Grammatical convergence, the achievement of structural similarity among languages, is widely considered to be an inevitable consequence of bilingualism, and code-switching, the alternation of different languages in a single speech event, has been assumed to be a key mechanism in such convergence. This study investigates the role of code-switching in promoting grammatical convergence in the long-standing New Mexican bilingual community, where there is regular code-switching between Spanish and English. This research targets patterns of first person singular subject (yo 'I') expression, and tests the hypothesis of grammatical convergence via code-switching through comparing the structure of variable first singular subject expression in (1) monolingual varieties of Spanish and English, (2) bilinguals of varying levels of bilingualism, and (3) the presence versus the absence of code-switching. Given observed effects of monolingual and bilingual structural priming, or repetition of the same grammatical form, this research is expected to demonstrate that cross-linguistic priming plays a role in ostensible convergence by modestly increasing the rate of a parallel structure in the presence of code-switching without involving change in language-particular grammatical patterns.
This research advances the understanding of the language varieties spoken by bilinguals. In recording the speech of New Mexican bilinguals, it will contribute to the documentation of the oldest surviving Spanish dialect on the continent. The materials that arise from this research can be incorporated into the development of pedagogical resources for heritage language teaching at various levels throughout the United States. Finally, while labels such as "Spanglish" may imply that switching between languages involves haphazard language mixing and leads to grammatical deterioration, by showing the systematic structure of bilingual discourse, this research promotes greater public awareness of the realities of US Spanish and helps to counter the stigmatization of bilingual language varieties.
Gauging convergence on the ground: Code-switching and priming in the community Is grammatical convergence between bilinguals’ two languages inevitable and does code-switching inherently promote it? This project advances the study of bilingualism through the constitution of a community-based corpus of bilingual speech in which the role of code-switching in convergence is evaluated through a novel on-line measure: comparisons based on the proximity of speakers’ spontaneous use of multiword sequences of the other language. Bilingual community members, spontaneous speech and prosodically-based transcription The contact site for the study is the northern part of the U.S. southwestern state of New Mexico, where Spanish and English have co-existed as the main competing languages for over 150 years (Bills & Vigil 2008). The longstanding nature of this contact situation allows for a compelling assessment of hypothesized convergence. Participants in the study are all at least third-generation Nuevomexicanos ‘New Mexicans’. The 23 women and 18 men, born between 1923 and 1989, were selected to cover a range of demographic backgrounds (miners, ranchers, schoolteachers, and others), allowing extra-linguistic constraints on the variability to be assessed. Crucially, all meet the criterion of regular use of both languages with the same interlocutor in the same domain. Because bilingual behavior is known to be sensitive to the communicative situation, spontaneous speech data is recorded through the community-based method of sociolinguistic interviews (Labov 1984) conducted by in-group members (Poplack 1993). To enable accountable analysis, the recordings are comprehensively transcribed. Prosodically-based transcription, where each line of transcript (aligned with the audio recording) corresponds to an Intonation Unit, permits an objective assessment of the smoothness and syntactic units of code-switching in conversation (where the notion of the sentence is unserviceable). These data constitute the New Mexico Spanish-English Bilingual (NMSEB) corpus (Torres Cacoullos & Travis, In Preparation), which comprises a total of 29 hours of speech, approximately 340,000 words. NMSEB is unique in its roughly even distribution of speech produced in Spanish and English, by the same speakers in the same conversations. In addition to single-word other-language insertions, NMSEB records copious switching between multi-word strings of Spanish and English, seen in (1), testimony to the success of our data collection techniques. (1) Miguel: ... y en inglés, ‘... and in English, when I'm speaking, when I'm speaking, ... hay veces que quiero poner ... there are times when I want to put una .. Spanish word in there. a .. Spanish word in there.’ [NMSEB 04 Piedras y gallinas, 1:11:03-1:11:24] Tests for convergence: comparisons of structured variability NMSEB has fostered tests of the role of code-switching in convergence for a range of linguistic variables (e.g., International Journal of Bilingualism, special issue edited by Travis and Torres Cacoullos, 2015). Subject pronoun expression is one of the most trumpeted loci of Spanish convergence with English, on the strength of reported higher rates. Here we consider not only rates (which oscillate according to considerations extraneous to the grammar such as genre), but the direction of effect of constraints on variable use. Thus, we assess inter-linguistic grammatical similarity via intra-linguistic variability, comparing contextual factors contributing to the selection of a given variant across each of the languages in contact (Figure 1). Parallel favoring effects of factors operationalizing putative shared constraints or functions provide evidence for grammatical similarity. For first singular (1sg), while we find shared constraints, such as subject continuity, which can be construed as a cross-linguistic tendency, a language-specific constraint is operative in English such that, outside of coordinated contexts, unexpressed subjects only occur in Intonation-Unit initial position. Since this prosodic constraint is absent in monolingual Spanish it constitutes a "conflict site" (Poplack & Meechan 1998) which supplies a hard test for convergence. New Mexican Spanish shows parallel favoring effects with monolingual Spanish and not with English on this site of structural conflict, ruling out convergence. What of the much-surmised role of code-switching? We introduce a synchronic test based on comparisons of rates and constraints in the presence vs. absence of maximally proximate code-switching, defined as the use of multi-word English sequences in the same or preceding clause (indicated with underlining in (1)). Systematic comparisons show no effect for code-switching on neither the rate nor the constraints of Spanish subject pronoun expression. However, we find both language-internal and cross-language coreferential subject priming, such that unexpressed subjects tend to be followed by unexpressed subjects, while preceding pronouns—Spanish yo and English I—favor a subsequent yo. This suggests that what others have interpreted as intrinsic code-switching effects for Spanish subject expression may in fact be the result of particular priming effects, and, specifically, the occurrence of fewer unexpressed primes under proximate code-switching (Figure 2). These results refute convergence via code-switching, and give rise to the "Contextual Distribution via code-switching" hypothesis as a new direction in the study of code-switching.
