2020 Workshops

On Tuesday 31st March, we will host a series of workshops on various topics related to phonetics. Each session will be two hours and we will have three workshops running concurrently in each of the slots. To cover the cost of lunch and coffee, there will be an additional charge of £6 - you'll see this on the online store. *Once you have registered for the conference via the online store, please indicate which workshops you would like to attend via this Google Form

(1) Are you thinking what I’m thinking? An introduction to a new methodology for evaluating judgements on the typicality of speech variant productions in given populations

Sula Ross, Erica Gold, Kate Earnshaw (University of Huddersfield)

When we listen to an individual speak, we often find ourselves listening for cues that can help us to identify which accent-group that individual belongs to. In doing so, we may also capture realisations that stand out as being atypical for a specific accent variety (e.g. a vowel quality being further forward than expected, or an atypical /r/ production). We may be confident in stating that individual variation in a speech variant is present, but what if you were asked to quantify how atypical their production was? Would your judgement, based on your knowledge and experience of that accent, differ to that of another phonetician?

Forensic phoneticians, as expert witnesses, are required to make judgements regarding the similarity and typicality of evidential recordings. Experts are often required to base their typicality assessments on estimations of the population distribution using their experience and expertise (Rose, 2002; Hughes & Wormald, 2017), as case-specific data collection (Rose, 2007; 2013) in generally impractical (French et al, 2010), and there is insufficient availability of forensically-relevant corpora (Hughes, 2014). It is highly unlikely, due to the nature of the variability within speech and the speed of language variation and change, that this will change (French, 2017). However, research has identified that the estimates provided by experienced forensic phoneticians can vary (Ross et al., 2016), but the extent of variation is currently unknown.

This workshop will introduce a new proposed methodology that is currently applied in various scientific fields where data is either a) unavailable or b) there is uncertainty due to insufficient amounts of data (Slottje et al., 2008) such as volcanology (Aspinall & Cooke, 2013) and health economics (O’Hagan et al., 2006). This process involves expressing opinions, or judgements, in the form of subjective probability distributions (O’Hagan et al., 2006) which are then calibrated during feedback sessions.

The workshop requires no prior knowledge of either probability or forensic phonetics, as a short introduction to both fields will be presented using practical examples. A demonstration of the methodology will then be provided using data from the newly-completed West Yorkshire Regional English Database (WYRED; Gold et al. 2018). I welcome all feedback from all members of BAAP.

(2) Using fine phonetic detail to inform the diagnosis of Mild Cognitive Impairment

Traci Walker (University of Sheffield)

To celebrate BAAP2020 at York, in line with a long established tradition in British phonetics, this workshop will focus primarily on fine grained impressionistic description of connected speech. Participants will analyse data collected by a research team at the University of Sheffield. Participants in the project have been referred to the memory clinic and asked to speak to a ‘digital doctor’: a pre-programmed talking head on a computer screen.

These data have been used to successfully classify patients with dementia from participants with non-organic memory complaints based on the automated analysis of conversational speech samples (Mirheidari et al. 2017, Mirheidari et al. 2018, Reuber et al. 2018, Walker et al. 2018). To date, all classifications are based on automated analyses of the participants’ responses, without any detailed impressionistic phonetic analysis.

The research team is now working on a four-way classification of participants: healthy controls, neurodegenerative dementia, functional memory disorder, and Mild Cognitive Impairment (MCI), a condition indicating cognitive decline worse than normal aging but not consistent with an Alzheimer’s disease diagnosis. The experienced consultant neurologists on the team describe the speech of people with MCI as more ‘effortful’. Therefore in this workshop we will focus on phonetically describing the talk of people with MCI, with special attention to the initiation and release of plosives. Diagnosing and monitoring people with MCI is important because 21-60% of people will later convert to a dementing illness, the most common of which is Alzheimer’s disease

I invite participants to bring a laptop and headphones to this discussion-based workshop in which I will provide data to analyse and discuss collectively in a consensus listening exercise. Some but by no means all of the topics open for discussion are the occurrence and nature of the following in the data:

  • Duration of closure

  • Type of release

  • Secondary articulations

  • Juncture

  • Differences across speakers and groups (age, dialect, sex, patient group)

(3) Voice quality synthesis with the UCLA Voice Synthesizer and KlattGrid

Jo Pearce (University of Glasgow)

Laryngeal components of voice quality, called phonation, can form linguistic contrasts as well as serve as cues to the identity, stance, or emotional state of the speaker. Investigating the various functions of voice quality often requires looking at how listeners perceive voice quality, but doing so requires the ability to create stimuli for a perception experiment with controlled differences in voice quality.

This workshop will explore how to take a voice and re-synthesize it with differences in voice quality using two pieces of software: the Praat implementation of the Klatt Synthesizer, called KlattGrid, and UCLA’s software for the analysis and synthesis of voice quality. These programs allow the user to take a voice and separate the glottal source from the filter, so that changes can be made to the source and thus to the resulting voice quality.

After looking at why you might want to synthesize voice quality differences and discussing the advantages and drawbacks of different methods of synthesizing voice quality differences, we will demonstrate both pieces of software and guide you through using both KlattGrid and the UCLA Voice Synthesizer for yourself. We will provide sample audio files to work with and the workshop will be taking place in a computer room with the software installed, so all you need to bring is yourself!

