‘Doing’ Romance Linguistics

studied the effects of code-switching at the level of phonetic implementation for Spanish-English bilinguals as they produced blocked monolingual and code-switched sentences as in (7). We measured voice onset time (VOT) for /p t k/ when speakers were producing in monolingual mode and in three sites (pre-switch, switch, and post-switch) when they were producing in bilingual mode. All participants showed a significant effect for language, with Spanish VOTs lower than English VOTs, as well as a significant effect for site, with the segment at the switch site presenting the greatest converged values. The findings show that bilingual maintain separate phonetic inventories and that their VOTs are adjusted in code-switching, demonstrating patterns of divergence, convergence, interference


Introduction
My path in 'doing'Romance linguistics has been productive, though not always uncomplicated. My early endeavors were firmly grounded in my training in formal linguistics, and I was primarily concerned with syntax, specifically generative syntactic theory, which dominated the field. At that time, doing Romance linguistics meant upholding or refining universal principles and theoretical constructs, and my investigations extended beyond Romance to Germanic, Asian, and Semitic languages. A number of these works considered the abstract features that license various types of nominals in Spanish and English, including overt and unexpressed (pro and PRO) subjects and [±wh] phrases, accounting for null-subject and verb-second properties, among others (see Toribio 1990Toribio , 1992Toribio , 1993. In subsequent studies, I investigated the mechanisms that underlie derivations that feature nominative-marked objects and dative-marked subjects in Spanish, Icelandic, German, and Korean (see Harbert & Toribio 1991and Toribio 1988. And still others of my publications examined non-canonical case marking and honorific morphology in Sinhala and Japanese, proposing each as instances of agreement relations (see Toribio 1989 andGair 1991). My ensuing analyses in the areas of code-switching, second language morphosyntax, and native language attrition brought additional observations to bear on the linguistic knowledge of early and late bilinguals. Introspections on grammatical well-formedness for French-Arabic, Spanish-English, and Spanish-German codeswitching motivated proposals on the functional architecture of phrases and on the cartography of the left periphery of the clause (see Belazi, Rubin & Toribio 1996, Rubin & Toribio 1995, 1996, Toribio & González-Vilbazo 2014, and data from French-English and Spanish-English bilinguals of varying levels of competence unveiled information regarding the components of the syntax that could be most stable or impermeable in incipient attrition (see Zapata, Sánchez &Bullock &. While these efforts towards characterizing defined structures and advancing abstracted, cognitive representations were important and often influential, I harbored uneasiness over the field's rejection of the 'non-sanctioned' linguistic forms that I knew to circulate in vernacular speech, and I felt unsettled as I reflected on why certain speech was chosen to be modeled, while other language behaviors were designated as infelicitous, target-deviant, spurious, or simply wrong. I began to confer less importance to the theoretically stipulated, 'well-formed' utterances of idealized speakers and language 'experts' and greater attention to the patterns of usage attested among diverse groups of speakers and contexts. In more recent years, then, my 'doing' Romance linguistics has also meant attending to vernaculars, individuals, and communities and their ecologies. This work has been especially rewarding, as it speaks to my background as an Afro-Hispanic native speaker of a racialized dialect of Spanish and in my experiences as a 'generation 1.5' immigrant whose repertoire comprise vernacular and 'standard' varieties of Spanish, English, and Spanglish; it also speaks to that of my students (many of them incipient educators and scholars), who hail from marginalized communities and whose colloquial language practices are stigmatized and undervalued by academic institutions. In these pursuits are identified two discernible threads of inquiry, the first centered on Dominican Spanish and its speakers and the other on oral and written code-switching; each is discussed in turn below. These undertakings are situated within the broader discipline of linguistics, to include sociolinguistics and sociology of language, but the approach is necessarily interdisciplinary, informed by insights and practices of the social and applied sciences.

