Voices in Society: A Workshop on Prosody

This week I got to spend three busy days in the always busy (yet energy-infusing) London, more specifically at Queen Mary University, to learn more about prosody. The workshop was called "Voices in Society", it was organised by PhD candidates Andy Law and Elisa Passoni, and Dr Esther de Leeuw and it was funded by the ESRC.

Excited as I always am about learning more about prosody, this workshop was really overwhelming in terms of all the things there are out there to continue learning about! I will make a brief review of some of the areas and points raised during some of the talks, and I will, whenever possible, add links to research material, tools or website of the presenters, if any of my readers wishes to go further into any of the topics.

***

The usual disclaimer (because I have to): As this is a personal blog on personal and academic interests of mine, what follows is not a formal review of the workshop, and it was written as a means of summarising my "take-home points". It is, of course, also meant to share my experience with my teacher-readers. It was written in almost one sitting (on the train from London to York, actually!). And of course, any errors or evaluative comments are entirely my own, and do not represent the views of my University.

***

Day 1: Intro to Prosody, and Rhythm

Dr Esther de Leeuw discussed the huge realm of prosody by reviewing some of the basics. We were (re)introduced to the subdomains of prosody, and their perceptual and acoustic correlates (with the usual warnings, like F0 not strictly being "pitch", and the fact that our perception of pitch is not linear, but logarithmic). We got to play around with the thresholds of human perception of pitch, using the Hearing Test of the University of South Wales here. Dr de Leeuw also addressed some of the distinctions (or lack of) in the literature, such as the suprasegmental-prosody labelling, and the differences between tone and intonation. 

We tried an exercise of perception of the Three Ts from a scene in "The King's Speech", and simple as it was, it still revealed some of the inter-speaker differences in perception we may find, especially in terms of segmentation and the level vs rise contrast.

In the second session, we did some Praat analyses of tonal alignment, pitch span and pitch level. It turned out to be a complex exercise, but the following sessions were very helpful in clarifying some of the uses of the measurements made (especially for tonal alignment, which is something I never thought of considering for CA).

Dr Lawrence White discussed rhythm and everything involved in it in great detail, and with some very cool videos! Dr White addressed the now debunked stress-timing vs syllable-timing dichotomy by referring to some studies that analysed rhythm production and perception cross-linguistically. Phonotactic constraints and syllable constitution, as well as the fact that many studies were conducted from the basis of stress as it happens in English -thus disregarding the role of stress in other languages-, were found to be responsible for the initial categorization of these two distinctions. Dr White proposed a metric that entails determining the interaction between percentages and ratios of vowels over consonants and vowel duration and standard deviation. We got a really detailed and exciting demonstration of segmentation in spectograms according to manner of articulation to help us make the right measurements.
After the teaching session, Dr White discussed some of the rhythm perception studies on different languages he conducted on babies, some entertaining SASASA perception tests (where syllables in an utterance are replaced synthetically with SASASA), and a few findings regarding the role of phrase final vowel lengthening.

Day 2: ToBI, PRAAT scripting, and research on L1 attrition

Prof. Mariapaola D'mperio delivered a great introductory workshop on AM and ToBI. There was a very interesting historical review of how ToBI came to be, and what were the great findings in Pierrehumbert's 1980 thesis, as well as very convincing arguments as to the representational goals and uses for prosodic typology. Some very interesting points were tackled in passing, such as compression and truncation and alignment. A great part of the session was devoted to discussing the elements of prosodic structure, and then we tried a few exercises which can be found in the MIT site.
Even though I had some very basic (very, very basic) background in ToBI, I have to admit I was prejudiced, but Prof. D'imperio said all the right things: the importance of listening (over merely looking at the PRAAT waveform), the fact that phonetics and phonology have fuzzy edges at times, and that what is contrastive is at times difficult to define, as well as the need for multidisciplinary work on prosody. Once again, I confirm that a great part of what you can "hear" is affected by the model in which one has been originally trained: I hear step ups where perhaps some people who do ToBI hear glides; I understand the notion of rise-falls  on a nucleus as a different notion from what some people in ToBI woud call a LH*L-L%, for instance, where there may be a low "onset" (in British terms) and a fall on the nucleus. I think this session was a great eye-opener for me to understand many things that in the past I found contradictory in papers, and to confirm that we (either ToBI or British-school people) hear the same phenomena, but we process it and label it differently (with all the implications that this entails, of course...).
In the afternoon, Prof D'imperio discussed her research, and she mentioned some very interesting studies that address the complexities of chunking and phrasing (especially in French), and how different types of cues work differently in different languages (fascinating!)

