Phorum 2017

August 28, 2017 - James Kirby (Edinburgh)

"Microprosody and tonogenesis"

Although the microprosodic basis of tonogenesis is well established (e.g. Matisoff, 1973; Hombert et al., 1979), the mechanics of exactly how the transfer of pitch occurs are not well understood. The classical phonologization model (e.g. Hyman, 1976) proposes that this process involves the exaggeration of low pitch from a voiced consonant to the following vowel: in other words, a textbook example of a phonetically gradual, lexically abrupt Neogrammarian sound change. In this talk, I will discuss a recent investigation into the production of onset-induced F0 perturbations (Kirby and Ladd, 2016) which I think suggests an alternative explanation: namely, that this process is more likely to be phonetically abrupt. Time permitting, I will also discuss the implications of our study for models of laryngeal phonology more generally.

September 7, 2017 - Special Phorum 

**Thursday 12:30 - 2pm, 1229 Dwinelle**

Eric Wilbanks & Sharon Inkelas (UC Berkeley) - "Directionality effects via distance-based penalty scaling"

Directionality effects are a known challenge for surface constraint-based models of phonology. We describe a novel approach to modeling such effects in constraint-based grammars through the use of distance-based penalty scaling. Under our proposal, faithfulness and markedness constraint violations are scaled according to the distance of an offending segment from a specified edge. Distance is operationalized as an abstract scale indexing the distance of each segment in a word from the designated edge. (Distance can be measured in segments, moras, or syllables.) The values on this scale generate multipliers for constraint violations incurred by a segment in a given position. For a left-tropic distance-scaled constraint, a violation by an initial segment thus incurs a lower penalty than one by an identical segment in noninitial position. We apply our model to several cases of directional long-distance assimilation and compare our approach with alternative models of such effects.

Nicholas Rolle (UC Berkeley) - "Transparadigmatic Output-Output Correspondence"

This paper proposes Transparadigmatic Output-Output Correspondence (TrOOC), building on grammatical paradigm correspondence and uniformity in Burzio (1998) and Pariente (2002). Under TrOOC, a root with an inflectional affix acts as the base and is in correspondence with that same root plus derivational morphology in the same inflectional context, i.e. a Corr-OO relation x between Base [root1-infl1]x <-> Output Candidate [ [root1-deriv]-infl1]x. When this correspondence holds, if an Ident constraint is highly ranked it enforces identity between the two forms along some specified phonological dimension. I show that unlike Paradigmatic Output-Output Correspondence which enforces correspondence among cells within a single inflectional paradigm (i.e. [root1-infl1]x <-> [root1-infl2]x ), TrOOC enforces correspondence across counterpart cells in equivalent but separate paradigms.

TrOOC is used to account for unexpected similarity in word accent between morphologically related words in the Bolivian language Ese’eja. Within Ese’eja verbs without derivational morphology, inflectional morphemes idiosyncratically assign accent to the initial, penult, or ultima of the unmodified stem, subject to morphological accent competition and resolution, and footing constraints. In contrast, when the stem contains derivational morphology the position of accent exceptionally remains faithful to the position in inflected verbs without derivational morphology, and cannot be derived strictly through inflectional accent assignment mechanisms. I propose a constraint Ident-TrOO(melody) which enforces identical accent melodies between forms standing in a Transparadigmatic correspondence, i.e. if they have the exact same root and inflection. I discuss the implications of this for OO Correspondence theory generally, and demonstrate the inadequacy of strict IO cyclicity (e.g. Stratal OT) to account for these data.

September 11, 2017 - Martha Schwarz1, Morgan Sonderegger2, Heather Goad2 (1UC Berkeley, 2McGill)

"Representing a four-way contrast: Nepali, voiced aspirates, and laryngeal realism"

Theories of laryngeal realism (Honeybone 2005, Iverson and Salmons 1995, Beckman et al. 2011, 2013) argue for a tight correspondence between a segment’s phonetic cues and the (laryngeal) phonological features that represent it. Consequently, the ‘p’/‘b’ contrast in French, expressed phonetically by vocal fold vibration during the stop closure, is represented by a [voice] feature whereas the ‘p’/‘b’ contrast in English, expressed phonetically by long and short lag VOT, is represented by a [spread] feature. Tests used to diagnose a language as employing either [voice] or [spread] (Beckman et al. 2011, 2013) predict potential conflicts for segments specified for multiple laryngeal cues, i.e. voicing and aspiration.