Dominican Spanish
One line of research in which I have been invested examines Dominican Spanish and its speakers, and especially the speech of residents of northern rural border areas between the Dominican Republic and Haiti, where language and cultural contact, isolation from urban centers, and absence of normative pressures have yielded linguistic properties that attest to the uniqueness of the dialect. This research draws on naturalistic methods (e.g., interviews and narrative story telling) and controlled elicitation (e.g., repetition of words and sentences) administered to hundreds of individuals to disclose the incidence and diffusion of the segmental and prosodic properties that identify regiolects and sociolects of Dominican Spanish (see and Toribio & Clemons 2020. My studies corroborate the existence of three regional dialects distinguished by the phonological processes that affect realization of coda liquids, as in (1). Of these processes, the lambdacism of the capital region (Santo Domingo) is accorded the most social prestige, and the liquid gliding of the Cibao region the least; in fact, the latter production is the most salient to the Dominican ear, and it is often over-corrected by those Cibaenos who produce it. As another example of variation, coda /s/ elision (2a) is the norm for Dominican speech in general, but it is greatest among lower-class and male speakers; indeed, coda [s] realization among men is considered effete and/or effeminate. Also attested is non-etymological [s] production, which is heard in natural speech (2b) and can be elicited in controlled tasks (2c); in usage, [s] insertion is interpreted as reflecting a speaker's ignorance. Nevertheless, a series of studies coauthored with Barbara Bullock demonstrate that Dominicans are not 'lost coda-s speakers' as has been suggested in the literature and that their realization of etymological /s/ and insertion of the non-etymological sibilant are governed by social and linguistic variables (see Bullock & Toribio 2010, 2014, 2015and Bullock, Toribio & Amengual 2014 In the realm of syntax, my scholarship examines numerous properties of Dominican Spanish that are altogether absent or not systematically reproduced in other varieties of the language (see Toribio 1993, Toribio & Bullock 2015. The extracts that follow immediately below and throughout this section (drawn from speakers of diverse social profiles) showcase the unexpected expression of overt referential pronouns and a pleonastic ello, the unanticipated placement of subjects in infinitivals and interrogatives, a noteworthy focalizing ser, and emphatic negation and affirmation constructions. While many of these had been remarked on by Henriquez Urena in the first half of the 20th century, their perseverance in the Dominican vernacular had garnered little scholarly attention.
(3) a. ... Ello hay personas que lo aprenden bien, pero hay otros que no porque cuando uno mezcla un idioma con el de uno se le hace dificil realmente para uno pronunciar las cosas. 'There are people who learn it [English] well but there are others who don't because when you mix a language with your own it becomes truly difficult for you to pronounce things.' b. Bueno yo me quedé, digo, "Mire, ¿usted esta loco es? ¿Que usted es? ¿Usted es un brujo?" 'Well I stood, I said, "Look, are you crazy? What are you? Are you a witch?"' Taken together, such morphosyntactic properties may indicate a departure from what is generally presumed for Spanish grammar with respect to core and interface syntax. It is conceivable that the Dominican variety presents a different grammar (e.g., one that is non-pro-drop and non-V2, etc.) and that the occurrence of forms that align with what is prescribed for Spanish is a consequence of education and self-monitoring.
To be sure, the above-referenced phonological and morphosyntactic properties hold potential for correcting received knowledge and providing new data for formal linguistics and sociolinguistics. Notwithstanding its potential import, however, Dominican Spanish is held in low esteem in the Dominican Republic and in the larger Spanish-speaking world. In interviews and attitude surveys that I conducted in the Dominican Republic, speakers consistently express deep linguistic insecurity and especially adverse assessments of certain local lects. Children and adults from across social strata in the Dominican Republic and in the diaspora disparage their own Spanish as deficient; some individuals reason that their speech is too far removed from its European origins, while others lament that Dominican speech is tainted by the African substratum of Haitian Creole, most noticeable along the Dominican-Haitian border (4). The downgrading of border Spanish is further verified via a matched-guise task, in which listeners negatively evaluated the verbal guises of speakers from the border, whom they assume to be uneducated, poor, and black, based on 'fronterizo' prosodic properties. These attitudes are also reflected in behavior, as children decline to learn the neighboring Haitian Creole, which they deride as gibberish (5) (see Bullock & Toribio 2008 'I don't want to learn Haitian Creole because we don't need to know that language, because here there is an abundance of Haitians but rarely do you need them and they also speak Spanish. [...] It doesn't sound good to me at all.' Importantly, these racialized ideologies (which are founded in the country's history of nation-building) persist when Dominican immigrants are integrated into the black/white dichotomy of the United States. In New York, where Dominicans have largely settled, Dominicans remain loyal to their Spanish, invoking the language in performing a non-Black identity (6a, 6b), and they express a preference for 'standard' European American English, eschewing the African-American English that would 'rerace' them as Black based on appearance in this diasporic context (6c) (see , 2003, Jensen et al. 2006).
