Today I was really lucky to get to attend a fantastic workshop/talk on pronunciation at my Uni called "Pronunciation Matters (?)", delivered by Prof. Sam Hellmuth. I was really eager to attend because I had the chance of auditing Prof. Hellmuth's classes this term and I can attest to her great knowledge, experience and teaching skills (and many of you know I am not easy when it comes to praising people!).
Below you will find a short account of the talk in my usual colloquial style and in one sitting (so sorry about any potential typos!). As I always say, any misunderstanding of the content presented is my own fault, and evaluative comments on the presentation are attributable to me alone (unless explicitly attributed to someone else!).
The session began with Sam's exploration of her journey from being a learner of Arabic -and a few embarrasing moments of her own- to becoming a teacher. This first anecdote lead to questions regarding pronunciation goals and how we can measure pronunciation gain, issues which were discused during the presentation by the presenter and the participants (as there a few slots for group discussion, which was really exciting!). (BTW, if you want to learn more about pronunciation goals and the complexity behind their selection, you can always go back to my early 2015 blog posts: here and here)
The first group discussion already triggered those heated pron-teaching-related discussions we are all familiar with: accentedness vs intelligibility, the difficulties in defining intelligibility, what accents we should teach, what accents learners want to learn...After the discussions, Sam also mentioned some well-known references in the EFL and ELF worlds, such as the paper on L2 accent and credibility issues, and accent discrimination problems in the US in the job market, among other things.
The following step in the presentation included a quick review of studies on three forms of "pronunciation gain measure": (foreign) accentedness, intelligibility, comprehensibility (these terms have been widely defined in the papers and books by Derwing and Munro if you need to trace them back, more currently discussed in the 2015 Pronunciation Fundamentals book, which guides some of the discussion by Sam Hellmuth in this lit review). There was a very interesting overview of different research findings which could potentially point to interesting teaching priorities, which I will try to summarise below:
- accentedness, that is, the "perception of how different the accent in question is to the L1 accent used as reference" was highly dependent on suprasegmentals, though of course segmentals also play a part (Anderson-Hieh et al 1992)
- Quick detour! Here's a fun fact that Sam presented: an accent can be even detected in speech recordings played backwards, according to Munro et al 2013 (which could be attributed to voice quality, or in my view, to base of articulation)
- intelligibility, that is, "the extent to which someone understands whath was said", is not dependent necessarily on levels of accentedness. That is, a native-like accent does not necessarily make an intelligible accent. Suprasegmentals appear to have a diminished role in ensuring intelligibility according to the studies by Munro and Derwing (1999) and others. (I have my objections here, as usual...but they would need a whole post)
- comprehensibility, in other words, "the listener's perception of how difficult it is to understand an utterance", can also be affected by prosodic choices.
- An alternative measure could be that of fluency, but it was not planned as part of this presentation (and I agree, it does deserve a presentation of its own!).
Consequently, these variables can be affected by different types of "pronunciation errors" and carry important teaching implications regarding goals, and the teaching of segmental and suprasegmentals.
Another quick reference was made to the (in)famous Critical Period Hypothesis and the findings by Munro and Mann (2005) stating that it is possible to improve your pronunciation after puberty (which, to be fair, is what most of my teacher trainees and myself have done! We ought to consider ourselves heroic!?). (BTW, more on the CPH and pronunciation in my review of Linda Grant's "Pronunciation Myths" book here). This could be seen as part of the "good news" regarding pronunciation improvement.
A very interesting comparison ensued, between two possible selections of features to teach when it comes to pronunciation:
> NS hearer
1. word and sentence stress
2. syllable structure (phonotatictics)
3. vowel length distinctions
4. major consonant distinctions
5. vowel quality distinction (those with a high functional load)
6. minor consonant distinctions (those with a low functional load).
