1.1 Experimental approach
Over the last ten years, my research has been defined by two principal qualities. First, it is part of the phonetics-phonology interface. This interface refers to phonetically-guided research in phonology. In general terms, it means that abstract frameworks used to describe human speech must be grounded in fine-grained, minute phonetic detail in languages (e.g. my research in svarabhakti vowels). Until the late 1990s, there was quite often a separation between those who did phonetics (i.e. more concrete) and those who did phonology (i.e. more abstract). My work is part of a movement that synthesizes both approaches.
Second, with the exception of my work with a historical basis, my research is defined by its usage of laboratory phonology. This approach differs from past research that often cited data from previous studies. In this approach, the researcher analyzes data from participants whose speech was recorded in a controlled environment. Data are then laboriously transcribed with phonetic symbols and then analyzed. In my research, I collect data in the field and analyze it using up-to-date methods and software; for example, I conduct all spectrographic and waveform analysis in Speech Analyzer 2.6, all audio file editing in Sound Forge, and all statistical analysis in the software most commonly-used in my field, a program simply called ‘R’.
1.2 My approach
My research interests center on the properties of sounds within a dynamic language system. Languages generally contain around forty sounds used in human speech. Studies in the last fifteen years have completely changed our view of how we process these sounds in oral communication. They have illustrated with empirical data that sounds in a word are not a collection of independent, static units, like pearls on a necklace. Rather, they are interdependent, dynamic units, called ‘gestures’ in Articulatory Phonology (henceforth, AP) (Browman & Goldstein 1992; Gafos et al. in press). In this approach, a gesture is a dynamically defined articulatory movement that produces a constriction in the vocal tract. For this discussion, a gesture is roughly used to represent the movements required to produce a given sound in human language. In AP, the vowel is the underlying gesture in a syllable and consonants (i.e. constrictions) are ‘placed’ onto the vocalic gesture.
1.3 Research motivations
My work is motivated by the premise that human speech has a very intricate timing relationship between adjacent consonants, called ‘consonant clusters’. Consonant clusters represent an intriguing challenge to the linguist in that each language allows for different consonant cluster combinations, both within and across a syllable. My research seeks to answer three fundamental questions. First, what is the governing force behind these changes in timing relationships? Second, how do changes in timing relationships between sounds (i.e. gestures), particularly adjacent sounds, alter a particular language? Third, at a theoretical level, what implications do these changes have for a gestural-based approach? In what follows, I discuss how I answered these questions in further detail.
2. The Specifics of My Research Program
2.1 Consonant cluster timing: Svarabhakti vowels in Spanish
The bulk of my research has focused on a vowel-like element, called an intrusive or svarabhakti vowel (henceforth, SV) that occurs between two consonants. Crucially, the SV is a measure of the amount of overlap between two consonants; longer SV duration entails less overlap between consonants. In the case of Spanish, SVs occur in consonant clusters comprised of a consonant and specifically an ‘r’, as in ‘pronto’ [p«R] 'soon' in the same syllable (i.e. tautosyllabic); the ordering is reversed across a syllable boundary (i.e. heterosyllabic).
In previous research, the presence of this vowel was simply noted, but its duration and what factors affect duration were not. It was seen as a ‘stepping stone’ to get from one consonant to the next. My research is novel in that it tests SV duration and suggests factors that affect its duration in both contexts mentioned above. I note that my work particularly on consonant clusters across a syllable boundary was the first known study to consider this context in Spanish; there was almost nothing known about either the acoustic properties of the SV in this environment or the possible effects of the following consonant.
In very general terms, my research suggests that in the case of same-syllable consonant clusters, what caused greater distance between the two consonants (i.e. longer SV duration) was the phonetic quality of the first consonant. My findings in the case of consonant cluster across a syllable boundary suggested that SVs were statistically longer (i.e. much less overlap of adjacent consonant) across a word boundary. Finally, in 2009 my research on consonant clusters in Spanish addressed the limitations of a formal-theoretic representation of Spanish consonant clusters and remedied the problem by offering new categories needed to incorporate recent findings on the issue.
2.2 Consonant cluster timing: Historical change in Galego-Portuguese
When compared to its other sister Romance languages, one feature almost exclusive to Galego-Portuguese is the change from /Cl/ to /CR/ diachronically, as in Latin eclesia to Galego-Portuguese igreja ‘church’. Though the phenomenon is well-understood in the field, the reasons behind rhotacization (i.e. an ‘l’ becomes an ‘r’) are not. Previous studies did not adequately account for the data. I offered an articulatory account of its occurrence over time to emphasize restrictions set by syllabic timing. I then applied the analysis to word type. I show in the study why rhotacization does not occur in one class of words (i.e. patrimonial or learned transmissions), but does indeed occur in another class (i.e. semi-learned transmissions).