b. She starts to talk Spanish, and I'm like, "Aren't you black?" And she's like, "No, I'm Dominican. I'm 100% Dominican." I was like, "Oh, snap." I just kept quiet. c. A mi me gusta oir los anglosajones. ... La consejera me preguntó, "¿Ella es afroamericana?" Le dieron un training para hablar y mejoró bastante, y ya no la confunden [por afroamericana]. 'I like the way Anglos speak ... The counselor asked me, "Is she African-American?" They gave her some speech training and she improved a lot, and now they don't confuse her [for African American].' As to be expected, the attitudes and ideologies that dictate Dominicans' language usage on the island and in the U.S. are implicated in language contact and change, as well as in language learning and shift. But there is evidence that transmission of the heritage language may be eroding and that notions of racial identity are being re-defined as Dominican-Americans build alliances with peers/communities of color. My on-going work re-examines the linguistic and social allegiances of second-and third-generation Afro-Dominicans in New York towards uncovering the linguistic variables that might index ethno-racial affiliations in the absence of recourse to the heritage language. Is there an identifiable monolingual Dominican English in New York, akin to other established ethnolects such as Chicano English? If so, what stable linguistic features uniquely characterize this lect and distinguish it from other local varieties, e.g., second-language or bilingual Spanish and African American English.

Bilingual code-switching
A parallel line of research to which I am committed is focused in the area of language contact phenomena, most especially code-switching, calquing, and convergence. While much of my previous work in this area examined isolated, selected, and/or constructed exemplars (see Bullock & Toribio, eds., 2009, in more recent examples of 'doing' Romance, I have strived to contribute findings based on rigorous methodologies and larger data sets in order to discern patterns (and define the possibilities and limits) of contact outcomes. As one example, Barbara Bullock and I studied the effects of code-switching at the level of phonetic implementation for Spanish-English bilinguals as they produced blocked monolingual and code-switched sentences as in (7). We measured voice onset time (VOT) for /p t k/ when speakers were producing in monolingual mode and in three sites (pre-switch, switch, and postswitch) when they were producing in bilingual mode. All participants showed a significant effect for language, with Spanish VOTs lower than English VOTs, as well as a significant effect for site, with the segment at the switch site presenting the greatest converged values. The findings show that bilingual maintain separate phonetic inventories and that their VOTs are adjusted in code-switching, demonstrating patterns of divergence, convergence, interference, and hypercorrection, as determined by individual factors (e.g., proficiency, language practices, etc.). Notably, the most proficient bilinguals produced the most converged segments. I also collaborated with Barbara Bullock in creating corpora and tools for the analysis of data in which multiple languages are overtly represented. One initiative, the Spanish in Texas Corpus Project (spanishintexas.org), profiles Spanish as it is spoken throughout the state and provides open learning resources for researchers, educators, and the general public to promote appreciation of the richness of local varieties of Spanish (see Toribio & Bullock 2013. The corpus contains nearly a hundred interviews, collected by Latino students in their home communities. The interviews are fully transcribed and tagged for part-of-speech, and nominal and verbal morphology; thus, they are easily mined for the study of contact phenomena. As one example, the Texas corpus registers the use of the light verb agarrar (literally 'to grasp') calqued on English 'get' (8), and our comparison with corpora from Mexico confirm its extension in the U.S. context (see Bullock, Serigos & Toribio 2020). The Texas corpus also presents ample variation, e.g., in the incorporation of English verbs, which occurs via morphological integration (English root+ar) and via codeswitching with a light verb hacer 'to make/do' (hacer+English V), as in (9). The metadata that accompanies the speaker sub-corpora allows for further analysis of the social variables (e.g., generation, education, occupation, residence) that are often associated with contact outcomes. Surprisingly (or not), the corpus comprises only 3% English language sequences (a figure that includes English proper names, song and movie titles, and talk-turns that are entirely in English), but there are identifiable networks of speakers for whom borrowing, calquing, and code-switching co-occur; such is the case for the speakers below, who represent three generations from El Paso (see Bullock &. (8) a. Para no dar mayores gastos, o sea, para que no gastaran tanto, porque la colegiatura y los libros pues eran bastante caros, entonces, yo me esforcé para agarrar becas. 'So as not to incur greater expenses, that is, so that they [parents] wouldn't spend so much, because the tuition and books were very expensive, I strived to get scholarships'. b. Allí habia three houses, four houses que mi papá había hecho. Pero mi papá murió en nineteen thirty-two. Mi mamá murió cuando yo tenía fourteen months old. Y me cuidó mi abuelita y mi abuelito desde esa edad, de siete a los catorce años. Luego, como no quiso agarrar ayuda del gobierno, nos llevaron a Juárez. 'There were three houses, four houses that my father had made. But my father died in 1932. My mother died when I was 14 months old. And my grandma and grandpa took care of me from that age, from 7 to 14 years. Then, since he didn't want to get help from the government, they took us to Juarez'.
