ui....

       

    

 

    

 

 

 

 
   

 

 

  

  

 

131 Mme

MCI CEHK ANS TATE NUIVERSITY LIBRARIE

IIIIIII IIIIIII IIIII IIII II II

3 1293 00579 9980

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

I

I

.
w

LIBRAR 9"“
Mi‘h‘fi'" State,
L University

 

 

This is to certify that the

thesis entitled
Syntactic Aspects of Bilingual Language Production
I presented by
Helga Loebell
has been accepted towards fulfillment

of the requirements for

Master of Arts degfimin Psychology

m m

Major professor

Date 3/2111989

0-7639 MS U is an Affirmative Action/Equal Opportunity Institution

 

PLACE IN RETURN BOX to remove thls checkout from your record.
TO AVOID FINES return on or before date due.

 

 

 

DATE DUE DATE DUE DATE DUE

 

‘ ‘4~_ ' 'AUC ‘ 5 ‘ p
‘ ‘ App 1 9 “fjf‘f‘

‘~

 

 

 

 

 

 

 

. I

 

APR 1 2.23m
I] V U y r!

 

FM
0917 n

 

 

 

 

 

 

 

 

 

MSU Is An Affirmative ActlorVEqual Opportunity Institution

 

SYNTACTIC ASPECTS
OF BILINGUAL LANGUAGE PRODUCTION
. By
Helga Loebell

A THESIS

Submitted to
Michigan State University
in partial fulfillment of the requirements
for the degree of

MASTER OF ARTS

Department of Psychology
1 989

ABSTRACT
SYNTACTIC ASPECTS OF BILINGUAL LANGUAGE PRODUCTION
By
Helga Loebell

Bilinguals have knowledge of two linguistic systems. The phenomena
associated with code switching suggest that both systems can be active and
available during language production. Two experiments were conducted to
examine the implications of this for syntactic processes in bilingual speech.
Experiment 1, using fluent German/English bilinguals, investigated whether
producing a construction in one language influences the probability of using
the same constmction in the other language to describe a subsequent,
unrelated picture. It was found that dative sentences in one language primed
the later use of dative forms in the other language. Experiment 2 attempted to
extend this experiment to a monolingual German setting, with native German
speakers producing the priming sentences and picture descriptions in
German. Although the priming effects failed to achieve significance, they
followed the general pattern of the cross-language results for datives.
Implications for models of bilingual sentence production are discussed.

ACKNOWLEDGEMENTS

I would like to thank Dr. Kathryn Bock for her guidance and
encouragement. Her interest and constant willingness to discuss any issue
or problem arising out of this research has helped greatly. My other comittee
members, Dr. Thomas H. Carr, Dr. David E. Irwin, and Dr. Rose T. Zacks, are
thanked for their willingness to always help and answer questions. Thank
you also to Ken Smithson, whose assistance with computer problems has
helped the progress of this work greatly.

I would also like to express special thanks to all the bilingual and
monolingual German speakers that have served as subjects in this study.
Without their interest and their eager participation this project would have
been neither possible nor as enjoyable.

Lastly, I would like to thank my children Nicole, Stephanie, and David for
reminding me daily how fascinating bilingualism is.

m

TABLE OF CONTENTS

Page
LIST OF TABLES ........................................................................................ vi
CHAPTER
I. INTRODUCTION AND LITERATURE REVIEW ............................. 1
The Characterization of Bilingualism ............................................. 2
Language Transfer ............................................................................ 3
The History of the Construct of Transfer ........................ 4
The phenomena and interpretation of
Transfer ............................................................................... 5
Code-Switching .......................................................................... 7
Language Interaction in Models of Language
Use ................................................................................................ 12
An Experimental Paradigm to Study Syntactic
Transfer ........................................................................................ 18
ll. EXPERIMENT 1 .................................................................................. 22
Method ......................................................................................... 22
Subjects .................................................................................. 22
Materials ................................................................................. 22
Procedure ............................................................................... 24
Scoring ................................................................................... 26
Design and Data Analysis ................................................... 27
Results ......................................................................................... 27
Discussion ................................................................................... 31
Ill. EXPERIMENT2 .................................................................................... 33
Method .......................................................................................... 34

iv

Subjects .................................................................................. 34

Materials ................................................................................. 35

Procedure ............................................................................... 35

Scoring ................................................................................... 35

Design and Data Analysis ................................................... 36

Results ......................................................................................... 36
Discussion ................................................................................... 39

IV. GENERAL DISUCSSION ................................................................... 41
Conclusion .................................................................................. 46
Footnotes ..................................................................................... 48
APPENDICES ............................................................................................. 49

A. List of Bilingual Subjects, with Occupation, Age,
and Length of Residency in an English Speaking

Country ............................................................................................. 49

B. Dative and Transitive Priming Sentences and
Pictures, Experiment 1 .............................................................. 51
C. Questionnaire ................................................................................ 57
LIST OF REFERENCES ............................................................................ 59

LIST OF TABLES

Table Page
1 Number of Utterances in Dative Priming
Conditions, Experiment 1 ......................................................... 43

2 Number of Utterances in Transitive Priming
Conditions, Experiment 1 ......................................................... 45

3 Number ot Utterances in Dative Priming
Condtions, Experiment 2 ......................................................... 54

4 Number of Utterances in Transitive Priming
Conditions, Experiment 2 ......................................................... 55

vi

Syntactic Aspects of Bilingual Language Production

CHAPTER I
INTRODUCTION AND LITERATURE REVIEW

Bilingualism has intrigued researchers and theoreticians from a variety of
fields for decades. This interest is most prominently reflected in controversies
over its merits. Some studies seem to indicate that speaking more than one
language can lead to increased linguistic flexibility and intellectual diversity
(Bain & Yu, 1980 ;Cummins, 1977; Landry, 1974; Peal & Lambert, 1962).
Others have labelled bilinguals morally and intellectually deficient, suffering
from weakened self-discipline and general decay of character ( Weisgerber,
1966). But regardless of whether bilingualism is a handicap or an asset,
questions arise about how the knowledge of multiple languages is
represented and used. How are the languages of a bilingual speaker
organized? To what degree and in which domains of language functioning
are the languages interrelated? How do they interact in language use? This
paper will focus on problems associated with structural aspects of language
patterning in bilingual speech production.

In recent years interest in the processes operating when bilingual
speakers switch languages within sentences (code-switching) has sparked
research on the syntactic aspects of bilingual language processes. This
research, to be examined in more detail below, converges on the hypothesis
that both languages of the bilingual are simultaneously active during the
speech formation process. This, in turn, leads to a reconceptualization of the
phenomenon of transfer, the imposition of the the pattern of one language on

another, and ties it to the phenomenon of code-switching. The purpose of this
paper is to assess this simultaneous activation hypothesis in a controlled
setting and to outline its implications for a model of bilingual language
production. The first section of the introduction lays out some terminological
conventions. The second reviews the role of language transfer in second
language acquisition and bilingual production. In the last part of the
introduction, the relevance of code-switching to explanations of syntactic
processing in bilingual speech will be assessed. The results of an
experiment designed to examine syntactic processes in bilingual speech will
be reported. A second experiment replicated and extended this experiment
in a monolingual setting. Finally, outcomes are evaluated and discussed with
regard to their implications for bilingual language production.

The Characterization of Bilingualism

lnformally, individuals who have substantial mastery of two languages
can be regarded as bilingual. There is no agreement among researchers
about what constitutes substantial mastery and how or even whether it relates
to labelling a person bilingual. The term 'bilingual' has been used very
narrowly by Bloomfield (1933), among others, to refer only to native-like
mastery of another language. A different view is that people who have even
minimal competency in at least one language skill (speaking, writing,
listening, or reading) should be regarded as bilingual (Macnamara, 1967).
With the former definition there would be few (if any) 'true' bilinguals; the latter
definition would include almost everyone who has ever come in contact with
another language.

As a practical matter, the all-or-none view of bilingualism is untenable
because of the difficulty of adequately characterizing asymptotic attainment of
a language. Recent studies have revealed that individuals judged to have
native-like competence in two languages nevertheless differ from native

speakers on more subtle measures of linguistic intuitions (Coppieters, 1987)
and processing strategies (Bates, McNew, MacWhinney, Devescovi, & Smith,
1982)

For the purpose of this paper, I will adopt the view that "...bilingualism is
not an all-or-none property, but is an individual characteristic that may exist to
degrees varying from minimal competency to complete mastery of more than
one language” (Hornby, 1977). However, the research to be considered and
proposed emphasizes individuals at the upper end of the continuum. These
are speakers who, while not necessarily 'native like' in all their linguistic
competences in both languages, are able to understand and converse
fluently in both. The first language learned by these speakers in childhood
will be termed the W. Any language that is acquired after the
first will be termed a We. Given these definitions, children who
acquire more than one language simultaneously in childhood will have more
than one native language. In the more usual case of asynchronous
acquisition of two languages, the term tamauanguagg will be used to
denote the language that a second language learner is trying to acquire.

\

Language Transfer

In the literature of bilingualism, transfer is the paradigm case of
crosslanguage influence. Language transfer occurs when a language user
appears to impose the pattern of one language on another (at any level of
structure, including both the forms and functions of elements; Gass, 1983).
Although most authors agree with this basic definition, some include
additional phenomena such as failure to use target language forms that have
no source language analogs (Kellerman & Sharwood Smith, 1986),
overproduction of target forms that have source language analogs (Selinker,
1983), and transfer of typological organization (roughly, basic word order,
Rutherford, 1983). In its broadest sense, any effect that one language has one

producing utterances in an other language might be termed 'transfer'.

In this broad sense, borrowing and interference might be regarded as
specific instances of transfer. Borrowing refers to the incorporation of specific
lexical elements from one language into another. This is often done to make
up for lexical gaps, as in English borrowings of terms like defame, and mun.
Though this may originate in transfer, the result is that the borrowed items
become part of the language into which they are incorporated. These lexical
borrowings often lose their language specificity during the incorporation
process, in that the new words are assimilated (by changing and adjusting
their pronunciation and morphology to fit the'host language). Interference, on
the other hand, is a form of transfer in which a second language learner uses
linguistic structures of the source language that are incompatible with the
patterns of the language being spoken. One common form of language
interference is seen in second language learners who fail to attain native-like
pronunciation. Instead, they retain an accent that reflects the phonology of
their native language.

MW

Within linguistics, particularly in applications of linguistic theory to
second language learning, research on transfer has been closely associated
with behaviorist accounts of language learning (Skinner, 1957) and the use
of contrastive analysis (Lado, 1957). By comparing the grammatical
structures of different languages, contrastive analysis tried to identify likely
difficulties of second language learners (James, 1980). However, language
learning turned out to be poorly explained in contrastive terms. This produced
a lasting disenchantment with the notion of transfer, a disenchantment that
was enhanced by dissatisfaction with anything remotely behavioristic (Dulay
& Burt, 1974; Felix, 1982).

Although there is nothing obviously objectionable in the contrastive
analysis program, it depends crucially on identifying psychologically real

structures and their roles in language use. Klein (1986) , for example, has
suggested that predictions of possible transfer should not be based on
comparisons of structural properties but rather on the way in which the
learner processes these structural properties. Contrastive analysis did not do
this.

The dismissal of transfer in second language acquisition, linked as it was
to contrastive analysis and behaviorism, reflected more the reaction against
behaviorist theories than a careful evaluation of the role of transfer in
language processing. Gass and Selinker (1983) pointed out that the
negative associations connected with transfer should not lead to a denial of
the fact that existing linguistic knowledge will influence the way second
language learners acquire another language. This is implicitly
acknowledged in the terminological revisions that have appeared, with some
researchers suggesting the substitution of terms such as 'mother tongue
influence' (Corder, 1967), or 'crosslinguistic influence' (Kellerman &
Shanlvood Smith, 1986) for the tarnished construct of transfer.