In order to test these conflicts, we apply the same diagnostics to Nepali’s four-way laryngeal contrast between voiceless, voiceless aspirated, voiced, and voiced aspirated stops, for which Iverson and Salmons (1995) propose the following representation:

Voiceless: [ ]
Voiceless aspirated: [spread]
Voiced: [voice]
Voiced aspirated: [voice], [spread]

We find that the diagnostic tests do support this representation, with the caveat that the voicing contrast is “stronger” than the aspiration contrast. We propose that the strength of [voice] over [spread] is best represented by temporally ordering the [voice] and [spread] features within a segment, and suggest a Q theory representation (Inkelas and Shih 2017) to do so.

September 25, 2017 - Jordan Ackerman1, Noah Hermalin2, Sharon Inkelas2, Darya Kavitskaya2, Stephanie Shih1 (1UC Merced, 2UC Berkeley)

"Pokémonastics: A cross-linguistic comparison of sound-meaning correspondences"

Sound symbolism (SSym), which is prevalent cross-linguistically, challenges the core assumption of the arbitrariness of the sign and raises the question: to what extent is SSym universal, vs. shaped by language-specific structure (Dingemanse et al. 2015)? A challenge in answering this question is the difficulty of controlling real-world referents across cultures. Most previous studies of SSym focus on language-specific phenomena (e.g. Hamano 1998, Bergen 2004, Hinton et al. 1994); cross-linguistic studies have focused on a narrow set of SSym relationships (e.g. D’Onofrio 2013, Iwasaki et al 2007). We address the challenge of cross-linguistic comparison by analyzing a rich new cross-linguistic dataset drawn from the Pokémon game franchise. Within this controlled universe we compare how English and Japanese use SSym in Pokémon names to encode the same range of meaning dimensions (Pokémon attributes).The 802 Pokémon creature names provide an ideal testing ground for SSym: Pokémon constitute a fixed set of cultural referents with quantifiable physical attributes, and 79% of their names differ across English and Japanese.

In an earlier study, Kawahara et al. (2017) found that moras and voiced obstruents correlate with size and weight in Japanese Pokémon names: a Pokémon named Bangirasu is likely to outweigh a Pokémon named Pippi. In the current study, we analyze a larger set of phonological properties and attributes in English and Japanese, including such key attributes as Pokémon evolutionary stage, power statistics, and gender. The results reveal both cross-linguistic and language-specific SSym.
 Three correlations are shared: name length and voiced obstruents with
 weight, and name length with evolutionary stage. Language-specific correspondences include bilabial consonants with weaker Pokémon in Japanese and low vowels with larger Pokémon in English. Significantly, no individual attribute-feature relationships are positively correlated in one language but negatively correlated in the other.

October 2, 2017 - Myrna Schwartz (MRRI)

"Phonological speech-error production in aphasia: A reappraisal"

My talk centers on results from our computational and anatomical studies in aphasia that were unanticipated and challenge current thinking about how spoken word production goes awry.  The findings were from lesion-deficit mapping with over 100 individuals with post-stroke aphasia, and they showed that patients’ rates of phonological errors in picture-naming and word repetition correlated with lesions in left hemisphere sensorimotor areas. The responsible lesions spared semantically relevant cortices (although naming is a semantically-driven task) and areas thought to house lexical-phonological representations (although our theoretical framework locates phonological errors within these representations). As will be discussed, the issues raised by this finding are central to contemporary neurobiological accounts of the language–speech interface. 

October 6, 2017 - John Alderete & Paul Tupper (Simon Fraser)

**Friday 12 - 1pm, 1229 Dwinelle**

"Phonological regularity, perceptual biases, and the role of grammar in speech error analysis"

This talk investigates a set of phonological patterns in the SFU Speech Error Database (SFUSED), with the goal of understanding if and how phonological grammar is involved in phonological encoding. It assesses 2,076 sound errors in English of various types (phonological substitutions, deletions, additions, etc.) for their phonological regularity (do they violate English phonotactics?) and their sensitivity to a variety of markedness constraints commonly used in speech error analysis. The results show that sound errors are much less regular (violate phonotactics more often) than reported in prior work, and they reveal no significant effects from eight distinct measures of phonological markedness. The higher degree of phonological irregularity is attributed to methodology, because the methods for collecting speech errors in SFUSED are demonstrably less prone to perceptual bias. While phonological encoding may require linguistic representations of planning units (e.g., segments), these findings suggest that some of the tools of phonological grammars, like syllable-structure algorithms and phonological markedness constraints, may not need be required in the retrieval of phonological segments in words.

