LLS Colloquium: Two short talks from Ben Gibb-Reid and Mohammad Bani Younes

On Wednesday 9th March 2022, Ben Gibb-Reid (University of York) and Mohammad Bani Younes (Al Al-Bayt University, Jordan) will each present a talk on their current research. 

Just can’t get enough: The phonetic variation of just with  applications to forensic voice comparison (Ben Gibb-Reid, University of York)

A specific difficulty of the task of forensic voice comparison (FVC) is the lack of direct  correspondence in the content of different recordings. That is, recordings are unlikely to share many of the same words. Therefore, a frequently used word (or other feature) in naturally occurring speech may be of value to the FVC practitioner. To examine the usefulness of a feature in FVC, it is necessary to understand how variable it is between speakers, and the factors that affect it in different discourse positions or prosodic contexts. A short discourse-pragmatic marker (DPM) such as just could be analysed in this way for suitability as a diagnostic feature in FVC. In previous research, other DPMs such as filled pauses (uh, um) have been analysed as forensic voice comparison features with promising results. For the present study, just, STRUT and filled pauses were analysed for 100 male Southern Standard British English speakers (DyViS corpus). Voice comparison tests show that just vowel individual variation discriminates between speakers better than STRUT, though worse than the filled pause um, showing some promise for FVC application. 

     Studies of the DPM just have focused on its various discourse functions and overall frequency of occurrence, and show adverbial functions of just (‘to be precise’ and ‘to reference the recent past’) as distinct from emphatic (restricting/intensifying meaning) or planning (when used as a ‘filler’) functions. Phonetic variation of just has not been the subject of detailed study (with the exception of the Canberra Corpus Collective, forthcoming). This study compares variation in both the discourse functions of just and the phonetic realisation of its segments, in a similar way to analyses applied to like. In an analysis of the present data just is, as predicted, highly frequent in informal talk with a  frequency of 0.88 per 100 words. Analysing the realisation of just with vowel and /t/ elision and formant variation shows that the pronunciation of the word is in some way linked to its pragmatic function though not to its prosodic context. When just is used for discourse functions, it is phonetically reduced, with vowel elision and a centralised vowel. However, just as a marker of restriction (i.e. ‘nothing other than’) has less vowel elision and when a vowel does occur it is nearer [ʌ].

A Universal typology of the disambiguation of disjunctive questions Mohammad Bani Younes (Al Al-Bayt University)

Alternative questions (altqs) and disjunctive yes-no questions (dynqs) in English and Arabic  are string-identical. For example, do you have a mobile or an iPad? can be interpreted either as an altq or a dynq in English. These two readings are disambiguated by prosody such as accent distribution and the shape of the final intonation contour; with final falling intonation [\] the question is perceived by listeners as an altq, but with rising intonation [/], it is considered a dynq. One reason why altqs and dynqs may need to be disambiguated by prosody in English is that there is only one disjunctive element (DE) (or), that can be used in both types of questions. Little is known about the disambiguating cues in other languages. 

Dayal (2016) suggests that it is only prosody that is responsible for the disambiguation and  implies that this is a universal tendency across languages. Meertens (2019), on the other hand, proposed that languages disambiguate altqs and dynqs in one of the three ways: prosody-alone, disjunction-alone, or combination of both. A perception study from four Arabic dialects (74 Jordanian (JA), 52 Egyptian (EA), 70 Kuwaiti (KA), and 48 Syrian (SA) Arabic) showed that Arabic dialects fit three different types. Type 1 includes dialects in which the two disjunctive elements seem each to be specialised to a specific disjunctive question (to one meaning each). Type 2 comprises dialects in which there is an indication that one disjunctive element is specialised to one type of disjunctive question while the other is not (i.e., one disjunctive element is specialised, and one is general). This type could in principle be divided into Type 2A in which the specialised disjunctive element maps to altqs, and Type 2B in which the specialised disjunctive element indicates dynqs. Type 3 includes dialects that show no specialisation of disjunctive elements (i.e., both disjunctive elements are general) (see author, 2020). JA, EA, and KA were found to belong to Type 2A, and SA to Type 3. This typology was explored against a literature review of nine languages, including Mandarin, Korean, Turkish, Finnish, Basque, etc., and was able to accommodate all these languages, suggesting universal application. An interesting observation is that there was no language belonging to Type 2B. Additional details about more languages will follow in the presentation.

The talk will take place at 4pm on Zoom, and there will be an opportunity to ask questions at the end - you can join using this link.