II I I . | | I. I I I

It has been found that lexical, phonological, syntactic, pragmatic,
rhetorical, conversational, and discourse factors from the native language can
all be identified as influencing the learner's struggle in acquiring a new
language (Gass & Selinker, 1983). In what follows, some instances of
syntactic transfer will be highlighted.

Syntactic transfer is evident when the structural aspects of the native
language influence sentence construction in the target language. Adjémian
(1983) investigated transfer processes involving the English verb 'fight'. The
verb 'fight' corresponds to the French verb 'se battre', a reflexive verb.
Transferring these reflexive properties of the verb to English, French second
language learners produced sentences like:

'They want to fight themselves against this [tuition increase]'.

Syntactic transfer is seen in the configurational features of sentences as
well. In producing the incorrect German sentence:

*Bringen Sie mir ein Glas von Bier. (Kufner, 1962)

(Bring me a glass of beer)

a native English speaker was influenced by the English construction 'glass of
beer'. The correct German constmction does not take a preposition: 'Bringen
Sie mir ein Glas Bier.‘ Other instances of syntactic transfer include the
incorrect ordering of adjectives and adverbs, depending on the mles of the
native language. For example, a French-English bilingual produced the
sentence

* I started only to learn Russian grammar in high school (Sheen, 1980).

(Je commence seulement a apprendre la Russe a I'école élémentaire)
Likewise, differences in adjunction and complementation cause native
German speakers touse incorrect English verbs forms, as in

*It has stopped to rain (Mack, 1986)
which corresponds to the German equivalent of : 'Es hat autgehort zu
regnen'.

Traditionally, the use of the term transfer has been restricted to this
one-way, native to target language transfer of linguistic elements and forms.
However, transfer may not occur only from the speaker's native to his or her
second language, but also vice versa, from the second to the native
language.

A study by Van Vlerken (1980, cited by Sharwood Smith, 1983) supports
this more symmetrical notion of transfer. In Van Vlerken's experiment,
English speaking school-aged children who had lived in the Netherlands for
a minimum of two years were given an English sentence building task, and
an English acceptability judgment task. An analysis of the word order errors
revealed that 83% of the errors produced were attributable to transfer from
Dutch to English. Sharwood Smith concluded that native language
competences are not fixed constructs, but should be regarded as "permeable"

(p.225). That is, transfer processes cannot be regarded as fixed, one-way
influences.

In general, speakers might draw on their knowledge of different
languages to produce sentences in another language. Transfer, as it occurs
in second language learning, seems to involve primarily influence of the
native language on the acquisition of the second language. The idea of
symmetrical transfer implies that this influence might be more bidirectional,
and suggests the possibility of considerable language interaction during the
production process. This contradicts the idea that the languages of the
bilingual are independent entities. The suggestion has been made
(Macnamara & Kushnir, 1971 ; Kolers, 1963) that the ability of bilinguals to
keep their langauges apart while talking to monolinguals indicates that the
languages are processed independently, with translation serving as the
connecting link. Others (Obler & Albert, 1978; McCormack, 1977), have
proposed that bilinguals cannot block out their knowledge of the other
language.

To investigate the rules and patterns of language interaction,
researchers have studied the tendency of bilingual speakers to mix
languages within utterances. In the next section I will address the question of
how languages might interact in language use In terms of this phenomenon
of code-switching.

Code-switching

Code-switching is characterized by the mixing of two languages. This
mixing might involve, 9.9., the incorporation of single lexical items in an
otherwise monolingual sentence:

"'Si va air shopping, vaya con Mickey' (Pfaff, 1979)."

(If you go shopping, go with Mickey),
or the switching of languages within the sentence:

"'Yo sé, porque I went to the hospital to find out... ' (Reyes, 1974)"
(I know, because I went ....... )
or the incorporation of idiomatic expressions:

'"Yo anduve in a state of shock por dos dias' (Pfaff, 1979)."

(I went around ..... for two days)

Such mixing of languages provides an opportunity to observe language
interaction during normal, time-constrained speech production. The
constraints on code-switching may reveal underlying organizational factors
that limit or promote the interplay of different syntactic systems.

The fact that bilinguals do switch, seemingly without effort, from one
language to another while producing fluent speech led to the proposal of a
mechanism in the brain which turns linguistic channels on or off depending
on the language of discourse (Penfield & Roberts, 1977). Code-switching
has been studied extensively for evidence of such a mechanism.

Some have tried to specify the operation time of the proposed switch.
Macnamara, Krauthammer, and Bolgar (1968) showed that expressing
'linguistically neutral' stimuli (such as numbers) in linguistic form took longer
for mixed language responses than for unilingual responses. The difference
was taken to represent the operating time of a consciously controlled output
switch. In a subsequent study, Macnamara and Kushnir (1971) compared the
reading times for monolingual passages with translation-equivalent mixed
language passages. The passages contained sentences with one, two, or
three language switches, with the sentences being either " basically English
with French words in place of English ones” (p.482), or vice versa. Half the
sentences represented true statements , such as 'Douze chases make une
dozen', and the other half contained false sentences, such as : 'Most fenétres
have cheveux.'. The reading times for switched passages were longer than
for either monolingual passage. The extra reading time forthe bilingual
passage, divided by the number of code-switches it contained, was taken as
an index of the operating time of an input switch. Macnamara and Kushnir

(1971) concluded that bilinguals have psychologically distinct language
systems, and proposed a two-switch model of bilingual functioning. It
included an automatic input switch (one does not choose to listen in a
particular language), and an output switch that is under voluntary control.
Both switches work sequentially and independently and require an
observable amount of time. '

The proposal has been attacked on several grounds. First, the proposal
of functionally distinct language systems that can be voluntarily switched on
or off may be untenable. Obler and Albert (1978), in a bilingual version of the
Stroop test, asked Hebrew/English bilinguals to label in English the colors of
color words written in Hebrew. The color words conflicted with the ink colors
they were written in. They found that even with dissimilar orthographic
systems such as Hebrew and English, subjects were unable to completely
suppress the interfering stimulus words, and consequemly produced errors.
To the degree that Stroop interference is an output effect due to response
competition (Morton & Chambers,1973; Palef & Olson, 1975; Posner,1978;
but see Dunbar & MacLeod, 1984 ), this finding casts doubt on any simple,
voluntary on/off switching mechanism.

Second, with respect to syntactic mechanisms, the study of Macnamara
and Kushnir has also been criticized for its artificially constructed
code-switches (e.g. Chan, Chan & Hoosain, 1982; Dalrymple-Alford, 1985).
Some of Macnamara and Kushnir's (1971) language-mixed sentences, such
as W, would never be uttered by a
code-switching English/French bilingual speaker. The use of such sentences
reflects an underlying assumption that where and when code-switching
occurs is irrelevant to its linguistic processing. Experimenters who have used
natural code-switched speech, rather than constructed sentences, did not find
any evidence for additional processing time for linguistically mixed texts as
compared to monolingual passages (Trmm, 1983).

Code-switching is probably constrained by the syntactic structure of the

10

languages involved and is a mle-govemed rather than a random word mixing
process (Pfaff,1979). Analysis of spontaneous speech has shown that the
placement of code-switches occurs in regular patterns (Pfaff, 1979; 'l'rmm,
1983). In addition, subjects asked to evaluate the acceptability of code-mixed
sentences consistently agree to accept some sentences while rejecting
others (Poplack, 1980). Such work has suggested several syntactic
constraints on code-switching. It is generally unacceptable to switch
languages within words (Poplack, 1980), as in switching between a root and
an inflectional morpheme (Berk-Seligson, 1986) or within the elements of the
verb, for example between an auxiliary and a main verb (Timm, 1983). In
contrast, code-switching is permitted at major constituent boundaries,
between sentences and clauses (Pfaff, 1979). Although the switching of
single words does occur in code-switched speech, the higher the
constituency of an element, the more likely it is to be code-switched (Sridhar
& Sridhar, 1980). 80 main clauses, relative clauses, noun phrases, and verb
phrases are frequently switched, but the code-switching of function words,
such as articles, quantifiers, and prepositions rarely occurs.

The systematicity of code-switching has led to the proposal that, through
exposure to code-switched speech, code-switching bilinguals form a third
grammar which is based on an integration of their separate languages
(Poplack, 1980). If this is true, then bilinguals who have extensive exposure
to code-switched speech should be better at judging the acceptability of
mixed sentences than speakers who have little or no exposure to
code-switching.

Lederberg and Morales (1985) tested this hypothesis by comparing the
grammatical intuitions of Mexican-American subjects who were born in the
United States, lived in a Mexican-American environment, and frequently
code-switched, with those of Latin Americans who moved as adults to the
United States, reportedly did not code-switch, and had no contact with
code-switching communities. The groups did differ in their judgment of the

11

acceptability of certain constructions. The Latin Americans were less willing
to accept code-switches within a noun phrase, or switches where the
language of the stern of the main verb and the language of the inflection
differed than the Mexican-Americans. However, the overall response
patterns were similar. Both groups agreed, for example, that a code-switched
verb is more acceptable than a code-switched auxiliary, and switches within
noun phrases containing Spanish adjectives are less acceptable than other
types of mixed noun phrases. Code-switching experience did not
significantly affect the pattern of grammaticality judgments. Lederberg and
Morales (1985) concluded that the rules of code-switching are primarily
based on the knowledge of the grammars of the two code-switching
languages, without the need to postulate a third grammar.

A quantitative analysis of mixing in conversations of Mexican-Americans
supports this argument (Pfaff, 1979). It revealed that code-switching occurred
mainly at points which did not entail a syntactic conflict between the two
languages. For example,

I put the knives en la mesa ('I put the knives on the table')
is acceptable, because the phrase structure rules for generating this sentence
are identical in English and Spanish, whereas the sentence

* He ran to the house Chiquita. ('He ran to the little house')
is unacceptable, because Spanish requires the adjective to follow the noun in
this case, whereas English requires the adjective to precede the noun.
Recorded interviews with a Puerto Rico-American woman similariy showed
that out of 400 recorded instances of code-switching, only one did not satisfy
either English or Spanish surface structure constraints (Poplack, 1981 ). The
remaining 399 instances obeyed the structural rules for both languages.

This conclusion, that the rules of code-switching are based on the
grammars of the languages involved, implies that a bilingual code-switching
speaker has simultaneous access to both grammatical systems. Traditionally,
an "ideal" bilingual was one regarded as having equal proficiency in two

12

distinct languages, languages kept segregated from each other.
Code-mixing was thought to be an indicator of deficient linguistic mastery
(Weinreich, 1963). However, in orderto fluently code-mix, bilinguals must
have sufficient proficiency to observe the linguistic constraints of both
languages. In the production of mixed sentences ".. not only the lexicons but
the entire rule system of both languages must be simultaneously active..."
(Sridhar &Sridhar, 1980, p.411).

Code-switching, then, may be the result not of deficient language
mastery, but a sign of high proficiency. As Trmm (1983) points out, less
proficient bilinguals do not code-switch the way fluent bilinguals do. Rather,
they tend to code-switch only at the level of single words, to fill lexical gaps.