(9) a. [A]lguien siempre tenía que estar ahí pushándome, para poder avanzar, ¿verdad? so, decidí agarrar una carrera técnica que es más, más, como dos años para agarrar tu Associate's, so, estaba como ... toda mi vida he estado mechanically inclined. 'Someone always has to be there pushing me to progress, right?, so I decided to get a technical career that is more, like two years to get your Associates, so I was always like all my life I have been mechanically inclined.' b. I guess todos fueron buenos maestros. Del que me recuerdo más sería en high school, Mr. Rodriguez, porque él sí me hacía push, mucho a que compitiera, que hiciera extra, extracurricular activities y todo eso. 'I guess they were all good teachers. The one I remember most would be in high school, Mr. Rodriguez, because he did push me, a lot to compete, to do extracurricular activities and all that'.
My most recent efforts in the analysis of language contact have been strengthened by the recruitment of tools and technologies from corpus and computational linguistics and by collaborations with students from STEM fields. Again, with Barbara Bullock, I convened the Bilingual Annotation Tasks (BATs) research group 1 a cohort of bilingual students from sciences and engineering with shared interests in language mixing, and together we embarked on developing natural language processing tools for working with multilingual corpora, which are increasingly available in open corpora and in digital and print media (see Bullock, et al. 2020). It cannot go unremarked that what we thought to be a seemingly simple task language identificationis in fact quite challenging: Is sale tagged as a Spanish verb or English noun? Is a tagged as a Spanish preposition or English determiner? We created an algorithm that allowed for language tagging of numerous and diverse textse.g., the Spanish in Texas corpus, epistolary texts (cf., Killer Crónicas, excerpted in (10)), and bilingual film transcripts (e.g., Bon Cop Bad Cop)-without human intervention. These language tags generate sequences of language spans and are the 1 The acronym is significant for our students as Austin is home to the largest urban bat colony in North America; the Mexican free-tailed bats migrate to roost under the bridge that crosses over the lake at the center of the city. basis of metrics designed to quantify mixing: They gauge the proportion of languages, the probability of mixing, and the regularity of mixing, and they return scores that allow for crosscorpora comparisons (see Guzmán et al. 2017). The language spans also allow for visualization and for easy comparison across corpora. The Figures 1-3 represent three corpora, of comparable length (~7,000 tokens each), which depict different mixing signatures: The oral conversational corpus in Fig. 1 and Fig. 2 show Spanish and English as the matrix language, respectively, while Fig. 3 profiles a written corpus (Killer Crónicas) with more equal distribution of Spanish and English, greater integration of languages in horter language spans, and a consistency of switching across this written text.
