This article addresses methodological concerns in research on grammatical aspects of code-switching. Data from code-switching have the potential for a unique contribution to linguistics by giving us access to combinations of linguistic features that may be difficult (or impossible) to observe in monolingual data. Nonetheless, the use of code-switching data for linguistic inquiry is not without issues. In this paper, we focus on three methodological questions specific to code-switching research: (i) project design, (ii) experimental procedure and (iii) participant selection. Drawing on experimental data from both published works and in-progress projects, we highlight potential solutions to each methodological challenge, concluding that several solutions are often required to mitigate the impact of confounding variables. In line with previous work (e.g. Grosjean 1998, Gullberg, Indefrey & Muysken 2009), we suggest that researchers clearly report on their methodology. Our overall goal is to contribute to a dialogue on best practices in code-switching research.

The aims of the present study were twofold. The study addressed the issues of simultaneous attention to form and meaning in second language (L2) written input and reactivity of think-alouds. Specifically, the study examined the comprehension of L2 learners of Spanish who either attended to lexical or grammatical forms while reading for meaning or read for meaning alone. Learners completed these tasks while either thinking aloud or not. Results indicated only a minimal effect for thinking aloud that did not appear to compromise the internal validity of the study. Additionally, results showed that attending to grammatical or lexical form while reading for meaning did not affect comprehension. Indeed, learners who processed these forms more deeply evidenced greater comprehension. These findings are considered in light of methodological issues and the larger issue of simultaneous attention to form and meaning in a L2.

Many languages, including English, exhibit a restriction on subject extraction over complementizers called the that-trace effect. Although extensively studied, this phenomenon remains a puzzle. Not all languages exhibit the effect; Spanish does not. Spanish also allows postverbal subjects, while English does not, which has been linked to the that-trace effect. Because Spanish/English differ in these properties, combining lexical items from both languages in a single derivation, as in code-switching, offers additional insight into the nature of the restriction. Two acceptability judgment tasks of Spanish/English code-switching reveal that a single Spanish functional head is insufficient to license either postverbal subjects or subject extraction. Instead, we argue, the that-trace effect and related properties arise from the interaction of two heads.

Among methodological concerns specific to code-switching (CS) research is the design of the target stimuli used in experiments with an acceptability judgment task. We argue here that research which makes use of CS data of this type must also incorporate monolingual stimuli into the experimental design, specifically monolingual stimuli judged by the same bilingual participants who judge the code-switched stimuli. We do so by reviewing two sets of experimental CS data we collected and exploring the role that monolingual stimuli can play in the analysis of that data. In each experiment, an analysis based solely on acceptability judgments of the CS stimuli leads to one interpretation, while incorporation of results from the monolingual stimuli leads to a distinct interpretation. We show that it is the interpretation integrating the monolingual acceptability judgments which is more valid, thereby arguing for the value of monolingual stimuli in design and analysis.

Various methodological concerns are specific to code-switching research; however, the modality of experimental stimuli has yet to be thoroughly investigated. This study explicitly tests if the mode of presentation does in fact affect participants’ judgments in Spanish-English code-switching using two different syntactic phenomena: (i) pronouns and lexical DPs, and (ii) wh-movement. The results are parallel, but not identical for the two modalities. We found no difference on a global level, indicating that written code-switched stimuli do not produce depressed ratings. We found a few individual differences when looking at specific structures within the two phenomena. In those cases, the aural condition enhanced the ratings of more acceptable sentences. Crucially, these differences did not affect the interpretation of the results.

Some languages have a fixed subject position, while others are more flexible. Languages like English require pre-verbal subjects; languages like Spanish allow subjects in postverbal position. Because this difference clusters with several linguistic properties distinguishing the two languages, subjects in Spanish and English have been a perennial issue in linguistic theory, touching central problems like the EPP, the nature of cross-linguistic variation, and the relationship between core functional heads. Our project contributes a novel source of evidence to these debates: Spanish/English code-switching. Code-switching, the use of two languages in one utterance, combines the languages’ lexical items and their attendant syntactic features in a single derivation. Because code-switching, like all natural language, is rule-governed, researchers can exploit judgments about the well-formedness of code-switched sentences to draw conclusions about the combinations of features they represent. We report on a formal judgment experiment testing subject position in Spanish/English code-switching as a function of the presence of two functional heads known (from monolingual evidence) to affect subject placement: the C(omplementizer) and T(ense) heads. By manipulating which head appears in which language, we test the availability of post-verbal subjects under different feature combinations. Our results show that post-verbal subjects are only available when both C and T are in Spanish; neither Spanish head alone is sufficient. This finding suggests that the features regulating subject position stem from neither head alone, which is problematic for traditional approaches to the EPP as a feature of T but in line with other recent research on null subjects.