A similar proposal has been made regarding native language
interference in second language learners (Wade, 1977). During an eight-
month observation of foreign language learning, Felix (1982) found that
interference increased with proficiency level in the second language. He
pointed out that this was true only for syntactic interference; phonological
interference was negatively correlated with proficiency level. Felix concluded
that the second language learner has to reach a certain competence in the
new language before native grammatical knowledge influences the second
language. This makes it unlikely that syntactic transfer involves only a
translation process. If it did, beginning second language learners should
show extensive syntactic transfer. Instead, it suggests that syntactic transfer,
like code-switching ability, may only emerge when a second language
learner attains some competence in the target language.

Language Interaction in Models of Language Use
The evidence is strong that the syntactic systems of both languages are

simultaneously active during bilingual code-switching. It is possible that such
simultaneous activity is also responsible for transfer between the structural

13

systems of the two languages. The simultaneous activation of both
languages might result in code-switching in fluent bilinguals, whereas this
activation might result in structural Interference in second language learners.
Code-switching and transfer, thoughtreated by researchers as separate,
independent phenomena might, thus, be both a manifestation of
simultaneously active languages. However, these structural factors have
received little attention in recent examinations of the transfer process. Rather,
researchers have focused on the roles of semantic and pragmatic factors.

The competition model represents one attempt to explain crosslinguistic
interference in terms of general, nonstructural forces ( Bates & MacWhinney,
1982; MacWhinney, 1987). According to the competition model, the
interpretation of a sentence relies on language specific cues based on the
relationships among four language elements: lexical items, morphological
markers, word order‘, and prosody. The utilization and relative dominance of
these cues in assigning thematic roles to a given element varies across
languages. In an attempt to explain crosslinguistic differences in sentence
interpretation, Bates et al. (1982) showed that monolingual English speakers
relied more heavily on word order cues in interpreting anomalous sentences
than did monolingual Italian speakers, who relied more on semantic and
pragmatic cues. That is, cue strength differences across languages can lead
to differences in strategies of sentence interpretation. One issue in transfer is
whether such strategies are applied to second languages, too.

Harrington (1987) found some support for the applicability of the
competition model to second language acquisition, showing that native
Japanese speakers seem to make greater use of animacy cues, whereas
~ native English speakers seem to rely more on word order cues, consistent
with the cue strength differences of the respective languages. Native
Japanese second language learners, responding to sentences in English,
showed both the animacy biases typical for Japanese sentence

14

interpretations, and the word order effects typical of English sentence
processing.

The competition model points to an important factor in multi-language
processing, that is, that the processing of a second language may reflect
encoding and decoding patterns appropriate for one's native language. It is,
however, based on the premise that there is no syntax per se, so that all
instances of transfer must be reducible to cue use keyed to semantic,
positional, and lexical form class information, not more abstract constituency
or syntactic role information.

In a different attempt to explain a transfer phenomenon, Jordens (1986)
drew on a discourse-functional distinction between 'given' information and
'new' information. He hypothesized that the given and new information are
independently produced, one before the other, and that some second
language acquisition errors are attributable to this independence. In
spontaneous speech samples from monolingual Dutch speakers, Jordens
found that native Dutch speakers acquiring German made case-marking
errors in their German utterances that could be attributed to their habitual
assignment of the preverbal noun phrase to subject function, as it occurs in
Dutch. Jordens interpreted this as evidence for the independent production
of 'given' and 'new' information. However, the same case marking errors
were also produced by native German speakers, suggesting that the
observed errors represent general breakdowns in sentence planning rather
than syntactic transfer.

A model that emphasizes the semantic underpinnings of transfer has
been proposed by MacKay (MacKay & Bowman, 1969; MacKay,1982). The
model was developed to account for increasing fluency in high proficiency
skills such as speech production, and how the skills acquired in practicing
one performance transfer to other performances. According to MacKay, the
process of language production proceeds in a hierarchical fashion. The
basic components for organizing complex actions are nodes, which are

15

organized into different subsystems. Action planning proceeds in a top-down
fashion, from the formation of a thought (arrayed over propositional nodes) to
the activation of relevant muscle movement nodes for the articulation process.

Phonological and muscle movement nodes are especially well practiced
for common words. However, higher level nodes, such as propositional
nodes, are less well practiced because, unlike phonemes or muscle
movement sequences, particular concepts do not occur as often in the same
form as do phonemes. As propositional nodes are less practiced than other
nodes, they should benefit most from transfer because other nodes have
attained asymptotic levels. In bilingual speech production, transfer, and with
it a facilitation in production speed, should occur at the level of propositional
nodes rather than at the level of muscle movement nodes, as the set of
muscle movements involved in the language production of a given thought
varies widely from language to language.

To test this, MacKay and Bowman (1969) had German/English bilingual
subjects produce a practice sentence in one language twelve times at a
maximal rate. Then subjects were given either word-for-word translation
equivalent sentences in the other language, or semantically equivalent
sentences with different word order. The researchers found higher speech
rates when the translations were phrased as word-for-word equivalents as
compared to semantically equivalent sentences. MacKay (1982) attributed
this effect to a contextually dependent integration of the meanings of
individual words, and concluded that the observed priming effect could not be
attributed to phonological or muscle movement priming, because those
varied as a result of the use of different languages. Instead, according to
MacKay, conceptual meaning nodes are activated during bilingual sentence
production, which in turn are connected to different phonological
representations.

Clearly, word-for-word translations of sentences from one language to
another not only share the same semantic properties, but also many syntactic

16

properties. Although MacKay and Bowman (1969) found an increased
priming effect for word-for-word translations as compared to sentences with
different word orders, they did not attribute this to the effects that syntactic
properties can have on the transfer process, but exclusively to semantic
factors. Although the conceptual similarity between the experimental
sentences probably contributed greatly to the observed transfer effect, the
increased facilitation for word-for-word translations leaves room for syntactic
transferto have contributed to the results.

Models of the code-switching process provide a number of clues to the
possible loci of syntactic transfer. Sridhar and Sridhar (1980) proposed an
interactionist model of overlapping systems to explain the patterns of bilingual
code-switching. A distinction was made between the host language, the
primary language of the discourse, and the guest language, the source
language for mixed elements. The model proposed a comparison stage,
where the guest constituents are checked for their syntactic compatibility with
the host language. According to Sridhar and Sridhar, the placement of a
guest element has to obey the rules of the host language, regardless of the
internal structure of the guest element. They suggested that bilingual mixed
language production involves an "assembly line process"(p.414) where the
syntactic fit of the guest elements into the host language is checked before
the mixed language is produced.

In this model, the constituent structures for the sentence are generated,
and elements of another language are inserted into this frame. Applied to the
phenomenon of transfer, such a model does not encompass the possibility of
structural influence across languages in fluent bilinguals. Although the model
involves the simultaneous activation of both syntactic systems, there is no
interaction between both systems at a level prior to the specification of lexical
elements of the guest language. These elements of the guest language are
checked for their syntactic compatability within the host language only after
the host language has been assigned to the sentence. Thus, one language

17

provides the stmctural frame of the sentence, with elements of the other
language being checked for their syntactic fit before insertion into the host
language. The model does not predict any structural interaction independent
of the lexical elements.

In contrast, Woolford (1983) has proposed a linguistic model of
code-switching in which bilinguals switch freely between grammatical
systems to generate a constituent structure tree. The two grammars of a
bilingual are not altered in any way, but operate just as they do in producing
monolingual speech. However, during code-switching each grammar
generates only partof the sentence. At places where the structural mles of
the two languages overlap, the speaker has joint access to both grammars
because the rules are in fact indistinguishable, eliminating the need for a
specific language assignment.

For the case of transfer, Woolford's model would make different
predictions depending on whether there is phrase structure overlap between
the languages involved. If there is no constituent structure overlap,
Woolford‘s model would not predict transfer, since only one grammar would
generate sentence parts where there is incompatibility of the grammars of
both languages. However, if there is constituent structure overlap between
languages, then transfer should occur as the speaker has simultaneous
access to both grammars. According to Woolford's model, both lexicons of
bilinguals remain entirely separate from each other. If a phrase stmcture mle
is drawn from an area of constituent structure overlap between both
languages, it cannot be distinguished as either language, and belongs to
simultaneously to both grammars. Thus, this model would predict structural
influence across languages at places of constituent structure overlap,
independent of lexical elements, as the sentence frame is formed prior to the
insertion of lexical elements.

Although models of code-switching emphasize syntactic aspects of
bilingual speech production, they have had no impact on explanations of

18

transfer. Models of transfer focus on semantic and pragmatic factors, which
may be necessary to explain transfer, but may not be sufficient. If the
syntactic systems of both languages can be simultaneously accessed during
bilingual production, then certain of the phenomena associated with
code-switching and transfer might both be explained in these terms. But to
my knowledge, the role of syntactic transfer in bilingual language production
has not been assessed under circumstances that minimize the contribution of
semantic and pragmatic factors.

An Experimental Paradigm to Study Syntactic Transfer

The studies on bilingual code-switching seem to point to the
simultaneous availability of the structural systems of both languages during
code-mixed speech, with shared constituent structures favored for switches.
The relevance of this to transfer can be assessed with a phenomenon that is
common in normal monolingual speech production. It shows up as a
tendency to repeat syntactic structures which have occurred in previous
sentences. In conducting sociolinguistic interviews, Weiner and Labov
(1983) found that the probability of a passive construction increased after the
occurrence of another passive in the previous five sentences. Levelt and
Kelter (1982) compared the responses to telephone questions such as 'At
what time does your shop close?’ with responses to questions like 'What time
does your stop close?'. The form of the answers depended on whether the
question was phrased in the prepositional form or not. The prepositional
question was likely to be answered with ‘At five o'clock', whereas the
nonprepositional question was more likely to be responded to with the
answer ' Five o'clock'.

Bock (1986,1987, in press), and Bock and Loebell (1988) have shown
that this tendency cannot be explained by word repetitions, persistence of
detailed meaning features, or event structure organizations. Instead, it seems

19

to involve the persistence of a syntactic structure. In the priming procedure
used in these experiments, subjects produced a priming sentence in one of
two different syntactic forms (e.g. 'The governess made a pot of tea for the
princess' or 'The governess made the princess a pot of tea). Following such
priming sentences subjects described a semantically and narratively
unrelated picture (e.g. a man reading a story to a child) which could be
described in either syntactic form (‘the man is reading a story to the child' or'
the man is reading the child a story'). The frequency of use of a particular
syntactic form changed as a function of the syntactic form of the priming
sentence, with matching forms being used more often than mismatching
forms. Similarly, Bock (1986) found that subjects' tendency to use a passive
sentence increased as a function of having heard a passive priming
sentence, compared to subjects who received the same priming sentence in
the active form. The priming trials were embedded in a mixed list of
sentences and pictures, introduced to the subject as a recognition memory
test. The subject repeated the sentences, described the pictures, and
indicated whether the item had occurred previously in the list. This was done
as to minimize the subject's attention to his/her speech. The obtained
findings of stmctural influence suggest syntactic forms can be primed, even
when the semantic content of successive sentences is dissimilar.

This experimental procedure allows an empirical examination of
syntactic transfer between languages in a controlled experimental setting,
minimizing the effects of shared content. If, during bilingual language
production, the activation of structural procedures in one language influences
the use of corresponding procedures in the other language, or if the same
procedures are involved at some level, then a syntactic priming effect should
be elicited across languages just as it was found within a language. This
prediction accords with Woolford's (1983) proposal that phrase stmcture rules
that are identical in two languages belong simultaneously to both languages,
as they cannot be distinguished as belonging to one particular language. In

20

contrast, if the processes that create syntactic forms in bilinguals are
separate, come from isolated systems, or are controlled completely by the
conceptual features of the thoughts that are communicated, then the syntactic
form of a priming sentence should not influence the syntactic form of a picture
description, unrelated to the prime in meaning, produced in a different
language.

