Talk by Donna Erickson

Dr. Erickson gave a talk at Haskins Laboratories on Thursday, June 11th, titled Articulation of Rhythm: A Multilanguage Perspective. The slides include a lot of interesting data on jaw movement patterns in English, Spanish, French, Japanese, and Mandarin, as well as speech produced by non-native speakers. It ties together a few different projects that Dr. Erickson has been spearheading over the last few years, working with EMA data to test the connections between jaw movement patterns and prosodic structure across different languages.

Two New Papers

We’ve added two new papers in the Papers for Discussion section. Bridging articulation and perception: The C/D model and contrastive emphasis is an attempt at drawing parallels between the emphasis and pauses listeners hear and the articulatory gestures speakers use to communicate emphasis. A practical guide to calculating syllable prominence, timing and boundaries in the C/D model gives an overview aimed at helping readers who are new to the model use its features in their own research.

Take a look, and please leave us your comments!

New article from Donna Erickson, and comments from Jangwon Kim

We’ve got a new article up from Donna Erickson, giving her point of view on the C/D model. Spoiler alert: she likes it.

Also, Jangwon Kim from USC sent in a few thoughts to share here:

Here are my thoughts on the C/D model. I could not make a paragraph, but I hope these items are useful for the discussion you will have with other authors of the article about the C/D model.

Pros:
The C/D model provides a nice framework to represent the prosody of an utterance, which can be driven from articulatory movements.
The C/D model is a comprehensive model whose conceptual description for speech production covers both planning and execution.
Provides simple and algorithmic way of representing high-level prosodic structure using articulatory movements.
Open questions
(based on my thoughts and emails with the C/D model people: J.C. Williams, Caroline, Osamu, Donna and Patrizia:
1. For computational implementation of this model, some elements themselves and relations between the elements need to be quantitatively (mathematically) specified.
– Example 1: impulse response function (IRF) relates the strength of syllables to surface-level articulatory movements (via articulatory gestures). However, this has not been quantitatively modeled in literature.
– Example 2: It has not been understood how para-linguistic factors, e.g., emotion, age and health, affect utterance features, and further how the utterance features affect other C/D model parameters.
– Example 3: There has not been theoretical way to determine temporal center of the syllable triangle for the syllables without onset and/or coda from articulatory movements.
2. The C/D model has not been validated in natural speech, but only in lab speech.
My questions:
1. What is the critical articulator for each consonant? Has it been defined in objective terms in literature? And reasons for such choice? Donna said (in email) that only one articulator for each consonant had been considered as critical articulator, but why it should be one? What if there are two critical articulators? How to determine the iceberg points, and derive the center of the syllables for bottom-up analysis?
2. What to do for consonants other than stop, fricative and alveolar lateral approximant? These consonants have not been considered in empirical study in the C/D model framework.