verb number error on come in ‘the last
sentence is accounted for by VIII.)
XVII. Tense substitutions. When the Spanish stimulus sentence contains
either no adverbial time reference, or that reference is ambiguous, and the
translation of that sentence contains a do or a main verb in a tense different
from that which was in the stimulus sentence, this tense substitution can be
attributed to performance and translation variables. Since the sentence con-
tains no adverbial time reference to help the subiect remember the tense of
the sentence when he is translating it, it is easy to see how lapses of attention
and memory can cause errors of the types illustrated in the following
examples :
Who doesn’t your brother to invite? (for didn’t)
What didn’t work here? (for doesn’t)
Who write this book? (for wrote)
(The number error on write is accounted for by VIII.)
Errors of Indeterminate Orgin

When an error in the translation of a Spanish stimulus sentence cannot
be accounted for by overgeneralization, transfer, or translation strategies, that
error is considered to be of indeterminate origin. Error types XVIII-XIX in
the taxonomy have been classified as errors of indeterminate origin.
XVIII. Incorrect use of the infinitive marker to. When to + V is used in
place of do + V, a conjugated verb, or modal + V, those errors do not appear
to be attributable to any of the three strategies which have been discussed
thus far. In a sense, we might be able to understand how to + V comes to be
used where other verbal constructions are required if we realize that students
learning English as a second or foreign language tend to learn the to + V form
when learning how to translate the basic infinitives in their native languages.
This tendency to consider to as part of the verb might be responsible for its
overuse in attempted English sentences such as the following:
What the boy to understand? (for What does the boy understand?)
The student to study in the library each night.
Who to see Angela? (for Who can Angela see?)
XIX. Incorrect placement of the negative marker. If not is inserted after
a main verb instead of after the modal that is in the sentence, that error
cannot be accounted for by the three strategies which have been discussed
previously. Neither overgeneralization, transfer, nor translation strategies can
explain why the error in Who will be not at the 10 o’clock? (for Who won’t
be here at 10 o’clock?)was made.
Errors N o t Considered
The taxonomy of error types which has just been discussed was
designed to provide a framework for the classification of errors made in the
Aux’s and VP’s of the translated Spanish stimulus sentences. Errors occurring
in other parts of a sentence were not investigated. Mistranslations of vocabu-
lary, incorrect past tense forms, errors in adverb, determiner, and preposition
usage, and other errors of similar types were noted and classified, but these
errors were not considered in the results.
XX. Errors not considered. Some examples of the kinds of errors which
were not considered are presented below:
John can swim right. (for well)
Ricardo don’t has the enters. (for tickets)
The boys aren’t see the T.V. tonight. (for watch)
What do the guests brings t o the soup? (for dinner)
The children can t o come after. (for now)
Jose and Carlos speaked with me yesterday.
He can well to swim.
What will be not at the 10 o’clock?
The boys don’t will watch the T.V. this night.
Who know t o this man?
To whom did Maria and Elsa persuade?
They are eating every day at 6 o’clock. (for Do they e a t . . . 2 ) (The
present progressive was not investigated, so this error in progressive
question formation was not considered in the analysis of the
corpus.)
Did studied the lesson last night? (omission of subject)
The student study in the library every night. (for students)

Overgeneralization and Lang Transfer

Uploaded by

Copyright:

Available Formats

Overgeneralization and Lang Transfer

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Overgeneralization and Lang Transfer

Uploaded by

Copyright:

Available Formats

THE USE OF OVERGENERALIZATION AND

TRANSFER LEARNING STRATEGIES BY

An orally administered test requiring the written English

In a recent article, Taylor (1974b) discusses the possibility

‘This is a revised and expanded version of a paper presented at the 9th

4. Errors which appear to indicate a reliance on native

Error analysis is not a new way to investigate the process of

analyzing errors, however, we are frequently unable t o compare the

Materials and Testing Procedure

that the subjects had to reflect on the problem and required an

3 pl sb: 3 pl sb: 3 pl sb:

Negatives (17) 3 sg sb (19) 3 sg sb (21) 3 sg sbt& Negative 3 sg sb: 3 sg sbi 3 sg sb:

3 p l sb: 3 pl sb: 3 p l sb:

>third person sbsubject ob=object *not acceptable in Standard English

THE ERROR ANALYSIS

The translations of the eighty Spanish sentences by the

A Taxonomy of Error Types

Five categories of errors were established: overgeneralization

Results and Discussion

verbal endings (notably their use of -s on plural verbs), and their

Wb j e c t Wh-ques t i o n s 60 37 132 32 192 69

*Neg o b j Wh-questions lo5 33 101 26 206 59

did the elementary subjects. A comparison of the means of these

Statements .62 1.00 9 38 0.00

H . 0 5 one tailed). These results also confirm the hypothesis

Figure 1 . The means of the proportions of overgeneralization

intermediate transfer errors, we should be careful not to compare

Several important conclusions can be drawn from this in-

From the point of view of language learning, these results

2. Los estudiantes n o tienen dinero.

3. Carmela no preguntb al professor.

*first sitting-50 sentences and 5 pre-test sentences

5. iQuiPn n o estudi6 10s verbos?

6. iVive Alfredo en Venezuela?

7. El senor Garci’a trabaja muy duro.

8. El senor Martihez vivi6 en Caracas.

9. i N o escribio Rafael la carta?

10. ‘Compraron tus hijos la casa?

13. Los estudiantes estudian e n la biblioteca cada noche.

67. El senor Garcia escribi6 la carta anoche.

46. JosP y Carlos hablaron conmigo aver.

5. Pablo saldrd a las o ch o manana.

29. Juan puede nadar bien.

22. Los hombres estardn ah i a las ocho.

28. ‘Estudia Margarita e n el ELI?

62. ‘Cornen Enrique y Rafael a las seis todos 10s dias?

75. iEstudi6 Javier la lecci6n anoche?

24. ‘Prepararon las mujeres el desayuno?

6. ‘Puede el director hablar conmigo ahora?

47. ‘Sabra Alejandro la respuesta manana?

38. ‘Pueden 10s estudiantes hablar inglds?

80. ‘Vendran Margarita y Carlos a la fiesta?

21. Roberto no camina a clase todos 10s dias

76. (omitted for technical reasons)

70. Ricardo no tenia las entradas.

42. Los chicos no durmieron bien anoche.

56. Enrique no puede comprar la guitarra.

15. Juan no ira a1 cine contigo.

36. Roberta y Elsa n o pueden cantar bien.