German and English are two languages that share several syntactic
features, and are thus well suited for such an experimental assessment.
Although German is a highly inflected language that shows greater
constituent order flexibility than English, several grammatical functions are
expressed similarly. Like English, German has two semantically comparable
dative constructions. Thus,the sentence:

'The boy sent a letter to his pen pal'
can be expressed as an exact constituent-mapping equivalent in German:

'Der Junge schickte einen Brief an seinen Brieffreund'2
The double-object form of the English dative:

'The boy sent his pen pal a letter'
corresponds to the German:

'Der Junge schickte seinem Brieffreund eine Brief.‘

Likewise, the transitive, active sentence

'The janitor cleans the floors daily'
corresponds to the German sentence

'Der Hausmeister reinigt die Boden téglich.‘

However, German passive constructions are formed by moving the verb to the
final position of the sentence. So, the sentence

'The floors are cleaned daily by the janitor'
translates to the German sentence

'Die Boden werden téiglich von dem Hausmeister gereinigt'
where 'gereinigt' is the past participle of the verb 'reinigen' ('to clean'), and
occurs at the end of the sentence.

21

The phrase stmcture rules for generating the German passive are thus
not identical to the comparable English phrase structure rules. So, according
to Woolford's (1983) model, the phrase structure I’UIOS for generating passive
constructions do not belong simultaneously to both languages. It the overlap
of phrase structure rules is cmcial, then cross-linguistic syntactic priming
effects would not be expected for German/English passive constmctions.

Investigating the importance of phrase structure rules in monolingual
sentence production, Bock and Loebell (1988) found evidence that phrase
structure correspondence is essential for syntactic priming. The sentences:

a. Susan brought a book to study.

b. Susan brought a book to Stella.
have similar word orders and similar metrical patterns, but the constituent
structures vary considerably. lf phrase structure similarities are necessary for
syntactic priming, then a priming sentence like (b) should increase the
probability of subsequent use of the prepositional dative form, whereas
priming sentences like (a) should not, relative to a control form. This is what
Book and Loebell found. Conversely, sentences with constituent structures
similar to target forms but representing different thematic roles showed
comparable priming effects, relative to control forms. Thus, 'the wealthy
widow gave her Mercedes to the church' and 'the wealthy widow drove her
Mercedes to the church' behaved similarly as primes, even though the
thematic roles of W differ. Such findings suggest that the processes
that create sentence structures are to some degree dissociable from
underlying thematic relationships. 80, monolingual priming effects seem to
tap phrase structure similarities. Across languages, then, phrase structure
similarity should yield cross-language priming effects, if the procedures for
generating the structures are shared by both languages.

CHAPTER II
EXPERIMENT 1

Method

Subjects, 48 native German speakers living in Michigan served as
subjects. Each subject received $10 for participating in the two experimental
sessions. In order to qualify as a subject, the German speakers had to have
lived a minimum of 2 years in the United States. Most subjects had studied
English extensively in school prior to coming to the United States. All of them
were able to converse fluently in both languages. Their median age was 39
years (range 18-66), and the median length of residence in English speaking
countries was 16 years (range 2-36). Appendix A lists the subjects, their age,
occupation, and length of residency in an English speaking country.

Materials, The German materials were analogous to the English
materials. For brevity, only the English ones will be described.

The experimental priming materials consisted of 16 sets of dative
sentences, each paired with a dative target picture, and 16 sets of transitive
sentences, each paired with a transitive target picture. Within the pairs,
systematic narrative or semantic relationships were avoided. Appendix B
contains a complete list of the priming materials.

Each of the 16 dative sentence sets included one prepositional dative
(e.g. the girl read the newspaper to the blind woman), and the corresponding
double-object form of the dative (the girl read the blind woman the
newspaper). Each item was paired with a picture that could be described in
either a prepositional or in a double-object dative form. All the dative priming
pictures depicted an action that involved an agent, a patient, and a

22

23

beneficiary of the action (e.g. children giving flowers to a teacher). Half the
pictures depicted an action proceeding from left to right (the actor situated in
the left part of the picture, and the beneficiary in the right part of the picture).
In the other half of the pictures, the action proceeded from right to left.

The remaining experimental items consisted of 16 transitive sentences
pairs, each including an active and a corresponding full passive form. Each
of these was paired with one of 16 pictures that could be described with a
sentence in either an active or a passive form. All the transitive pictures
involved an agent and a patient. All of the pictures depicted an event with an
inanimate agent, one half with an animate patient, the other half with an
inanimate patient. Half showed the agent on the left, and the other half on the
right side of the picture.

In addition to the experimental priming materials, there were 80 filler
items, 40 pictures and 40 sentences. Most of the filler pictures depicted
actions or states that were described with intransitive sentences, 9.9. a cat
sleeping, two girls running. The filler pictures, as well as all experimental
priming pictures, consisted of black and white line drawings. The filler
sentences were in a variety of constructions, such as locatives, reflexives, and
complement constructions. All the sentences had natural literal translations
in German. Sentences that had language specific words in them, such as
names that exist only in one language, were excluded. Most of the pictures
and many of the sentences were adapted from experiments performed by
Bock(1986)

From the pool of experimental priming and filler items, two lists of 224
items were created. They contained only English sentences. The
grammatical correctness of the English sentences was checked by a native
English speaker. Both lists had identical filler pictures, filler sentences, and
experimental target pictures. Each list had an equal number of each type of
priming trial (8 prepositional datives. 8 double-object datives. 8 active
transitives, and 8 passive transitives). Each priming sentence set was

24

represented one and only one priming trial in each list. Across lists, both
priming sentences from the set with which a picture was paired occurred
once. The sentences from the same set appeared in the same location in
both lists. Each priming sentence immediately preceded its paired picture.
These priming trials were always separated by five filler items. Consecutive
priming trials never involved the same type of sentence/picture pairs
(transitives versus datives). There were no more than three consecutive
sentences or pictures, and care was taken that consecutive Items were
semantically and narratively unrelated. The filler items were randomly
arranged within the lists, with each filler item occurring twice within each list.
One-sixth of the filler items were repeated in the first third of each list, an
additional one third were repeated in the next third of each list, and the final
half of the filler items occurred as repetitions in the last third of the list.

The pictures for each English list were photOCOpied and taped onto
15x22.5 cm index cards for presentation to the subjects. Blank index cards
occurred in the list positions where sentences were presented. The
sentences were read by the experimenter.

Two additional lists contained the literal translations of the English
sentences into German. Their translations were checked by two native
German speakers (other then the experimenter) for naturalness and
grammaticality.

Engaging, The experiment was divided into two sessions for each
subject. Both parts took place in the subjects' homes. Half the subjects
received one of the German lists in the first session, and the other half
received one of the English lists first. In the second session, the subject
received the alternative list (not the translation-equivalent list) in the other
language. The two sessions were separated by a minimum of one week, and
each of them was recorded on audio tape. All the experimental meetings
were conducted by the author who speaks fluent English, though it is
accented.

25

The subjects were told that the experiment tested the influence of the two
languages on picture and sentence recognition. They were instructed, in
German, to carefully examine each picture and listen to each sentence in
order to be able to recognize them later. Subjects indicated after each picture
and each sentence whether they had encountered that particular item before
in the experiment. This was the recognition-memory mask for the experiment,
used to minimize the subject's attention to his or her speech. Three practice
trials established that the subject correctly understood the procedure.

Subjects turned the index cards at their own pace. When they reached a
picture, they described what was happening in the picture in one sentence
without using personal pronouns, and then said 'yes' or 'no' depending on
whether they had encountered the picture before during the session.
Subjects described the pictures in the language that differed from that used in
the list, so German primes (for example) were followed by English picture
descriptions, and vice versa.

Whenever a subject reached a blank card, the experimenter read a
sentence, at a normal rate and with normal intonation. Subjects were
instructed to repeat the sentence after the experimenter and to indicate with
'yes' or 'no' whether they had heard the sentence before in the current
experimental session. When repetition errors occurred, the experimenter
repeated the entire sentence, and the subject said it again. In the rare cases
when the subject was unable to retrieve a word needed to describe a picture,
the experimenter supplied the relevant word. The experimenter provided
feedback after each experimental item to indicate whether the subject's
yes/no recognition response was correct.

After the first experimental session, the subjects answered questions
concerning their language history. Appendix C gives the questionnaire,
which was adopted and modified from a questionnaire by Schwanenflugel
and Rey (1986). The questionnaire was phrased in German, and the subjects
responses, also in German, were recorded. Finally, after completing the

26

second experimental session, subjects were asked, in English, whetherthey
had recognized any similarities between the sentences and the pictures,
whether they had mentally translated from one language to the other, whether
the sentences appeared to be in any way unusual, and whether they thought
that the sentences they heard influenced their picture descriptions. None of
the subjects reported noticing any structural similarities between the priming
sentences and experimental pictures or thought the sentences unusual.
None of the subjects reported mentally translating from one language to the
other before responding.
I Scoring, The descriptions of the experimental pictures were transcribed
from the audio tapes and scored for their syntactic form. Only the first full
sentence in each picture description was scored. All responses in the
categories described below had to include a complete clause without
personal pronouns, except for omissions of articles and the capula. Only
those picture descriptions were used for which the corresponding syntactic
form was also possible. For example, to be scored as an English active, a
sentence had to have a grammatical passive counterpart with its subject as
the by-phrase object and its direct object as the subject.

Dative picture descriptions produced in English were scored as
prepositional datives when the agent was the subject, the patient was the
direct object immediately after the verb, and the beneficiary followed the
direct object as the object of the prepositional phrase. Descriptions were
scored as double-object datives when the agent was the subject, the
beneficiary was the direct object, and the patient was the second object.
Prepositional- and double-object datives produced in German were scored
according to the same criteria.

Transitive picture descriptions produced in English were scored as active
when the description included the agent of the depicted action in subject
position and the patient in direct object position. Passives were scored when
the patient was in subject position, and the agent in object position within a

27

by-phrase, and the main verb was preceded by a form of be or get. Transitive
picture descriptions in German were scored as active according to the same
criteria as the English descriptions. However, in the German passive, the
main verb is placed at the end of the sentence, following the agent. The
patient is in subject position, the agent in object position within a

prepositional phrase (in German van), and the main verb is preceded by a
form of werden (corresponding to the English verb 'be' ).

In cases when a subject asked the experimenter for lexical help before
producing a sentence, the responses were scored as 'other'.

We. Every subject described 8 transitive pictures
in each of the two transitive priming conditions (active and passive) in each
language (German priming sentences with English picture descriptions;
English priming sentences with German picture descriptions), and 8 dative
pictures in each of the two dative priming conditions (double object and
prepositional) in each language. The design was completely within subjects,
with every subject contributing to every experimental condition.

Analyses of variance were performed on the number of responses
produced by each subject in each experimental condition. Separate
analyses were done for dative utterances, with prepositional and double
object datives as dependent variables, and for transitives, with active and
passive utterances as dependent variables. Analyses were also done
separately treating subjects and items as random effects.

Results
The totals for each utterance type for the dative priming conditions are

presented in Table 1.

 

Insert Table 1 about here

 

There was a general tendency for the form of the utterance to match the form

28

Table 1

II I IIIII . D I. E" G If

Priming condition

Emerimenu

Form of utterance

 

Prepositional Double-object

 

English Prepositional dative
English Double-object dative

German Prepositional dative
German Double-object dative

German utterances

42 196
39 21 1

English utterances

210 52
185 79

 

29

of the prime, regardless of language. This effect was strongest for double
object utterances. The number of double object utterances was higher when
the preceding prime was a double object form than when the prime was a
prepositional form (E (1,47)=5.31, n< .05 for subjects; and E(1,15)=4.84, n<.05
for items).