(4) The phonetics of talk in interaction

Richard Ogden, Marina Cantarutti (University of York)

The workshop will be an introduction for phoneticians to some of the basics of conversation analysis. The main topic will be turn-taking, but we will also touch on other topics like self-repair and sequential organisation. The workshop will primarily be hands-on: there will be a short overview of some of the issues and methods, followed by an extended data session where we work intensively on a stretch of conversation, aiming to understand its organisation in terms of its phonetic and prosodic design, and how these relate to other orders of organisation, such as turn construction, sequential organisation, and the linguistic (including lexical and syntactic) implementation of social action. One of the things we'll highlight is the orderliness of everyday talk. We will primarily use audio data from British English. The workshop is open to anyone, but there's a limit of 12 participants (allocated on a first come first serve basis). There will be a bit of reading to do before the workshop.

(5) Introductory Bayesian methods (using brms in R)

Clara Cohen (University of Glasgow), Eleanor Chodroff (University of York), David Howcroft (Heriot-Watt University)

We will provide a conceptual understanding of what Bayesian data analysis can do that frequentism cannot do, and offer some structured toy examples to show how to run a model in brms. A rough estimate of the timeline of events is as follows:

Hour 1: Conceptual Bayes with hand calculation

  • Bayes rule, and why it’s philosophically more aligned with our analyses than frequentism (probability of model given data, rather than data given model). 10 minutes

  • A starting example with coin flips. 20 minutes

  • Exercises on the coin-flip example. 10 minutes

  • A brief introduction to sampling from posterior distributions, when things can’t be solved analytically: 10 minutes.

  • Types of priors: 10 minutes

Hour 2: Experimental Bayes using brms

  • Walk through a simple example (simple regression, one or two parameters), no repeated measures. Introduce the logic of the brms analysis, with the specification of priors and the specification of the model. 15-20 minutes

  • Walk through the output of the model and how to interpret it. 10-15 minutes

  • Participants work through their own scaffolded exercise(s): 25 minutes

(6) How can we design voices for technology?

Selina Sutton (Northumbria University)

In our daily lives we now regularly come across voices that are output by technology. This includes home (e.g. Alexa) and mobile (e.g. Siri) voice user interfaces, self-checkout supermarket tills, and automated train and bus announcements. Yet the design of the voice has been minimally considered in comparison to other elements of the technology, and there is little guidance for how companies can select a voice to use in their product.

In this workshop we will ask 'How can we design a voice for a speech-output technology?' and draft a design process in response. Through a series of interactive activities, we will familiarise ourselves with the typical methods used to collect system requirements and test prototypes (many of which will be familiar, such as experiments, and ethnographic observation). We will decide which ones should be used, how and when in the process of designing a voice for a piece of technology. No prior knowledge or experience in this area is required, just an interest in the topic.

Please note: This is a data gathering workshop for a research project. Thus, informed consent is required to participate. Data collection methods will be audio recording, photography and making observational notes, all of which will be anonymised/pseudonymised prior to data analysis. For more information, please see the information sheet and consent form here, or contact the lead researcher, Selina Sutton (selina.sutton [at] northumbria.ac.uk)

(7) Developing a career involving Phonetics: Advice on researching, publishing, and collaborating

Marina Cantarutti, Justin Lo (University of York)

This workshop is aimed primarily at doctoral researchers who wish to learn more about pursuing a career in research or teaching that involves work with Phonetics. A panel composed of experienced ECRs, senior academics and members of editorial boards will share their experience and advice when it comes to developing a career that involves Phonetics, offering perspectives from both sides of the academic recruitment and publication processes. We will discuss a range of issues, such as interdisciplinarity, career pathways, applications, open science, and publication advice. After brief presentations by the panellists, participants will be invited to present their questions. Towards the end of the session, there will be an open forum to discuss a proposal for a doctoral network for students working with Phonetics.

Following on from ICPhS 2019 in Melbourne, the following two workshops have been designed in response to the need identified to establish whether or not there is a future for the century-old IPA certificate. BAAP provides an ideal forum for asking this question. At the same time, the workshops offer a traditional ear-training experience for interested participants. Duration and content of each component will aim to reflect participant expertise and preferences. Participants will be provided with hard copies of the IPA Chart and one list of symbols for transcribing English for reference during the workshop.

The workshops are offered in two parts, run consecutively. However, the content is 'stand-alone', so you can attend one part and not the other if you wish.

(8) Ear-training, transcribing and the IPA certificate examination (Part I)

Patricia Ashby (University of Westminster)

This workshop will outline the examination; this is important for individuals wondering if their students (or even themselves) are suitably trained to take and pass this award.

Introduction to the IPA’s Certificate examination

An outline of the examination, emphasizing (with interactive examples) the predominance of practical skills required to succeed and accessing the IPA’s Examination web page. Practical phonetic skills, including transcribing from a written text, take up 140 of the 200 points available in the examination; only 60 points are allocated to knowledge of phonetic theory.

Taking the ETD paper

This session will provide an opportunity to experience an actual ETD (ear- training-dictation) examination. Materials will be timed to enable feedback for self-evaluation and discussion at the end of the session.

(9) Ear-training, transcribing and the IPA certificate examination (Part II)

Patricia Ashby (University of Westminster)

It’s all nonsense!

Although provision is made in the examination for the possibility of non-English dictation based on a language unknown to the candidates, it is impossible to police this and nonsense words are the invariable option of choice. Participants will be able to opt here for either:

Option 1 exclusively exam-level nonsense ear-training, or:

Option 2 overview of an eight-session intensive training programme with experience of focused practical exercises – from beginner to expert!

What do we want? When do we want it?

Complementing Part I, Part II will combine ear-training (substitutions and nonsense words) and discussion, enabling participants to voice their ideas, opinions and impressions regarding the format, content, value and future of the IPA’s English Certificate award.