1. consonant inventory
2. some phonetic detail
3. consonant clusters
5.sentence stress (especially for contrast)
Note: Jenkins' proposal is described in more detail in the ELF blog: https://elfpron.wordpress.com/2013/11/21/what-is-the-lfc/
A brief explanation of functional load followed, together with some comments on possible criticism to current teaching materials that place too much emphasis on contrasts like /ð,θ/ which have a low functional load in English. And here I need to input my own voice: I agree entirely with this point, but I think the problem is not the inclusion of this pair as part of the materials, but the actual grouping of features. I completely understand the fact that these two sounds are articulatorily similar and differing in voice, BUT I think a more teaching-friendly approach would present this sound together with those that may bring trouble for learners, such as /d/ vs /ð/ or /s/ vs /θ/ for Spanish speaking learners of English, for example.
Finally, in this lit review, some of the most interesting bits appeared: the discussion of how possible ways as to how we can actually measure pronunciation gain, and what different bits of research have found.
- Saito (2013) has reported that training phonetic perception may be a way of improving in production, which is a claim that has been also tested in other previous studies in other languages. (BTW, if you want to read my report on a talk by Saito earlier this term, click here).
- The Beginner Language Learner Survey conduected in June 2014 with learners of German at the Uni of York compared students' answers regarding their feelings when it comes to pronunciation: native speakers of English learning German valued comprehensibility more over intelligibility and accentedness, though the three of them were rated high. Those learners of German whose L1 was not English valued intellibility more highly over accentedness (with quite a low rating in comparison) and comprehensibility.
- A very interesting piece of research by Sam Hellmuth and Florence Edwards compared learners' accentedness ratings to establish whether pronunciation improvement could be due to their "year abroad" experience or to individual cognitive abilities. The data, which included a questionnaire regarding their "learning context factors" (age of acquisition of L2, experience abroad, motivation for a native-like accent, etc), and the results of a non-word repetition task (Gupta,2003) pointed at very interesting findings: students who appear to have been graded as more native-like are in fact those who have a better "phonological working memory" (PWM), the ability to retain phonetic detail in your mind and reproduce it. These were presented as the "bad news", since the "year abroad" experience did not necessarily lead to a more native-like accent, and this would relegate accentedness achievements only to the "gifted". However...
- a second study, this time on "stress deafness", implies that there are many things that can be learned and taught, and that the phonological working memory can be trained, and that the knowledge of other languages is really beneficial when it comes to pronunciation matters. This other study was initially done for French by Dupoux et al (2007) and it was replicated by by Bethany White (2013 graduate), and it tested native English speakers' perception of stress in English, Spanish and Japanese non-words. In short, the results revealed that students had a lot of trouble in the "sequence recall task" (retrieval task of strings of words with different stress patterns and a final "OK" which blocked access to acoustic memory), and those students who had some knowledge of Spanish did slightly better for Japanese. This study also established that in fact stress "deafness" in another language is a phonetic "business", as lack of reduction (lack of schwa) was a defining factor in those who had trouble identifying stress in Spanish and Japanese. (I have to admit I chickened out of asking this because I may have been entirely wrong, but as we listened to some of the non-words which were supposed to be controlled for intensity and pitch changes, I could not help noticing that the Japanese renderings did show some pitch difference across syllables (naturally!?) and I myself believe this can also affect perception, for obvious reasons. But I did not ask at the time, so treat this as just a fleeting impression!). This study has interesting implications for teaching, at least for English speakers learning another language, since it can be used to redefine the teaching of stress by actually helping students to tune in to phonetic features over stress rules, at least for perception purposes. And this also supports the tenets of those models of second language learning based on inhibition of first language perception (Darcy et al 2015, for example)
Even though it was a very short session (time did fly by! I wanted more!), a lot of very interesting comments, findings and debates cropped up. Once again, this shows how "hot" pronunciation is as a topic (in all possible senses!), how much research is still needed, and on a personal note, how I wish all the work being done in Argentina could get researched on and published, as we have so much to say on pronunciation teaching, and so very little reported.