Revisiting the Concept of "Triggering of Code-Switching
Open Access
- Author:
- Trawick, Sonya
- Graduate Program:
- Spanish
- Degree:
- Doctor of Philosophy
- Document Type:
- Dissertation
- Date of Defense:
- May 22, 2022
- Committee Members:
- Paola Migliaccio-Dussias, Major Field Member
John Lipski, Major Field Member
Rena Cacoullos, Chair & Dissertation Advisor
Janet van Hell, Outside Unit & Field Member
Paola Migliaccio-Dussias, Program Head/Chair - Keywords:
- bilingualism
linguistics
code-switching
trigger
sociolinguistics
quantitative
corpus - Abstract:
- The present dissertation revisits an established concept initially introduced to explain how and why bilinguals might code-switch, going back and forth between languages. The Triggering Hypothesis (first introduced in Clyne, 1967) claims that when a bilingual encounters a word with ambiguous or dual language affiliation, such as a cognate, the resulting raised level of activation of the language not being spoken will cause an increased likelihood of following speech being produced in that other language. Though initial work was confined to anecdotal analyses lacking accountable quantitative methodologies (e.g., Clyne, 1967, 1977, 1980, 1987, 2003), important contributions have since been made by Broersma and colleagues (Broersma, 2009; Broersma & De Bot, 2006; Broersma et al., 2009, 2020), reporting support for the increased occurrences of code-switches around cognates and offering findings such as that more trigger words further increase the likelihood of code-switches in a clause. Nonetheless, majors gaps in the inquiry into triggering remain, and the present dissertation aims to provide methods for closing the gaps, as well as the results of familiar and novel means of testing the triggering hypothesis. Using the New Mexico Spanish-English Bilingual corpus (Torres Cacoullos & Travis, 2018: Chapters 2 & 3), the present dissertation provides the first study of triggering in a corpus representing a unified speech community. Taking advantage of its prosodic transcription (Du Bois et al., 1993), it addresses methods for defining unambiguous multi-word code-switches as candidates for the triggering hypothesis, and then dedicates ample space to a discussion of identifying and categorizing candidate trigger words, namely cognates. Within this latter discussion is a review of cognates in previous studies, and how interpreting the results of cognate studies requires careful consideration of phonological overlap, assigning translation equivalents, dialectal variants, and frequency, among other variables, as well as justification for including lone other-language items (LOLI) as trigger words and not automatically conflating single-word insertions with multi-word code-switches. A series of analyses inspecting distributions and regression models analyzes new variables and replicates work from Broersma et al. (2020). The present dissertation does not find especially strong evidence for triggering, though some results illuminate trends such as that trigger words with higher phonological overlap with their translations and LOLI status are more likely to co-occur with code-switches than low phonological overlap and Non-LOLI trigger words. After stopping to note that code-switches in surrounding contexts co-occur with code-switches much more so than candidate trigger words, the dissertation provides an initial exploration of the application of a burstiness analysis to the frequency of code-switching, demonstrating the complexity inherent to attempting to model code-switching. Though not finding persuasive evidence of triggering, the present dissertation contributes valuable discussions of how triggering has been examined thus far, as well as what kinds of conclusions can be made from different treatments of bilingual data. Focusing on an understanding of the northern New Mexico bilingual community’s speech patterns with or without the consideration of candidate trigger words, it makes a case that in general co-occurrences of code-switches and candidate trigger words may be better understood through correlation, by way of high resting levels of co-activation, rather than causation, as is proposed by the Triggering Hypothesis. This dissertation concludes by underscoring the value of establishing replicable units of measurement and applying these to well-defined code-switching data against which to better understand the potential effects of triggering and other proposed processes, a prerequisite for developing code-switching models.