For prepositional dative utterances, there was a similar tendency to use
the same structural form that was provided in the priming sentence to
describe pictures. However, it was only marginally significant over subjects
(E(1,47)=2.79,_p=.10), and not significant over items (E(1,15)=2.75, n=.12).

The pattern of results was roughly the same regardless of the language
of the prime, as reflected in the absence of significant interactions between
language and priming condition for the production of both prepositional
(E(1,47)=1.90, p>.10 for subjects, and E<1 for items) and double object
datives (E(1,47)=1.34, p>.10 for subjects, and E< 1 for items).

There were more prepositional dative utterances in English
(E(1,47)=110.47,,p< .01 for subjects, and E(1,15)=85.13, p<.01 for items), and
more double object datives in German (E(1,47)=91.52, Q<.O1, and
E(1,15)=75.41, p<.01). These differences reflect a language bias: The
prepositional dative is much more restricted in German than in English.

The results for the transitive priming conditions are given in Table 2.

 

Insert Table 2 about here

 

There was no influence of the priming sentence in one language on the
syntactic form of the picture description in the other language (for all actives
F's < 1; for passives E(1,47)=2.34,n>.10 over subjects, and
E(1,15)=1.24,p>.10 over items). Interactions with language were also
nonsignificant (all F's < 1).

The fluency of the bilingual subjects is attested by their questionnaire
responses. 56% of all subjects could not specify one particular language as

30

Table 2

III III 'I"E"I:l"
Experiment;

Form of utterance

 

Priming condition Active transitive Passive transitive

 

German utterances

English Active Transitive 89 117
English Passive Transitive 89 105

English utterances

German Active Transitive 98 140
German Passive Transitive 94 132

 

31

being their strongest language. 42% of the subjects used English at home as
the language of communication, with an additional 30% using both
languages. Asked to subjectively rate their level of competence in their
languages on a scale from 1(rather bad) to 7(native like), subjects ratings
their English skills yielded a mean of 6.5, and their German skills a mean of
6.8, reflecting a high degree of perceived bilingual competence.

D' .

This study has provided some evidence that for sentence forms with
similar constituent structures, in different languages, syntactic transfer may
occur even when the possibilities for semantic transfer are minimized. For
these forms, in the speech of very fluent bilinguals, the structural aspects of
either language may transfer. Although this effect appears stronger when the
native language primed the second language than vice versa, this interaction
did not reach significance. If such results can be replicated and extended, it
contradicts the common view that transfer is associated only with a weak
second language. Instead, transfer may be more closely symmetric, and at
best bidirectional.

The priming effect was more clearly evident for the double-object dative
utterances than for the prepositional datives. This may be attributable in part
to prescriptive German grammar rules that forbid the prepositional dative form
when the preposition is followed by a dative-inflected noun phrase.
Addressing the usage of this form, Kufner (1962) wrote: "Once again we must
make it clear that this possibility is a peculiarity of English and it cannot be
transferred to German. Extensive drills are needed to avoid this mistake
(p.44)." These prescriptive grammar rules might contribute to a weaker
tendency for prepositional dative utterances to be primed by preceding
utterances of the same type. However, most of the native German speakers
who participated in the study used this construction and none of them
reported that it sounded awkward.

32

The transitive constructions failed to produce syntactic correspondence
in the subsequent picture descriptions. For passives, this might be due to
cross-language differences in constituent representations. The phrase
structure of the English passive is different from the German, in that German
puts the main verb at the end of the sentence.

The weakness in this account is that English and German active
sentences do, share the same constituent structure, and they also did not
show a priming effect. A more parsimonious summation is that the priming
effects for transitives were simply too weak to reveal differences between
passives and actives. Book's work (1986) suggests that transitive priming
effects may be less stable than dative priming effects, so that any transfer of
their syntactic structure might go undetected in a cross-language condition.

However, the absence of transfer in the mapping of thematic roles to
grammatical functions in passives is consistent with other research on the
effects of thematic role assignment on the priming of syntactic constructions
(Book and Loebell, 1988). As in intralanguage priming, cross-language
phrase structure similarity may be the most important component of transfer.
This accords with Woolford's (1983) proposal that phrase structure mles that
are identical in two languages belong simultaneously to both languages, as
they cannot be identified as belonging to one particular language. The
priming of a structural representation in one language would then lead to an
increased probability of using the same structural configuration in a
subsequent sentence in either language.

CHAPTER III
EXPERIMENT 2

The first exeriment presupposes that structural priming occurs within
each language. Thus, in a monolingual setting, the stnrcture of one language
should predispose the production of the same structure in a subsequent
sentence. Bock (1986, 1987, in press) has shown that such processes occur
for English speakers. The following experiment assessed whether such a
structural influence can also be found within a monolingual German setting.

The dative priming effect in Experiment 1 may be attributable, at least in
part, to a phrase structure overlap of the two languages involved, German
and English. Such effects should also occur within a monolingual setting
where there is phrase structure overlap. Moreover, active and passive
priming should be obtained. To assess the reliability of within-language
syntactic priming, the second experiment examined structural persistence
effects in a monolingual German setting.

This experiment also examined a complexity of German dative forms.
The inflectional system of German allows three possibilities for forming
datives (prepositional, double object, and inflectional). Two of these
correspond transparently to the English dative forms:

'Der Junge schrieb einen Brief an seinen Brieffreund.‘

('The boy wrote a Ietterto his penpal.)

and

'Der Junge schrieb seinem Brieffreund einen Brief.‘

(’The boy wrote his penpal a letter.')

However, the third possibility has no exact English equivalent:

33

34

'Der Junge schrieb einen Brief seinem Brieffreund'.

This form corresponds roughly to the prepositional form, but with the
preposition represented in appropriate noun inflections.

Experiment 1 employed only German prepositional dative constructions,
(constructions containing a free preposition), as priming sentences, since
only this dative construction has a stmctural equivalent in English. In the
English-German priming condition the inflectional dative form was
occasionally used (out of 514 dative picture descriptions in 26 (5.06%)
responses). However, the numbers were roughly equivalent across priming
conditions.

Experiment 2 investigated priming effects for all three German dative
constructions in a monolingual German setting. Each of the three forms
served as priming sentences, followed by a picture that could be described in
any of the forms. The results of this experiment should help to establish
whether constituent structure is sufficient for syntactic priming to occur. If it is,
all of the sentence forms examined should tend to be used more often after
similar than after dissimilar primes.

To the extent that factors such as thematic or grammatical roles are
involved, prepositional and inflectional forms should prime each other, as
well as themselves. This is because the assignment of thematic roles to
grammatical functions is the same in both dative forms. If the German
inflectional dative is analogous to the German prepositional dative at some
level of processing that trades in such roles, they should behave somewhat
similarly as priming sentences, eliciting both prepositional and inflectional
descriptions more often than double object descriptions.

M91090

Subjects, 48 monolingual native Germans living in West Germany
participated as subjects. Although all West Germans have some contact with
English, all the subjects were drawn from a largely monolingual German

35

environment. None of the subjects reported fluency in another language
besides German, and none of the subjects had regular contact with speakers
of other languages. An attempt was made to equate the subjects of
Experiment 1 and 2 with respect to their age and educational level. The
median age of the subjects was 41 (range 1872).

MM There were 16 sets of transitive priming sentences, each
paired with a transitive picture. These materials were identical to the German
materials used in Experiment 1.

In addition to the transitive priming sets, there were 18 sets of dative
priming sentences, each paired with a dative picture. Each set consisted of
three different forms of the priming sentence, the prepositional form, the
inflectional form, and the double object form, as described above.

Each priming trial was separated by 12 filler items, including one priming
trial of the other type, with no two priming trials of the same type occurring
consecutively within a list. To accommodate the two additional dative priming
trials in this experiment, there were twenty additional filler items at the end of
the list to separate these priming trials. These fillers were similar to the other
filler items.

The priming sentences, the priming pictures and the filler items were
assigned to six 244-item lists. The constraints on list construction were
identical to those in experiment 1, with list arrangements that
counterbalanced dative and transitive priming trials within and across the
lists.

Mum. The procedure followed that described for Experiment 1,
except that each subject received only one list and repeated the sentences
and described pictures only in German.

Scoring, The scoring of the transitive and dative picture descriptions was
carried out as described in Experiment1. Additionally, an inflectional dative
required the theme to be followed directly by beneficiary, with the theme
marked with the accusative and the beneficiary with the dative inflection.

36

W Each subject described 8 transitive pictures in

each of the transitive priming conditions (active and passive) and 6 pictures
in each of the dative priming conditions (prepositional, inflectional, and
double object). The design was again completely within subjects. The
analyses were done separately with subjects and items as random effects.

13850115

The total numbers of dative utterances as they occurred in each
condition are presented in Table 3. Too few prepositional and inflectional
forms were produced to merit analysis, so only the double object forms were
evaluated statistically.

 

Insert Table 3 about here

 

Analyses of variance were performed on the number of responses
produced by each subject in each experimental condition. For double
objectutterances, though there was a tendency toward increased occurrence
after double object primes, this difference was not significant
(E(2,94)=.71,n>.10, and E(2,34)=1.61,Q>.10 respectively for subjects and
items).

The results for the transitive priming conditions are given in Table 4.

 

Insert Table 4 about here

 

There was again a slight tendency for the primed syntactic form match the
syntactic form of the following picture description, but again none of the
effects reached significance; for actives( E(1,47)=1.00,p>.10, and
E(1,15)=0.67,n>.10, for subjects and items respectivelY); or for passives
(E(1,47)=0.71,p>.10, and E(1,15)=0.92,Q>.10, for subjects and items
respectively).

37

Table 3

II I [llll 'DI' E" C II

 

 

Expedmantz
Form of utterance
Priming condition Prepositional Double-object lnflectional
Prepositional dative 12 128 15
Double-object dative 6 146

lnflectional dative 6 138

 

38

Table 4

II I Hill 'I .I. E" E II
$030018an

Form of utterance

 

Priming condition Active transitive Passive transitive

 

Active Transitive 115 86
Passive Transitive 106 93

 

39

D' .

The monolingual experiment produced priming patterns that were
generally in line with the predictions of the constituent structure hypothesis,
but the trends were weak and unreliable. Overall, more double object
utterances were produced after double object priming sentences than the
other types of priming sentences, and more prepositional dative utterances
followed prepositional priming sentences.

The only departure from the predicted results appeared for inflectional
datives. lnflectional utterances more often followed prepositional priming
sentences than others. Such a result suggests that the assignment of
thematic roles to grammatical functions might have contributed to the choice
of syntactic form in the production process, since prepositional and
inflectional datives share thematic to grammatical role mappings. However,
so few inflectional datives were used that this conclusion must remain
tenuous, at best.

The small number of prepositional and inflectional dative utterances is
traceable to their role in German. Just as prescriptive grammar rules do not
allow for prepositions to be followed by dative objects (see Kufner,1962), the
inflectional form of the dative is similariy problematic. It may sometimes be
used to put emphasis on the second noun: Collins' German Grammar (1985)
states that ”A direct object usually follows an indirect..but the indirect object
can be placed last for emphasis, providing it is not a pronoun (p.224)."
However, the experimental conditions rarely should have satisfied
the contextual conditions necessary for such constructions, resulting in very
few appearances of the form.

