Gender and Phonetics in the Spanish/English Code-Switched Determiner Phrase

The current work is divided into two parts, both pertaining to the mixed Determiner Phrase (DP). The first half looks at the gender in the Spanish D + English N DP (e.g., la table). The second half focuses on the phonetics in the English D + Spanish N DP (e.g., the mesa). I delve into both parts in turn. In the first half, I propose a new way to account for the variability in gender in DPs like la table; that is, the quality of childhood input—what I call childhood input nouns—has a potential effect on the gender in mixed DPs. Some researchers have argued that bilingual speakers of Spanish and English have the following strategies when choosing gender in DPs like la table: the analogical criterion, the masculine default, and the phonological shape of the word (Burkholder, 2018; Liceras, Fuertes, Perales, Pérez- Tattam, & Spradlin, 2008; Otheguy & Lapidus, 2003; Valdés Kroff, 2016). While I do not completely abandon these tried and true strategies, I propose that heritage speakers of Spanish in Chicago will prefer the feminine determiner with nouns that have both, and not just one, of the following two characteristics: (i) the Spanish equivalent of the English noun is feminine, and (ii) the noun is what I call a childhood-input noun. As far as I am aware, this quality of childhood input has not been a variable that has been studied before. In order to test this new variable, 21 heritage speakers of Spanish in Chicago performed a Two-alternative Forced Choice task. They were given English sentences where the only switch was the Spanish determiner. Furthermore, the English nouns were: (i) feminine in Spanish and childhood-input nouns; (ii) feminine in Spanish non-childhood-input nouns; (iii) masculine in Spanish; and (iv) nonwords. The participants, then, had to choose what gender—whether masculine or feminine—they preferred with each stimulus noun. A frequentist analysis was performed. The results suggest that these participants significantly preferred the feminine Spanish determiner with the English nouns that were feminine in Spanish and were childhood-input nouns when compared to the other conditions, where the masculine was preferred. Thus, this experiment provides evidence that the childhood-input characteristic does indeed have an effect on the gender that is chosen in these mixed DPs. More importantly, these results suggest that there is evidence that the childhood-input characteristic of a word has an effect on the morphosyntactic flexibility of nouns whereby they can be used with feminine and masculine gender, as per López’s account. The second part of the dissertation focuses on the phonetics of DPs like the mesa. There has been a debate as to whether this DP is an acceptable switch (Moro, 2014, among others). Moro, for example, states that there must be feature checking between the noun and the determiner. In the case of the mesa, the noun has gender but the determiner does not. There is no feature checking, which leads to a crash in the derivation. However, this DP has been found, though infrequently, in natural speech corpora, as I demonstrate in this work. López (2020) proposes the following novel account: this DP is infrequent in production because it has a phonetic constraint that is motivated by the presence of a nominalizing n head with a gender feature. For a balanced bilingual population, this DP is an acceptable switch, but the entire DP must be pronounced with Spanish phonetics. López also considers “common currency” nouns, which are nouns that originated in Spanish but have become part of the English lexicon. According to López, these nouns have achieved full morphosyntactic flexibility whereby they can be pronounced with Spanish or English phonetics. In order to test López’s novel account, I analyzed two code-switching corpora—the Bangor Miami Corpus and the Multilingual Phonology Lab corpus—as well as testing heritage speakers of Spanish in Chicago using a Director-Matcher task. In the corpora I searched for the mixed DP of interest (e.g., the mesa). I also looked for monolingual Spanish DPs (e.g., la mesa) and monolingual English DPs (e.g., the table). I acoustically analyzed the vowel sound in the determiners and compared them to each other. According to López (2020), the mixed DP should be pronounced with Spanish phonetics, therefore similar to the Spanish monolingual DP. Through frequentist analysis, the normalized data showed that there was a significant difference between the mixed DP and the Spanish and English monolingual DPs. So, while the corpora data provided evidence for López’s account, there were some limitations. The number of tokens was small, and the acoustic quality of the recordings were not suitable for fine acoustic analysis. Therefore, experimental data was needed. I designed an experiment where 23 heritage speakers of Spanish in Chicago participated. They had to perform a modified Director-Matcher task. In this task, the participants produced the mixed DP (e.g., the mesa), monolingual Spanish DPs (e.g., la mesa), monolingual English DPs (i.e., the table), and complex DPs (i.e., the green mesa). Results from a Bayesian analysis showed that there was a significant difference between the pronunciation of the vowel sound in the determiner in the mixed DP and the English monolingual control, as well as the Spanish monolingual control. When it came to the adjective condition (e.g., the green mesa), it was significantly similar to the English monolingual control, but it was significantly different than the Spanish monolingual control. The “common currency” nouns were also analyzed. Qualitatively, the results supported López’s theory. Quantitatively, there was a low number of tokens, which could have affected the validity of the results from a Bayesian Analysis. The results provide partial support for López (2020) in that the participants did not pronounce the DP like their English, but it was also different than their Spanish. Further acoustic analysis of the nouns would be necessary to make a stronger claim. Results from both parts of this work provide evidence that López (2020) is the right framework to gain insight into the architecture of the bilingual mind.