October 9, 2017 - Seung Eun Chang (UC Berkeley)

"Enhancement effects of clear speech and word-initial position in Korean glides"

The current study investigated the enhancement effect in Korean speakers’ clear speech and word-initial position, using acoustic analyses of the Korean glides /w/ and /j/. The results showed that the transitions of glides /w/ and /j/ at onset were enhanced in clear speech with an expanded vowel space. An expanded vowel space was also observed in the word-initial position, but the expansion was not statistically significant. However, the significant interaction between speaking style and word position revealed that the articulatory and global modifications in clear speech were noticeably greater at onset in the word-medial compared to the word-initial position. Also, the mid-front vowel /e/ shifted downward and leftward in clear speech, indicating that mid-front vowels are fronted and lowered in clear speech. As a language-specific issue, no phonetic evidence was found supporting the existence of two Korean glides, /we/ and /we/, even in clear speech and word-initial position, indicating a diachronic sound merger of these two glides. In addition, the glide /je/ after a consonant was neutralized into /e/ in casual speech. These findings suggest a relationship between speaking style effects, word position effects, and changing phonetic targets due to diachronic sound change.

October 16, 2017 - Patrice Beddor (U Michigan)

"Differential cue weighting or sound change? Devoicing in Afrikaans"

The initial portion of my presentation will briefly document the relation between f0 and prevoicing in the production and perception of plosive voicing in Afrikaans (based on a study conducted in collaboration with Andries Coetzee). Acoustic data show that Afrikaans speakers differ in how likely they are to produce prevoicing to mark phonologically voiced plosives: younger speakers are more likely than older speakers to devoice. However, both younger and older speakers produce large and systematic f0 differences after phonologically voiced and voiceless plosives to convey the contrast between the voicing categories. This pattern, which is partially generationally structured, is mirrored in these participants' perception: although younger listeners rely more than older speakers on prevoicing as a perceptual cue, all listeners use f0 (especially in the absence of prevoicing) to perceptually differentiate phonologically voiced and voiceless plosives.

I would be interested in turning the bulk of my presentation into a discussion of the interpretation of these results in terms of age-graded cue weighting vs. incipient tonogenesis. A challenge to the tonogenesis interpretation is that an increase in the size of the f0 difference after phonologically voiced and voiceless plosives has not been documented for Afrikaans. This is unlike Korean tonogenesis, where younger speakers of Seoul Korean produce larger f0 effects than do older speakers, and where there is evidence that f0 and phonation changes are occurring in tandem (Kang 2014, J. Phonetics). It is tempting to argue that Afrikaans is following a different trajectory towards tonogenesis in which the exaggeration of f0 was first stabilized for all members of the speech community, with subsequent—and ongoing—loss of voicing. However, the data are also consistent with age grading, that is, with the possibility that younger participants, as they age, may develop voicing repertoires similar to those of the older participants.

October 23, 2017 - CILLA Practice Talks

"Stress and Tone in Cushillococha Ticuna" - Amalia Skilton

Cushillococha Ticuna is a variety of Ticuna (tca; isolate) spoken in the town of Cushillococha, Peru. It has eight contrastive tone melodies on monosyllabic words (Anderson 1959); maximal tone density; and extensive grammatical tone. Drawing on fieldwork, I argue that this language also displays stress. Stress interacts extensively with tone, licensing additional tone contrasts and conditioning tonal alternations.

"Nasal segments in Tupí-Guaraní: A comparative synthesis" - Myriam Lapierre & Lev Michael

Complex nasal segments (e.g. [mb], [nd], and [ŋɡ]) and nasal harmony are among the most prominent topics in lowland South American phonology (Storto & Demolin 2012), with the Tupí-Guaraní (TG) family being one of the major genetic groupings that exhibit these phenomena. Despite the ubiquity of these phenomena in TG languages, however, works on this family have not reached consensus on how to analyze them, with similar systems of nasality-related phenomena receiving starkly different analyses. This talk presents an analytical synthesis of the phonology of nasal segments in TG, based on a comprehensive review of descriptive and analytical works on these phenomena in the languages of the family. We focus on the analysis of allophonic alternations between fully nasal stops (e.g. [m]), partially nasal stops (e.g. [mb]), and fully oral voiced stops (e.g. [b]), and argue that empirical generalizations regarding these alternations support a unified analysis at the level of the whole family.

October 30, 2017 - Margaret Cychosz (UC Berkeley)

"Language-specific Sources of Acoustic Stability in Phonological Development"

Acoustic variability in speech is ubiquitous. In adults it often stems from articulatory overlap and reduction. In children it can be attributed to anatomical difference from adults or motor planning immaturity. A complete understanding of the sources of child speech variability can shed light on how motor control and the vocal tract develop through adolescence. So here I explore an additional source of variability: phonological structure of the language being learned. Previous conclusions on vocalic development in young children were drawn from languages with large vowel inventories (e.g. English, French). But vowel inventory size and the acoustic dispersion of phonological categories may be negatively correlated. I compared spectral variability across adult and child speakers of Chuquisaca Bolivian Quechua, a highly-agglutinating three-vowel (/a, i, u/) language. Child participants aged 5;0-6;0, 7;0-8;0, and 9;0-10;0 completed a picture selection and description task and adults narrated the Duck Story. Results suggest that children’s acoustic variability does not reliably differ from adults’. However, all vowel categories were more dispersed in suffixes than root words. This is evidence that variability may not stem entirely from articulatory limitations – inventory size may also influence attainment of acoustic stability.