For transitives, there was again a small but nonsignificant tendency for
utterances to match the syntactic form of the prime. Though little can be made
of such weak effects it may be worth observing that the transitive priming
results differ from those obtained in Experiment 1. Whereas Experiment 1

40

obtained a somewhat reversed priming effect for passives, with utterances
showing no positive tendency to match the priming sentence, both active and
passives in Experiment 2 patterned in the expected direction. Thus, matching
constituent structure might be the best predictor of priming effects both within
and across languages.

CHAPTER IV
GENERAL DISCUSSION

Two experiments were conducted to investigate syntactic aspects of
bilingual language production. Experiment 1, using fluent English/German
bilinguals, showed that syntactic transfer can occur from English to German,
and vice versa, at places where there is structural overiap between
languages. In the dative condition the constituent structures of English and
German double-object and prepositional datives are identical. As predicted,
subjects more often produced a given syntactic form in their picture
descriptions when it matched the form of the preceding priming sentence than
when it mismatched, even though the priming sentences were in a different
language from the picture descriptions. This effect was reliable for
double-object datives but marginal for prepositional datives, perhaps
because of the marginal status of such forms in German. Although the
priming effects appear somewhat stronger for native language to second
language priming than vice versa, the interaction was not significant. At a
minimum, then, there appears to be some transfer from the weaker to the
stronger language.

At the level of constituent structure elaboration, Woolford (1983)
proposed that phrase structure rules common to both languages belong
simultaneously to both languages. Since the phrase structure rules for
generating English dative constructions have analogs in German, the
production of a dative sentence in one language should predispose the use
of the same sentence structure in a subsequent utterance in the other

41

42

language. The dative priming condition of Experiment 1 provided some
support for this claim.

In the transitive priming condition, no significant priming effects were
obtained. This was predicted for the passive sentence constructions, as the
constituent structures of these constructions vary in English and German.
However, active sentence constructions, which have the same constituent
structure in both languages, also did not produce a significant increase in the
number of active responses.

Experiment 2 replicated the first experiment in a monolingual setting.
Although the overall pattern of results favored more frequent use of forms that
matched primes among datives as well as passives, none of these effects
reached significance.

The results of the first experiment have implications both for models of
transfer as well as for attempts to explain code-switching phenomena.
Attempts to explain transfer have been restricted to transfer of semantic and
conceptual properties (MacKay, 1982; MacKay & Bowman, 1969; Harrington,
1987). These models do not address the possibility of syntactic transfer,
independent of semantic properties. However, under circumstances that
minimized semantic transfer, Experiment 1 showed that bilinguals transfer
dative structures somewhat independently of semantic properties.
Furthermore, these processes seem to operate in both directions in fluent
bilinguals.

In contrast to models of transfer, models of code-switching do address
syntactic processes in bilingual language production. Several investigators
have pointed to the rule-governed occurrence of code-switches (Pfaff, 1979;
Poplack, 1980,1981; Timm, 1983), and have suggested that the syntactic
systems of both languages might be involved in determining the acceptability
of code-switches (Sridhar & Sridhar, 1981; Woolford, 1983). However, the
phenomena of code-switching are generally regarded as distinct from those
of transfer. In light of these limitations of previous attempts to explain transfer

43

and code-switching, the present research might be regarded as a connecting
link. It incorporates the transfer of structure with the issue of syntactic
interaction in fluent bilinguals.

Although bilingual production is complicated by the need to segregate,
integrate, and coordinate information from two languages, it can be viewed
from the perspective of models of normal processes. Bilinguals have to
transform an abstract thought representation into a structured linguistic form.
On a non-linguistic thought-processing level, bilinguals probably have one
conceptual store that serves as the input to either language (Hakuta,1986).
On the other end, the phonological forms of words are separated. So, for
example, the concept of 'tree' has two phonological representations for a
German/English bilingual: The English /tri/ and the German /bawm/. Thus,
somewhere in the transition from a thought to an actual utterance, a common
system becomes divided.

Psycholinguistic models of language production have attempted to
specify how abstract linguistic representations are transformed into speech.
Such models suggest that the syntactic and lexical elements of a sentence
are to some degree separable. Some of the evidence for such separability
comes from naturally occurring speech-errors, such as "Start the boat on the
motor‘ [motor on the boat], and "She's already trunked two packs'Ipacked two
trunks] (Garrett, 1975). The natural account of such errors, found across
virtually all models of production (Bock, 1982, 1987; Dell,1986; Garrett, 1975,
1980, 1982; Stemberger,1985) is that the syntactic frame for an utterance is
specified somewhat independently of the words it contains.

If the syntactic elements of an utterance are specified to some degree
independently of the lexical elements, then the possibility arises that during
bilingual language production, the processes that create the syntactic frame
of an utterance do not vary from the processes as they occur in monolingual
production at places where there is constituent structure overlap between
languages. It bilingual production is similar to normal production, the

44

phonological forms of words may be inserted into the slots of the frame as it
develops. Considerable research has been concerned with whether these
words originate in a common lexicon, though the questions have more often
concerned semantics than phonology.

Some researchers have maintained that there are different lexicons,
functionally separate (Kolers, 1963, 1966; Paivio & Lambert, 1981; Paivio &
Lambert, 1988), while others maintained that they function interdependently
(Caramazza & Brones, 1980; Mo Cormack, 1977; Schwanenflugel & Rey,
1986). Kolers and Gonzalez (1980, p.54) concluded that ”on the basis of
census alone, there is more evidence for the interdependence than for the
independence hypothesis. However, Hakuta (1986) cautioned that it is often
unclear what experimental studies are tapping, and that methodological
problems and different procedures make it impossible to draw definite
conclusions.

Grosjean (1982) noted that recall tasks of isolated words might reflect
more about strategies for recalling word lists than about lexical organization.
Paradis (1980) pointed out that many studies fail to make an explicit
distinction between a conceptual memory store independent of language and
a linguistically constrained semantic store. Whereas the conceptual
representation of propositions is likely to be language independent, lexical
items may be specified fortheir meaning and grammatical category in a
language-dependent way. 'Data-driven' tasks, such as word fragment
completion studies (e.g. Watkins & Peynircioglu, 1983), might successfully tap
language dependent stores. On the other hand, conceptual representations
in bilinguals are likely to be shared, and studies which report no evidence for
language specificity, such as studies that show cross-language influences,
might reflect this conceptual system.

Paradis (1980) proposed a model in which the bilingual has two
lexicons, one for each language. These lexicons are in turn connected to a
conceptual store that represents experiences, concepts and mental

45

representations in a language-free way. According to Paradis, the greater the
number of conceptual features shared by a word and its translation
equivalent, the more they will tend to elicit the same responses, i.e., they will
tend to be associated with each other. Similarly, Potter,So, Von Eckardt, and
Feldman (1984) proposed that lexical items are connected both within and
between languages via a conceptual system that is independent of language.
The contradictory findings about the semantic organization of bilinguals might
reflect a confounding of lexical and conceptual factors which makes
experimental outcomes task-dependent.

Thus, there is little consensus on the nature of the organization of the
bilingual lexicon. Nevertheless, researchers have focused on the bilingual
lexicon to the exclusion of syntactic factors to explain bilingual language
production. The research presented here found evidence that syntactic
factors play a role in bilingual language production somewhat independently
of semantic factors.

Several questions remain. Although these experiments tried to eliminate
most conceptual factors, there may be hidden influences of conceptual
factors. It is possible that instead of or in addition to the phrase structure, the
event role structure conveyed by an utterance plays an important part in cross
language influences. Though event roles have been found to have little
impact on the formation of constituent structures in English sentences, it is still
necessary to look at these processes in bilingual production.

Furthermore, the sentence types employed to investigate syntactic
transfer were limited to datives and transitives. It remains to be seen if the
findings can be extended to include other types of sentences.

A further limitation of this study is its restriction to only two languages.
English and German, both Germanic languages, share many sentence
constructions with similar constituent structures. Constituent structure
similarity appeared to be a major determiner of transfer in these closely
related languages. Investigating transfer between less closely related

46

languages might reveal other factors that determine transfer.

Another possible extension of the present experiments might forge a
tighter connection between transfer and code-switching. Using the same
research paradigm, bilingual subjects could be asked to switch languages
during their picture description. Depending on the form of the preceding
priming sentence, the switching point in the picture description might change.
Furthermore, in a highly inflected language such as German, the inflections
on nouns and articles prior to the onset of a code-switch might reveal the
stmcture of the attempted utterance prior to the code-switch. These
inflections might be influenced by the syntactic form of the preceding
utterance.

Finally, although the present research has found some support for
cross-language transfer of syntactic factors, some effects were weak and
unreliable, and need to be replicated with more subjects.

Conclusion

Two experiments explored a set of hypotheses about the functional
separation of linguistic systems in bilinguals. Using an experimental
paradigm that minimized content overlap across utterances in different
languages, the experiments examined whether and how the production of
sentences with certain structural features in one language changed the
likelihood of producing sentences with similar structural features in a different
language. Experiment 1 provided some evidence for the hypothesis that the
processes that created the constituent structure representation of a sentence
might be shared across languages when the phrase structure rules have
analogs in both languages.

The literature on transfer has focused almost exclusively on the semantic
aspects of transfer, largely ignoring syntactic aspects. The present
experiments showed that structural transfer in bilinguals may be bidirectional.
As the literature on code-switching suggests, during code-switching both

47

languages are simultaneously active. The current results suggest that
transfer and code-switching might both be reflections of the same underlying
processes of interacting linguistic systems.

Gass and Selinker (1983) pointed out that more analytical studies of
transfer are needed to assess actual second language learners under clearly
specified conditions. Addressing code-switching, Sridhar and Sridhar (1982)
noted that ”syntax has been a stranger to the psychology of bilingualism.“
They argued that here is a pressing need in psycholinguistic research on
bilingualism to study sentential processing with adequate attention to
syntactic factors. This study takes steps in these directions.

48

Footnotes

1The use of 'word order' in the competition model refers to the serial
string of words within a sentence, rather than a structural representation of a
sentence.

2The sentence: 'Der Junge schickte einen Brief an seinen Brieffreund
does not contain dative case inflections. The case inflections indicate that
both 'einen Brief and 'seinen Brieffreund' are accusative noun phrases.
However, the constituent structure of this sentence is equivalent to the
English prepositional dative. Thus, this sentence structure will be referred to
as the German prepositional dative to indicate the compatibility of the English
and the German sentence form.

APPENDICES

APPENDIX A

List of Bilingual Subjects, with Occupation, Age, and Length of Residency
in an English Speaking Country

APPENDIX A

 

.‘10 3|. . 0:01 .IIM. 0.0. ;o: .10 :gouo :sozg 'g
E I. I S I . E I

Subject# Occupation Age Length of Residency in US.
1 Housewife 50 29
2 Graduate Student 27 2
3 Housewife 49 25
4 Graduate Student 28 6
5 Housewife 66 31
6 Graduate Student 33 15
7 Accountant 53 32
8 Professor 37 29
9 Librarian 45 24
10 Teacher 41 12
11 Professor 26 2
12 Graduate Student 27 3
13 Businesswoman 47 27
14 Housewife 35 1 1
1 5 Housewife 45 24
16 Graduate Student 25 3
17 Teacher 50 22
18 Professor 44 2
19 Graduate Student 23 8
20 Student 21 4
21 Psychologist 32 3
22 Professor 31 5
23 Student 18 4
24 Farmer 26 2
25 Real Estate Agent 50 25
26 Businesswoman 47 1 8
27 Secretary 58 30
28 Teacher 33 2
29 Secretary 59 32
30 Graduate Student 25 5
31 Graduate Student 28 5
32 Professor 32

49

36

33
34
35
36
37
38
39

41
42
43
45

47
48

Teacher
Graduate Student
Teacher

Teacher
Graduate Student
Librarian
Secretary
Secretary
Secretary
Professor
Teacher
Graduate Student
Professor
Professor
Professor
Professor

50

35
32
53
27
27
47
42
52
37
51
35
26
59
43

47

10
1O
28

29
29
25
26
24
1 1

46
1 8
20
27

APPENDIX B

Dative and Transitive Priming sentences and Pictures
Experiment 1

51

APPENDIX B

D' II .I. E" El IE'I
Expolimomd.

