The Berkeley Phonetics & Phonology Forum ("Phorum") is a weekly talk and discussion series featuring presentations on all aspects of phonology and phonetics.
Come to our first Phorum of the year with a 5 minute (or less) Ph-related musing, question, or update on what you did this summer.
Grammars with weighted constraints predict the existence of ganging effects: cases where two constraints combine to overcome the effect of one competing constraint. This talk presents a case study of one such ganging effect in French, using it to argue for an analysis in MaxEnt Harmonic Grammar. The basic pattern is reported in Charette (1991), who shows that epenthesis occurs if and only if two conditions are met: (1) the epenthesis site is followed by exactly one syllable; (2) the epenthesis site is after a cluster. Although Charette reports a categorical pattern, for many speakers, the pattern is subject to variation. Using experimental data, I show that epenthesis is more likely when it meets one condition, and most likely when it meets both. This can be straightforwardly modeled in a theory of variation with weighted constraints, such as MaxEnt. The MaxEnt analysis also predicts that the constraints conditioning epenthesis should play a role whenever there is phonological variation. This is borne out in French: both constraints show independent effects in schwa deletion and phonotactics.
According to Selkirk’s (2011) “match theory”, the mapping of syntactic structure onto prosodic domains is universal. What this means is that if a language chooses to implement the relation between syntactic- or phrase-structure in the phonology, certain syntax-phonology relations should be predictable (and others not possible). This potentially produces asymmetries, as in Luganda, where a verb forms a tone phrase with what follows (e.g. an object, adjunct, right-dislocation), but not with what precedes (e.g. the subject, adverbial, left-dislocation). The purpose of my talk is to raise the question whether the phrasal tonology of Lusoga, the most closely related language to Luganda, is syntactically grounded—or is free to apply without respect to syntax. I begin by briefly outlining the situation in Luganda, and then turn to Lusoga. Although extremely closely related, the two languages are quite different in their implementation (Luganda) vs. non-implementation (Lusoga) of prosodic domains. While I provide an analysis that accounts for this difference, and which respects Selkirk’s claims, I also show that there is one head-dependent syntax-specific condition in both languages that does not so easily fall into line. I conclude with discussion of the typology of phonological phrasing in Bantu.
Since the early 20th century, it has been proposed that phoneme inventories should change in ways that preserve those phoneme contrasts that do more 'work' to convey meaning (e.g., Gillieron 1918, Trubetzkoy 1939, Hockett 1967, Silverman 2010). In this talk, I’ll review recent evidence that supports this idea. When a phoneme contrast merges, words that were originally distinguished by that contrast also merge in pronunciation (like caught ~ cot in English). We have used a database of diverse languages to show that phoneme mergers are much less likely to occur when the phoneme pair distingushes many words (Wedel et al. 2013). Instead, we show that when a phoneme pair distinguishes many words, the pronunciation of those phonemes is more likely to change in ways that do not result in merger. These alternative kinds of changes result in alterations to the phoneme inventory, but maintain existing distinctions in the lexicon. These results suggest that speakers of languages preferentially preserve phonetic distinctions when these distinctions do more work to communicate word-distinctions in their language. To support this, in the second talk of the day I will show phonetic evidence that English speakers hyperarticulate sounds when those sounds distinguish their host-word from another word. More informally, all these results suggest that when linguists have an intuition that a particular sound contrast in their language is preserved because it plays an important role in communication – they may be right. I will end by providing some thoughts about ways we can extend this kind of research to a broader range of languages, and ask for ideas.
This paper develops a new integrated analysis of the phonological and syntactic proper-ties of nonconcatenative morphology in (Classical/Modern Standard) Arabic. The account centers around an algorithm for sub-word linearization at the syntax-phonology interface, here termed the “Mirror Alignment Principle” (MAP). The MAP determines the ranking of Alignment constraints (McCarthy & Prince 1993) in the phonological component based on asymmetric c-command relations in the syntax. Using the MAP, we can predict the exact position of all morphemes/segments in an Arabic verbal form based on their syntactic functions and structures without any recourse to templates (cf. McCarthy 1979, 1981).
I illustrate how this syntax-phonology mapping generates idiosyncratic behavior of the Reflexive morpheme, and can explain the morphophonological distinctions between the two different types of Causatives found in the language. Additionally, I will demonstrate that this framework allows for a more complete and internally consistent phonological account of the verbal system than previous approaches.
Tutrugbu, a Kwa language of Ghana, shows a pattern of cross-directional vowel harmony that is amenable to either a phonological or morphosyntactic analysis of progressive harmony in the language. Tutrugbu roots trigger leftward, regressive assimilation of prefixes for the feature [ATR] while initial (subject) prefixes trigger rightward, progressive assimilation for the feature [RD]. Moreover, Tutrugbu appears to exhibit a variation of “sour grapes” vowel harmony (Padgett 1995; Wilson 2003), where leftward ATR harmony either assimilates all vowels within its domain of application or none. This is predicted by some OT constraints (e.g. AGREE and ALIGN), and has been regarded as pathological since no languages have been shown to instantiate this pattern. If vowel harmony is phonologically-driven in Tutrugbu, then at least one instance of “sour grapes” harmony is attested, justifying the predictions associated with constraints like AGREE and ALIGN in this regard. Furthermore, the progressive labial harmony pattern strongly resembles other known labial harmony patterns found in Kaun’s (1995) typology. That being said, a number of factors suggest the realistic possibility that vowel harmony in Tutrugbu may not be phonological in nature, but rather morphosyntactic subject-verb agreement. If vowel harmony is morphosyntactic, apparent phonological blocking of agreement, as well as numerous parallels between syntax and phonology in Tutrugbu, as well as in neighboring Tafi (Bobuafor 2013) and Logba (Dorvlo 2008), receive an explanation (see Baker & Willie 2010 for a similar analysis). In this presentation, phonological and morphosyntactic analyses of the data are discussed, focusing on the kind of relationship phonology and syntax must have in order to account for the Tutrugbu data.
This talk will be an interim report on my book in progress, covering three data-based chapters on the first words of children learning six languages (US and UK English, Estonian, Finnish, French, Italian and Welsh), their adult target forms, the use of prosodic structures at a later stage (end of the single-word period) and some examples of templates (children learning American English, French and Welsh). Discussion will address the similarity of the early forms, the challenge of consonant variegation and evidence for ‘whole-word phonology’ and for memory as a key constraint in early word learning.
Recent years have seen a shift in the study of language contact to the coupled use of large typological datasets and computational techniques, in an effort to detect large-scale patterns and avoid subjective judgments of areality (O’Connor and Muysken, 2014; Wichmann and Good, 2014). In this talk, we demonstrate the utility of principal component analysis (PCA) for determining the major dimensions of typological diversity in the phonologies of a region, by applying this analytical technique to a database of South American phonological inventories. We identify areal and genetic patterns in the distribution of phonological segments across the continent, and show that this method is an informative, quantitatively rigorous, and non-subjective way of exploring large-scale typological patterns in a geographic region.