November 20, 2017 - Claudia Valdivia (UC Berkeley)

"Effect of Speaker on the Nonword Repetition Task (NWR) in Monolingual and Sequential Bilingual Children"

When children begin attending a formal school setting, the dynamic of their input changes from predominantly adult input to including significant amounts of peer input. This shift in input is especially salient for sequential bilinguals with a different home language for which starting school also means exposure to a new language. In this project, I try to investigate if monolingual and sequential bilingual children treat adult and peer voices differently in a nonword repetition task. 32 4-6yr old children of both language backgrounds were tested as well as 16 monolingual and bilingual adults. Though data collection is not complete, preliminary results suggest that there is no significant difference in phoneme accuracy between child and adult speaker. Post-testing questions reveal that both children and adults rate their peer-aged voices more favorably and make more identity-based claims. Preliminary findings about duration matching, pitch matching, and other phonological features of repetition will also be discussed.

November 27, 2017 - Colin Wilson (Johns Hopkins), with Eleanor Chodroff (Northwestern)

"A restriction on laryngeal realization within and across languages"

The same traditional phonetic category (e.g., voiceless aspirated bilabial stop) can have different realizations across languages and across speakers of the same language. However, there are important restrictions on this type of phonetic variation: for example, all other things being equal high vowels have higher intrinsic fundamental frequencies than low vowels (e.g., Whalen & Levitt, 1995) and aspirated velar stops have longer voice onset time (VOT) values than labials (e.g., Cho & Ladefoged, 1999). In this talk, we investigate variation of voiceless stop VOT in (i) a large multi-talker corpus of American English and (ii) a typological survey of 80+ diverse languages. Both sets of data provide strong support for a novel restriction on phonetic realization: in the grammar of a given speaker, members of the same voiceless stop series must have highly similar glottal spreading and timing targets. The restriction follows from models of the phonetics-phonology interface in which abstract features are mapped to phonetic targets by a computation that has limited sensitivity to segment-internal context. Furthermore, the restriction provides a rational basis for observed generalization of speaker-specific VOT values in adaptation and imitation experiments, and could be a useful source of prior knowledge for first language acquisition of laryngeal distinctions.

December 4, 2017 - Ryan Bennett (UC Santa Cruz)

"Recursive prosodic words in Kaqchikel (Mayan)"

Following the development of prosodic hierarchy theory (Selkirk 1984, Nespor & Vogel 1986), evidence has accumulated that prosodic categories may be recursively self-embedded (e.g. Selkirk 1995, Truckenbrodt 1999, Wagner 2010, Itô & Mester 2013, etc.). However, this conclusion is not universally accepted (e.g. Vogel 2009a), and even the need for prosodic categories has been recently disputed (e.g. Scheer 2012b).  In this talk I argue that the prefixal phonology of Kaqchikel, a Mayan language, provides clear and convincing evidence for unbounded (iterable) recursion of the prosodic word (w). Patterns of glottal stop insertion and degemination receive a simple, elegant treatment if recursion of the prosodic word is permitted. Theories of prosodic phonology which do without recursion are forced to resort to a series of ad hoc stipulations to account for the same facts. Both derivational (e.g. Kiparsky 1982) and transderivational (e.g. Benua 2000) analyses of these patterns fail on morphological grounds. The overall conclusion is that both abstract prosodic structure and recursion of the prosodic word are indispensable parts of any theory of word-level phonology.

December 11, 2017 - Sharese King (Stanford)

"Language, locality and economy: Understanding Vocalic Variation among African Americans from Rochester, New York"

In this talk, I examine the relationship between social change and sound change, juxtaposing the collapse of industry in a Rust Belt region to generational sound change. Particularly, I investigate the production of TRAP in oral and nasal contexts across a sample of 24 African American speakers from Rochester, New York. There is a change in apparent time such that younger speakers are lowering TRAP, while raising BAN. Furthermore, speakers’ stances toward their position in Rochester’s post-industrial economy influence the degree to which speakers lower and retract TRAP tokens. These results are couched in a larger discussion about the exploration of social and linguistic diversity across African American speakers.