Dl' . II I

1a.The girl bought a newspaper for the blind woman.
Das Madche kaufle eine Zeitung fiJr die blinde Frau.
b.The girl bought the blind woman a newspaper.
Das Madchen kaufte der blinden Frau eine Zeitung.
2a.The little boy wrote a letter to his pen pal.
Der kleine Junge schrieb einen Brief an seinen Brieffreund.
b.The little boy wrote his pen pal a letter.
Der kleine Junge schrieb seinem Brieffreund einen Brief.
3a.The grandmother sewed a dress for her granddaughter.
Die GroBmutter nahte ein Kleid fr'ir ihre Enkeltochter.
b.The grandmother sewed her granddaughter a dress.
Die GroBmutter nahte ihrer Enkeltochter ein Kleid.
4a.The old man left the valuable coin collection to his nephew.
Der alte Mann hinterlieB die wertvolle Mr’inzsammlung fiir
seinen Neffen.
b.The old man left his nephew the valuable coin collection.
Der alte Mann hinterlieB seinem Neffen die wertvolle Miinzsammlung.
5a.The woman handed the screaming baby to her husband.
Die Frau reichte das schreinende Baby an ihren Mann.
b.The woman handed her husband the screaming baby.
Die Frau reichte ihrem Mann das schreinde Baby.
6a.The rich farmer bought a horse for his son.
Der reiche Bauer kaufte ein Pferd fiir seinen Sohn.
b.The rich tamer bought his son a horse.
Der reiche Bauer kaufte seinem Sohn ein Pferd.
7a.The lawyer sent the contract to his client.
Der Rechtsanwalt schickte den Vertrag an seinen Klienten.
b.The lawyer sent his client the contract.
Der Rechtsanwalt schickte seinem Klienten den Vertrag.
8a.The father promised a car to his daughter.
Der Vater versprach ein Auto an seine Tochter.
b.The father promised his daughter a car.

52

Der Vater versprach seiner Tochter ein Auto.
9a.The musician sold some cocaine to the undercover agent.

Der Musiker verkaufte etwas Kokain an den Agenten.

b.The musician sold the undercover agent some cocaine.

Der Musiker verkaufte dem Agenten etwas Kokain.
10a.The land lady rented three rooms to the couple.

Die Hausbesitzerin vermietete drei Zimmer an das Ehepaar.

b.The land lady rented the couple three rooms.

Die Hausbesitzerin vermietete dem Ehepaar drei Zimmer.
11a.The young woman made a cup of tea for her aunt.

Die junge Frau machte eine Tasse Tee ffir ihre Tante.

b.The young woman made her aunt a cup of tea

Die junge Frau machte ihrer Tante eine Tasse Tee.
12a.The architect built a new house for his mother.

Der Architekt baute ein neues Haus for seine Mutter.

b.The architect built his mother a new house.

Der Architect baute seiner Mutter ein neues Haus.
13a.The man baked a cake for his wife.

Der Mann backte einen Kuchen ft‘rr seine Frau.

b.The man baked his wife a cake.

Der Mann backte seiner Frau einen Kuchen.
14a.The hostess fixed some dessert for her guests.

Die Gastgeberin bereitete eine Nachtisch fr'ir ihre Gaste.

b.The hostess fixed her guests some dessert.

Die Gastgeberin bereitete ihren Gasten einen Nachtisch.
15a.The wealthy widow left her money to the church.

Die reiche Witwe vermachte ihr Geld an die Kirche.

b.The wealthy widow left the church her money.

Die reiche Witwe vermachte der Kirche ihr Geld.
16a.The young man wrote an apology to his fiancee.

Der junge Mann schrieb eine Entschuldigung an seine Verlobte.

b.The young man wrote his fiancee an apology.

Derjunge Mann schrieb seiner Verlobten eine Entschuldigung.

53

[If E . IIE'I

1. Girl throwing ball to girl

2. Boy giving present to girl

3. Girl giving flowers to teacher

4. Woman showing dress to man

5. Liberian giving book to boy.

6. Salesman showing car to couple.

7. Boy handing plate to boy

8. Girl handing mug to boy.

9. Girl handing paintbrush to boy

10. Girl reading book to boy

11. Boy and girl giving flowers to man
12. Policeman giving ticket to man.

13. Boy giving ”K” to girl

14. Woman throwing bone to dog

15. Boy and girl showing picture to teacher
16. Nurse giving stethoscope to doctor

54

11' .III

1a.The engine turned the wheel slowly.
Der Motor drehte langsam das Rad.
b.The engine was turned slowly by the engine.
Das Rad wurde langsam von dem Motor bewegt.
2a.Many people attended the concert.
Viele Leute besuchten das Konzert.
b.The concert was attended by many people.
Das Konzert wurde von vielen Leuten besucht.
3a.The chemical waste poisoned the river.
Der chemische Abfall vergiftete den Flu B.
b.The river was poisoned by the chemical waste.
Der FluB wurde von dem chemischen Abfall vergiftet.
4a.The customs official opened the suspicious suitcase.
Der Zollbeamte cffnete den verdachtigen Koffer.
b.The suspicious suitcase was opened by the customs
omdd.
Der verdéchtige Koffer wurde von dem Zollbeamten geéffnet.
5a.The picture on the wall concealed the safe.
Das Bild an der Wand verbarg den Safe.
b.The safe was concealed by the picture on the wall.
Der Safe wurde von dem Bild an der Wand verborgen.
6a.The artist painted the nude woman.
Der Ktinstler malte die nackte Frau.
b.The nude woman was painted by the artist.
Die nackte Frau wurde von dem Klinstler gemalt.
7a.A gang of teenagers mugged the salesman.
Eine Gruppe Jugendlicher iiberfiel den Verkaufer.
b.The salesman was mugged by a gang of teenagers.
Der Verkéufer wurde von einer Gruppe Jugendlicher
tiberfallen.
8a.The sun warmed the streets.
Die Sonne warmte die StraBen.
The streets were warmed by the sun.
Die StraBen wurden von der Sonne erwarmt.
9a.The telephone call confused the young woman.
Der Telephonanruf verwirrte die junge Frau.
b.The young woman was confused by the telephone call.
Die junge Frau wurde von dem Telephonanruf verwirrt.
10a.The police car forced the truck driver off the road.
Das Polizeiauto drangte den Lastwagenfahrer von der StraBe.
b.The truck driver was forced off the road by the police car.

55

Der Lastwagenfahrer wurde von dem Polizeiauto von der

StraBe gedrangt.
11a.A policeman found the crying child.

Ein Polizist fand das schreiende Kind.

b.The crying child was found by a policeman.

Das schreinde Kind wurde von einem Polizisten gefunden.
12a.The bright light blinded the tennis player.

Das grelle Licht blendete den Tennisspieler.

b.The tennis player was blinded by the bright light.

Der Tennisspieler wurde von dem grellen Licht geblendet.
13a.A helicopter pursued the fleeing bankrobber.

Ein Hubschrauber verfolgte den fliehenden Bankrauber.

b.The fleeing bankrobber was pursued by a helicopter.

Der fliehende Bankrauber wurde von einem Hubschrauber

verfolgt.
14a.The author discovered some old manuscripts.

Der Autor entdeckte einige alte Manuskripte.

b.Some old manuscripts were discovered by the author.

Einige alte Manuskripte wurden von dem Autoren entdeckt.
15a.The janitor cleans the floors daily.

Der Hausmeister reinigt die Boden taglich.

b.The floors are cleaned by the janitor daily.

Die deen werden taglich von dem Hausmeister gereinigt.
16a.A dog found the frightened child.

Ein Hund fand das verangstigte Kind.

b.The frightened child was found by a dog.

Das verangstigte Kind wurde von einem Hund gefunden.

56

I .I. E . IIE'I

1. Lightning striking church

2. Lightning hitting golfer

3. Firehydrant squirting firefighter
4. Flyswater killing fly

5. Tornado hitting barn

6. Missile hitting plane

7. Truck towing car

8. Ball hitting boy

9. Train hitting bus

10. Wave hitting boat

11. Ambulance hitting policeman
12. Torpedo hitting ship

13. Tank running over soldier.

14. Train about to run over woman
15. Avalanche burying skiers

16. Crane wrecking building

APPENDIX C

Questionnaire

57

APPENDIX C

Q I' .

1. What is your age?
Mean= 39 years
Range=18-66

2. At what age did you learn English?
Mean=-12.0 years
Flange= 4-35

3. Did you learn English in a formal (school, books) or informal
( by living in the country) setting?

lnformally- 12.5%

Both- 87.5%

4a. What language is spoken in your home?
English=- 41.6%
German:- 14.6%
Both languages: 29.1%

4b. What is the language (if any) you are most comfortable speaking?
English=14.6%
German: 29.1%
Equally comfortable= 56.2%

5. On a scale from 1 (bad) to 7(very good) how would you rate your
a. English skills Mean=6.5
b. German skills Mean=6.8

6. Do you frequently talk to other people who know both German and English and
when you do, do you frequently mix languages?

no mixing= 17.4%

frequent mixing= 82.5%

7. When you read, do you read in English or in German? Approximately how
many hours per day do you read:

English Mean= 1.5 hours

German Mean= 0.6 hours

58

8. Do you know other languages, besides German and English? How many
years did you study them ( or were you exposed to them)?

84.9% one or more languages besides German and English

15.1% none
All of the additional languages were learned in school. Three subjects indicated
that they were able to converse near fluency in another language besides
German and English.

LIST OF REFERENCES

LIST OF REFERENCES

Adjémian, C. (1983). The transferabIlIty of IBchal properties. In: S. Gass& L.

Selmker (Eds) Winn 250- -267)
Bowley, Massachusetts: Newbury House Publishers, Inc.

Bain, B. &Yu,A. (1980). Cognitive consequences of raIsmg children
bilingually: 'One parent, one language'. QanadlaugumaLQt
W 304-313

Bates, E. & MacWhinney, B. (1982). FunctIonalIst approach to grammar. In E.
Wanner& L. R. Gleitman (Eds.,)
(pp. 173-218). Cambridge. Cambridge University Press.

Bates, E., McNew, 8., MacWhinney, B., Delescovi, A., & Smith, S. (1982).
Functional constraints on sentence processing: A cross-linguistic study.

W 1.1.245-299-

Berk-Seligson, S. (1986). Linguistic constraints on intrasentential
code-switching: A study of Spanish/Hebrew bilingualsim. ngggaggjn

59mm 15, 313-348.
Bloomfield, L. (1933). Language. New York: Holt,Rinehart and Winston.

Bock, J. K. (1986). Syntactic persistence in language production. Qggnitjm
Esmhglmx. 1.8. 355-387-

59

60

Bock, J. K. (1987). Coordinating words and syntax in speech plans. In A. Ellis
(Ed)ELQnLQ§§JnJD§.D§¥QDQIQQLQflanQanfi(VOI 3 pp 337- ”390)

London: Erlbaum.