2.3 Consonant cluster timing: Pali
It should be noted that I also viewed gestural timing in consonant clusters in a non-Romance language, namely Pali, a Middle Indo-Aryan language. A salient characteristic of Pali is consonant cluster simplification, resulting in a cluster with the same consonant repeated (i.e. consonant gemination). There is one exception to this rule, namely when one of the two consonants is a nasal. My research offers articulatory evidence to explain why the timing relationship of the consonants is altered when one consonant is a nasal. In a subsequent publication on Pali, I look at how consonant gemination is affected across a morphological boundary. The study sheds light on the relationship between phonology and morphology and observes that previous accounts of monomorphemic words do not explain consonant gemination across a morphological boundary.
2.4 Linguistics work outside of consonant cluster timing
In 2010 and 2011, my research turned away from consonant clusters and focused on a lexical-semantic issue. In diachronic terms, Spanish has often created two words (i.e. doublets) from one Latin word to create nuance in meaning. It has generally avoided semantic ambiguity, instead preferring to create another word to capture the difference in meaning. That said, one variety of Spanish, US Spanish, is now allowing one word to carry multiple meanings. My research illustrates first the diachronic avoidance of semantic ambiguity and then illustrates how it is now being employed in this variety. It greatly adds to the field in that it illustrates a phenomenon that Spanish had previously gone to great lengths to avoid.
3. Current and Future Directions
I currently have one manuscript under review. My work in same-syllable consonant clusters in Spanish has led me to propose a continuum along which we may now view SV duration. Using an extensive corpus of twenty-nine subjects, the data suggest speakers alter timing relationships of consonant clusters based on a combination of two articulatory factors (i.e. manner and place of articulation). Though studies have suggested certain factors that affect timing changes of the cluster, this is the first study to suggest it is best viewed along a continuum.
Future research will focus on four succinct areas for which there is a current gap. First, the relationship between syllable timing and the prosodic (π-) gestural model, set forth by Byrd & Saltzman (2003); Byrd et al. (2005); and Byrd et al. (2006). My previous research in Spanish has addressed the model’s effectiveness in describing word-internal vs. across-a-word boundary SV duration. However, we currently have no known studies for Spanish that address higher prosodic tiers, such as the prosodic word, an utterance, etc., a fact that makes this topic ripe for my future research.
Second, phonological vowel reduction is rarely evidenced in Spanish. Dialects that are commonly-cited as evidencing this phenomenon are Mexican Spanish, especially around Mexico City, and Andean Spanish. While conducting the study on consonant clusters in Spanish, I collected data from Guatemalan Spanish and observed vowel reduction. Preliminary analysis suggests that phonological vowel reduction is occurring in this variety as well, especially in word and utterance final positions. I will analyze the data and measure vowel duration in unstressed syllables to determine which vowels undergo reduction. In addition, I will clarify in which environments, in the middle of the word or word-final, that reduction more commonly occurs.
Three, I will extend my analysis of consonant gemination in Pali to certain Spanish dialects such as Andalusian and Caribbean Spanish that evidence a similar phenomenon. I am particularly curious to see if the same articulatory and prosodic conditions are creating the same phenomenon in a different language. Finally, I plan to extend my diachronic analysis of Galego-Portuguese to the only other Romance language, namely Sicilian, in which a similar (but slightly different) phenomenon occurs.
In short, my work is defined by its breadth of linguistics research. I have conducted studies on consonant clusters in two Romance languages and one Middle Indo-Aryan language. I have published work that analyzed language formation in historical terms, along with work that focused on current phenomena. I have also published work solely based on a formal-theoretic approach and other work solely on acoustic data.
My work in Spanish, Galego-Portuguese and Pali supports the recently-made claims that many forms of sound variation in languages are motivated by changes in the timing relationship sounds have with one another. My research has greatly contributed to the field in (at least) two ways. First, at the micro level, I have offered analyses with empirical data that detail the factors that alter the timing relationship of adjacent consonants. Second, at the macro level, I have explained how these changes affect the processing of human speech. Additionally, I have expanded my work beyond consonant clusters to current work in semantics (i.e. word meaning) in U.S. Spanish.