Hybrid model: investigating bilingual language production through code-switching
View/ Open
Date
02/07/2018Author
Selles, Anthony
Metadata
Abstract
Bilingual language production is an area of psycholinguistic research that has
received recent attention. Experimental evidence from bilingual word production
tasks has shown that both languages share representation at the mental lexicon,
meaning that concepts will lead to the activation of the target lemma from both
languages. Investigations into how bilinguals organise two grammatical systems has
largely come from cross-linguistic syntactic priming. Syntactic priming is a
phenomenon in which speakers are likely to repeat a syntactic structure in which they
have recently experienced: cross-linguistic syntactic priming is when a speaker uses
a syntactic structure in one language because they have recently experienced that
structure from the other language. Together, the study of the bilingual lexicon and
syntactic representations have led to the development of models of bilingual
language production.
A more recent experimental paradigm is the forced code-switching task in
which participants are required to code-switch in some experimental trials. The
forced code-switching task is the experimental method used in this thesis. This thesis
aims to use this experimental task to test my proposed model of bilingual language
production, the Hybrid model. The Hybrid model proposes an architecture of the
bilingual lemma stratum that differs from previous models of bilingual language
production. The Hybrid model assumes that lexical items from one language can be
produced using the syntactic structure of the other language.
In this thesis I report seven experiments testing the proposed lemma stratum
of the Hybrid model. Experiment 3.1 investigated the production of prenominal
adjectives of English and postnominal adjectives of Spanish during code-switching
between Spanish and English to see whether speakers would use the lexical items
from one language with the word order of the other language. The results showed
that speakers almost exclusively used the word order dictated by the language in
which they produced the lexical items. This did not support the proposed lemma
stratum of the Hybrid model.
Experiments 4.1 and 4.2 investigated gender agreement of possessive
pronouns during code-switching between Spanish and English to see if the
possessive pronoun from one language could be produced using the gender
agreement rules from the other language. The results showed that English-Spanish
and Spanish-English bilinguals sometimes produced possessive pronouns in one
language with the gender agreement rules from the other language. It was
demonstrated that this effect was not due to a misunderstanding of the gender
agreement rules of the participants’ second language. These results support the
proposed lemma stratum of the Hybrid model.
Experiments 5.1 and 5.2 investigated gender agreement of possessive
determiners during code-switching between French and English to see if the
possessive determiner from one language could be produced using the gender
agreement rules from the other language. The results showed that English-French
and French-English bilinguals sometimes produced possessive determiners in one
language with the gender agreement rules from the other language. It was
demonstrated that this effect was not due to a misunderstanding of the gender
agreement rules of the participants’ second language. These results support the
proposed lemma stratum of the Hybrid model.
Experiments 6.1 and 6.2 investigated the production of determiners during
code-switching between German and English. Of specific interest was whether
English determiners would be produced more often than German determiners
because German determiners hold case information whereas English determiners do
not. In Experiment 6.1 participants were forced to code-switch before an accusative
NP. The results showed that English determiners were sometimes produced within
the German NPs, but German determiners were not used within the English NPs. In
Experiment 6.2 participants were forced to code-switch before a dative NP. The
results showed that participants almost exclusively produced the determiner in the
same language as the target noun. Analysing the frequencies of the determiner used
within the experimental session, the different pattern of results between Experiments
6.1 and 6.2 may be a result of a competition for selection between determiner forms.
To conclude the thesis I discuss the implications of these findings, what they
mean for the Hybrid model, and directions for future research.