Book, J. K. (in press). Closed-class immanence in sentence production.

g 'l'

Bock, J. K. & Loebell, H. (1988). W. [manuscript submitted for
publication].

Caramazza, A. & Brones, I. (1979). Lexical access in bilinguals. W
W13. 212- 214

Chan, M, Chan, H. L H. &Hoosain, R. (1983). Input output swnch In bilingual
code switching JoumaloLEsxonoflmuistiomoamla 407-416

Collins Gem (1985). W. London: William Collins Sons 8.
Co.Ltd.

Coppieters, R. (1987). Competence differences between native and
near-native speakers. mnguggg, 63,, 544-573.

Corder, S. P. (1967). The significance of learner's errors. IRAL, 4, 161- 170.
Citedin: Selinker, L (1983) Language Transfer. In: S. Gass & L. Selinker

(EdS) LanguaoeJLansioLinJanouaoeJoamino Rowley. Massachusettsz

Newbury House Publishers, Inc.

Cummins, J. (1977). Cognitive factors associated with the attainment of

intermediate levels of bilingual skills. W m
No.1;2. 3-12

Dalrymple-Alford, E. C. (1985). Language switching during bilingual reading.
BritioMoumaloLEsxonoloox. 7.6. 1 1-122.

61

Dell, G. S. (1986). Aspreading actIvatIon theory of retneval in sentence
production We: 283- 321

Dulay, H. & Burt, M. (1974). Errors and strategies in child second language
acquisition. W 8,129-136.

Dunbar, K. & MacLeod, C. M. (1984). A horse race ofadIfferent color Stroop

interference patterns with tranformed words. JgumngLExpgfimgmal
W10. 622- 639.

Felix. S. W. (1982). WWW
Tiningen: Gunter Narr Verlag TI’Jbingen.

Garrett, M. F. (1975). The analysis of sentence production. In G. H. Bower (Ed)
InopsxonomLoUeammoanommixatiomVom pp 133- -177) New

York. Academic Press.

Garrett, M. F. (1980). Levels of processing in sentence production. In B.

Butterworth (Ed.),_Languagg_Emgugtigm (Vol.1 , pp.177-220).

Orlando/London: Academic Press.

Garrett, M. F. (1982). Production of speech ObservatIons from normal and

pathological language use. In A. Ellis (Ed. ) WM
WM). 19-76). Orlando/London: Academic Press.

Gass, S. (1983). Language transfer and universal grammatical relations. In.

S Gass& L Selinker (Eds) WW
(pp. 69-85). Rowley, Massachusetts. Newbury House Publishers, Inc.

Gass. s. & Selinker. L.(EdS-)(1983)- LamoootransioflnJanouaooJoamino
Rowley, Massachusetts: Newbury House Publishers, Inc.

Grosjean, F. (1982). W. Cambridge, Massachusetts:

Harvard University Press.

62

Hakuta. K- (1986) MILLoLoLIanouaoolnooooatoonolinouatism. New York:

Basic Books, Inc.

Harrington, M. (1987). Processing transfer: Language-specific processing
strategies as a source of interlanguage variation. Applies:

Esxonoflnouistmo 351-377

Hornby. Peter A (Ed ) (i 977) W
W. Academic Press, New York.

James, C. (1980). W. Essex: Longman Group Limited.

Jordens, P. (1986). Production rules in interlanguage: Evidence from case
errors In LZGerman. In: E. Kellerman & M. Sharwood Smith (Eds.,)

£2tossIInoIIIstIo.InfluoIIcom.soconoJrztnooaooaoouIsmon(no-91 110)

New York. Pergamon Press.

Keilerman. E. & Sharwood Smith, M.( 1986). QLQ§§|ingHi§1ig mflnem in
W. New York: Pergamon Press.

Klein, W. (1986). Wm Cambridge: Cambridge

University Press.

Kolers, P. A. (1963). lnterlingual word associations. Jgumglgtigmal
W; 291 -300

Kolers, P. A. (1966). InterIIngual faCIIItatIon of short-term memory. Jgumglgt
MoroaLLeaminojndNotbaLBonayioLi 314-319

Kolers, P. A. &Gonzalez, E. (1980). Memory for words synonyms, and

translations WWW
Momont. 6. 53-65

Kufner. H. L- (1962) W

Chicago: The University of Chicago Press.

63

Lado, Ft. (1957).L1ngm§_tlg§_agLQ§§_QunuLgs. Ann Arbor: University of

Michigan Press.

Landry, R. G. (1974). A comparison of 2nd-language Ieamers and
monolinguals on divergent thinking tasks at the elementary school level.

W53, 10-15.

Lederberg,A. Ft. 8. Morales, C. (1985). Code switching by bilinguals:

Evidence against a third grammar ioumaLoLEsILohoIingoistoaosoamh.
13, 113-137.

Levelt, W. J. M. & Kelter, S.(1982). Surface form and memory in question
answering. DoonIMEsxonoloox. 14. 78-106.

Mack, M. (1986). A study of semantic and syntactic processing in

monolinguals and fluent bilinguals. MW
15, 463-484.

MacKay, D. G. & Bowman, R. W. (1969). On producing the meaning in
sentences. AmonoanJoumanLEsxohoIooLoz. 23-39.

MacKay, D. G. (1982). The problems of Flexibility, fluency, and

speed- accuracy trade-off In skilled behavior. flaminglggigaLngigm 82,
483- 506.

Macnamara, J. (1967). The Bilingual's Linguistic Performance- A
Psychological Overview. MW 23(2), 58-77.

Macnamara, J., Krauthammer, M, & Bolgar, M. (1968). Language switching in
bilinguals as a function of stimulus and response uncertainty. Jgumalgt

ExpeLImontaLEsxonoIooLZQ 208- 215.

Macnamara, J., & Kushnir, S. L. (1971). The IIngu1stIc Independence of

bilinguals The input switch .IounaLofleLoaNoarmnganom
oenamnlo. 480- 487.

64

MacWhinney, B. (1987). Applying the competition model to bilingualism.
Aooliodfisxoholinouistios.8. 315-327.

McCormack, P. D. (1977). Bilingual linguistic memory: The
independence-interdependence Issue revisited. In: P..A Hornby (Ed.,)
Bilingualism. New York: Academic Press.

Morton, J. &Chambers, S. M. (1973). Selective attentIon to words and
colours W25. 387- 37

Obler, L. K. &Albert, M. I. (1978). A monItor system for bIIIngual language
processing. In: Paradis, M. (Ed), AspagtsgLBilingualismmp. 165-161).

Columbia, S. C.: Hombeam Press.

Paivio, A. & Lambert, W. E. (1981). Dual coding and bilingual memory.
JournaLofJLomaLLoaminoanolomaLBebaxioLzosaz-sse.

Paivio, A, Clark, J. & Lambert, W. E. (1988). Bilingual Dual- -Coding Theory

and Semantic Repetition Effects on Recall. ,lgumalgLExpanmamal
EsxonolooxLLeamImMomomnofiognlmJA. 163-172.

Palef, S. R. &Olson, D. R. (1975). Spatial and verbal rivalryinaStroop-like
task CanadIanJoumanLEsxcnolmfl. 201 -209

Paradis, M. (1980). The language switch In bilinguals. Psycholinguistic and

Neurolinguistic perspectives. In P. Nelde (Ed)l.anguagas_in_cgn1agt_ang
minim. Wiesbaden: Franz SteinerVerlag

Pearl, E, 8 Lambert, W. E. (1962). The relation of bIlIngualIsm to IntellIgence
EsvonolooioaLMonootaonszo. 1-23.

Penfield, W., & Roberts, L. (1959). Bpaagnangmmmagtianisms. Princeton,

N.J.: Princeton University Press.

65

Pfaff, C. (1979). Constraints on language mixing: lntrasentential
code-switching and borrowing In Spanish/English. Language, 55,
291 -318.

Poplack, S. (1980). Sometimes I'll start a sentence in Spanish y termino en
Espafiol: Toward a typology of code-switching. linguistics 18, 581-618.

Poplack, S. (1981). Syntactic structure and social function of codeswitching.

In R DuranlEd)Latino_Languagoano_Qon1munIoatlxo_Boh.aI/_ior
(pp..169-185) Norwood, N.J.: ABLEX Publishing Corporation.

Posner, M. I. (1978). WWW Hillsdale, N.J.:
Erlbaum.

Potter, M. C, So, K, von Eckhard, B., & Feldman, L. B. (1984). Lexical and
conceptual representation in beginning and proficient bilinguals.

.loumaLofJLocoaLLoaminoandNoroaLBonavionza. 23- 38

Reyes, Rogelio (1974). Blugiasjngnigangjganisn. Harvard dissertation.

Rutherford, W. E. (1983) Language typology and language transfer In: S.

Gass & L Selinker (Eds ) LanguaoeJLansieLInlanguaoeJeamino
(pp. 358-371). Rowley, Massachusetts: Newbury House Publishers, Inc.

Schwanenflugel, J. & Rey, M. (1986). lnterlingual semantic facilitation:
evidence for a common representational system in the bilingual lexicon.

MW 25. 605-61 8.

Sharwood Smith, M. (1983). On first language loss In the second language
acquirer: Problems of transfer In: S. Gass & L Selinker (Eds. )languaga

mansjaLimanguagalaamingmp. 222- -.232) Rowley, Massachusetts:

Newbury House Publishers, Inc.

Sheen, R. (1980). The importance of negative transfer in the speech of near
bilinguals. lntemational Review of Applied Linguitics. Cited in: Grosjean,

F. (1982) Lifa_iiiii1n_‘[iiiio__l_anguagas. Cambridge, Massachusetts: Harvard

66

University Press.
Skinner, B.F. (1957). Win:- Englewood Cliffs: Prentice Hall.

Sridhar, S. N. & Sridhar, K. K. (1980). The syntax and psycholinguistics of
bilingual code mixing QanadianioumaLoLEsxonoloox. 31(4) 407-416

Stemberger, J. P. (1985). An interactive model of language production. In A.

Ellis (Ed-I EmoressmnonsxonoloouflanouagflVol. 1. 00143-186)
London: Erlbaum.

TImm, L. A. (1983). Does code-swflchmg take tIme?l1isganic_,loumaLgt
BonaxiouLaLSoioncosfi. 401 -416

Van Vlerken, M. (1980). Adverbial placement in English: A study of first
language loss. Master's dissertation, English department, Utrecht
University. Cited In: Sharwood Smith, M. (1983) On first language loss in
the second language aquirer: Problems of transfer. In: S. Gass, 8 L

Selinker (Eds ) WWW Rowley,

Massachusetts: Newbury House Publishers, Inc.

Watkins, M. J, & Peynircioglu, 2. F. (1983). On the nature of word recall:

Evidence for linguisitc specificity. WWW
W22 385- 394

Weiner, E. J, & Labov, W. (1983). Constraints on the agentless passive.
W19. 29-58.

Weinreich, V. (1963). Languagasjnfionlagt. The Hague: Mouton.

Weisgerber, L. (1966). Vorteile und Gefahren der Zweisprachigkeit.
W W 73- 89.

Wode, H. (1977). On the systematicity of L1 transfer In L2acquisition. Citedin:
S W FeliX(1982)Esxonollnouistiscne_Asoektodos

67

W. Tiibingen: Gunter Narr Verlag TI'ZIbingen.

Woolford, 5- (1933)- Bilingual code-switching and syntactic theory. Lingiiisflg
lnouimwal. 520-535.

”‘ifliiililllllllllllllllllllllllll/ll