Dr Stephen Welburn made an introduction to PRAAT scripting, which makes automatising tasks in PRAAT much simpler. It was only a brief tour, but I'm certainly going to try and start designing my own scripts! 

Dr Anouschka Foltz made a very interesting discussion of her studies of L1 attrition, which broadly defined is the loss of native-like proficiency of speakers of L1 as speakers grow to be bilinguals (perfect description of me, I am no longer a reliable speaker of Spanish, especially in terms of intonation!). Dr Foltz discussed the intonation of questions in Arabic speakers of English, and then presented another study of perception of contrastive accent and eye-gaze anticipation.

Day 3: Stats, R, and more research studies

Dr Anouschka Foltz made a quick, yet clearly-presented tour of different functions in R, including basic stats, mixed models and Growth Curve Analysis. Lots of work for me to continue doing in this area!

Dr Candide Simard presented her research consisting in the documentation and description of an Australian aboriginal language, Jaminjung (see Wikipedia entry here). It was a truly fascinating presentation that discussed the vicissitudes of doing this kind of work with a very special community, with specific cultural traditions, and a language which has got less than 50 native speakers at the moment. Dr Simard modelled the intonation systems of this language with reference to topic and focus (theme and rheme), communicative function, and to other emerging structures. 

Dr Erez Levon covered the topic of High Rising Terminals in London speech in young men and women. Dr Levon attempted to find clear acoustic differences between HRT fulfilling different functions, but the findings were not signficant in this respect. What was found in this particular context is the tendency for women to use uptalk to request alignment, and for men to apply it for affiliation-seeking purposes, among other more relevant functions (which I have to admit I did not get to take down). I think my attention was focused on these functions because I believe there could be a different realisation of the rise for alignment or affiliation projection, so I actually presented this as a suggestion in the Q&A session for the presenter to perhaps look into.
I am not a sociolinguist, but I sometimes find the men vs women criteria for exploration a bit problematic for reliable findings, and I would perhaps work on other variables myself. I think, in any case, that it is very interesting that there were no significant differences in the realisation of these rises. Who knows, perhaps conversation analysis could help untangle this point! (Sorry, could not help it!) (Plus, I do believe there must be something in the way these rises are produced that leads to their being interpreted differently!)

Dr Tiina Eilola presented some interesting studies on the way people perceive the language of emotions in bilinguals.  By referring to models such as the Revised Hierarchical Model (Kroll and Stewart 1994) and the Circumplex Model of Affect (Russell, 1980), Dr Eilola discussed her framework and findings on Gujarati speakers of English who moved to the UK. She explained the different processing of "positive", "negative" and "neutral" emotions, and the participants' evaluation of English as being the language of emotions for them over their own language (which reminds me of a lot of feelings I have in connection to my L1 and my L2)

***

I know I am missing out on a lot of detail here, but I just wanted to give you an idea of all the fascinating learning experiences, research and topics I have been trained in the last three days. If any of you happens to be interested in any of the presentations, and you wish to ask further, please use the comments box below, or check out the researcher's websites for related publications and possible ways of contacting them.
 A huuuuge thank you to Andy, Elisa and Esther for the great organisation of the event. And let's face it: prosody rocks!