(10) Estoy sentada en el comedor de nuestro departamento, taipeando en el laptop de Pierre which somehow, de milagro, did NOT get ruined by the sheets of water que inundaron nuestro departamento while we were in Chile. 'I am seated in the dining room of our apartment, typing on Pierre's laptop which somehow, by miracle, did not get ruined by the sheets of water that inundated our apartment while we were in Chile.' Adding Part-of-Speech tags enabled efficient testing of hypotheses regarding cooccurrence constraints on language mixing. For instance, our studies of language transition sites substantiate the regularity of switching at clause boundaries (identified by punctuation tags) and reaffirm the injunction against switching at the Pro/V and the Aux/V junctures across all corpora. These analyses further reveal a consistent asymmetry for Det/N switches, which nearly categorically include a Spanish Determiner + English Noun, irrespective of the matrix language of the corpus (see Bullock & Toribio, forthcoming, Bullock, Guzmán & Toribio 2019). We also examined a corpus of 'Spanglish translations' of renown texts (El Little Príncipe, Don Quijote of La Mancha, Hamlet); this corpus presents a ratio of languages and a probability of mixing that mirror those of Killer Crónicas, but the code-switching flaunts the norms (transitional probabilities) established for all other corporawritten and oralwith frequent switching at Pro/V, Aux/V, Det/N and within words and idiom chunks (11). The latter corpus thus reflects ludic mixing, and the authors' claim of presenting the "language of a new generation" is unfounded. (11) Mi little amigo estallo laugheando otra vez: 'But adónde would él irse?' 'Anywhere. Él va a seguir su nariz.' Entonces el little principe remarco solemnemente: 'It wouldn't importar. Mi lugar es tinico.' Y agregó wistfulmente: 'Si sigues tu nariz, tú puedes ir muy far.' 'My little friend exploded laughing again. 'But where would he go?' 'Anywhere. He has to follow his nose.' Then the little prince remarked solemnly: 'It wouldn't matter. My place is tiny'. And he added wistfully: 'If you follow your nose, you will go very far.' Our algorithms and metrics facilitated the effective processing of larger bilingual datasets than I had worked with previously. In another fruitful collaboration, I examined the social stratification of English-language spans in a 3.3-million-word corpus of Puerto Rican Spanish-language press (see Bullock, Larsen Serigos & Toribio 2016). Moving forward, I'd like to draw on our tools and techniques to examine the incursion of English in the Mexican(-American) Spanish of contiguous and noncontiguous locales of Mexico and the US, as registered in social media, illustrated in the Tweets in (12). The research intends to scrutinize Tweets geo-located in urban centers (Mexico City; Los Angeles) and binational sister cities (Ciudad Juarez/El Paso; Mexicali/San Diego) and user-generated web-content (Youtube videos, blogs) originating in Mexico and the U.S., towards documenting the linguistic contours of Mexican Spanish lects that are influenced by the global and local encroachment of English. Finally, I hope to collaborate with Bullock and our research team on the task of developing techniques for detecting covert switching, i.e., structural convergence, in the Spanish and English texts produced by bilingual authors. Robust findings generated only by analyses of largescale data, will inform theories of language contact and language change and aid in putting to rest misinformation surrounding varieties such as U.S. Spanish, Spanglish, and Chicano English, among others.
(12) a. No sé si llorar porque estoy feeling o porque estos niños no saben escribir. 'I don't know whether to cry because I'm feeling or because these kids don't know how to write.' b. I'm starting to talk Spanglish like I'll start talking American después mi mexicana quiere salir. '... and then my Mexican wants to come out.

Summary
To summarize, my 'doing' Romance Linguistics speaks in different ways to the interrelation of language and (re)presentation, bringing perspectives from structural linguistics, sociolinguistics, and the sociology of language to the analysis of the understudied language(s) of minoritized populations; it also speaks to the need for diverse tools, methodologies, and interpretive frameworks for a fuller understanding of local and global language varieties. The overview attests to the benefits of collaboration and of engaging individuals and communities in the production of knowledge (and in the redressing of received knowledge) about their speech practices. The research discussed in the foregoing paragraphs has highlighted the rule-governed nature of Spanish vernaculars and the logics of language variation and language choice among Spanish speakers in the Dominican Republic and in the United States. Through such efforts, my 'doing' Romance not only responds to long-standing questions of academic interest but also promotes appreciation of Spanishes and, in so doing, aids in dispelling myths that promote linguistic insecurity and linguistic prejudice.