Despite many teachers agreeing that pronunciation training is important for learners, it has long been neglected in language classrooms (Murphy, 2014). Teachers report feeling uncertain about how best to teach it and also often note that there is little classroom time available for its instruction. Others believe that communicative language teaching is incompatible with focusing on the form of pronunciation, and so avoid it altogether (Isaacs, 2009). The net result is that when pronunciation instruction does happen, it is often incidental, primarily occurring in the context of particularly egregious breakdowns in communication.

Given these limitations and learners’ need for massive amounts of exposure and practice in order for more accurate pronunciation to develop, computer-assisted pronunciation training (CAPT) seems like an obvious complement to traditional classroom instruction. Not only can CAPT help raise learners’ awareness of English pronunciation features, but it also allows for the large-scale exposure and practice that is necessary for improvement.

Unfortunately, relatively few pronunciation software applications are available, and those that do exist are not always evidence-based (Derwing & Munro, 2015). Good pronunciation software should be capable of addressing individual differences in learning. So, for example, applications should be customizable to the specific needs of learners from particular first-language backgrounds. They should also be capable of focusing learners’ attention on sounds rather than only on words (Thomson & Derwing, 2016).

English Accent Coach

One such program is English Accent Coach (EAC; www.englishaccentcoach.com), a free web-based application that allows learners to autonomously develop their perception of English vowels and consonants. The name of the website reflects terminology used by end-users who search for it, rather than intending to imply that foreign accents are problematic. In fact, the aim of this application is to improve speakers’ intelligibility, not global accent.

EAC follows the High Variability Phonetic Training paradigm (HVPT). This technique is not as complex as its name makes it sound. Essentially, HVPT refers to a type of perceptual training that presents learners with training stimuli in numerous phonetic contexts, spoken by different talkers. It can be roughly conceived of as minimal pairs on steroids. Laboratory research has repeatedly demonstrated that this technique is more effective than traditional practices, in which training stimuli are spoken by a single talker, or in which training exercises focus on sounds in a limited number of contexts (Thomson, 2011). Unlike a simple minimal pair approach, HVPT promotes the extension of perceptual learning during training to new talkers and to new items. HVPT has also been shown to promote transfer of perceptual training to production (see Thomson, 2011, for a detailed overview).

EAC takes HVPT out of the laboratory and makes it available to learners on a large scale. Because the application is web-based, it can be used at learners’ leisure, as long as they have access to a computer with a high-speed internet connection, and a quiet space. EAC allows learners to focus on individual areas of difficulty in perceiving English vowels and consonants. Preset levels promote learning of target sounds in increasingly complex contexts, from isolated syllables (many nonsense syllables) to stressed syllables in disyllabic words. All training items are spoken by 30 talkers, to my awareness the largest number of talkers ever used in HVPT training. This degree of variability is intended to ensure that learners will not repeatedly hear the same training token during a single training session. During a typical session, learners hear tokens containing target vowels or consonants and must click on the phonetic symbol for the sound that they believe they heard (see Figure 1). After each click, the game provides immediate auditory and visual feedback on the accuracy of the user’s selection. The vowel game also includes a switch, which allows users to click on color key words instead of phonetic symbols (e.g., green contains the sound /i/,blue contains /u/).

Figure 1. Screen shot of English Accent Coach vowel training application

Research Findings

Research using EAC and a precursor application (see Thomson, 2011) has been very promising. Apart from helping to determine if such training is effective, this research has also shed light on how second language pronunciation develops. For example, training learners to perceive 10 English vowels in one phonetic context (i.e., following /b/ and /p/) was found to result in only limited transfer of learning to the same ten vowels following other consonants. While the positive impact of training on the perception of vowels in the training context is encouraging, these early results demonstrated that second language pronunciation may not develop sound-by-sound, but rather context-by-context.

In a large-scale follow-up study over a 2-month period (comprising approximately 10-12 hours of combined training over 40 sessions) 15 learners were trained to recognize 10 English vowels. For the first nine sessions, vowels were presented in isolated /h + vowel/ syllables. For the next 29 sessions, the same vowels were presented after other English consonants. Finally, training returned to the original /h + vowel/ context for sessions 39 and 40. During the first nine sessions with the /h + vowel/ context, the learners’ recognition of English vowels improved session-by-session. However, despite 29 intervening training sessions in other contexts (10–38), when the /h + vowel/ context was reintroduced during session 39, no further improvement in vowel recognition was found to have occurred. This can be attributed to the fact that from sessions 10-38, learners only ever heard the 10 English vowels in other phonetic contexts. At the same time, whenever any other phonetic context was repeated (e.g., vowels after /b/, /d/, and /g/, during sessions 10 and 17), vowel recognition scores in those contexts also improved.

In another study (Thomson & Derwing, 2016), my colleague and I investigated whether EAC training that presents 10 vowels in nonsense syllables leads to greater improvement in the pronunciation of 70 target words containing those vowels, relative to training that primarily presents vowels in the 70 target words. To answer this question, 31 learners were randomly assigned to either the phonetically-oriented group, the word-oriented group, or a control group. An elicited imitation task was used to record the learners’ productions of the 70 target words, before and after training (i.e., they heard the words in the sentence, The next word is _____ and had to repeat the word in the sentence, Now I say ______). We found that the phonetically oriented training group significantly improved in their pronunciation of the 70 target words, while the word-focused group—trained to perceive vowels using recordings of the 70 target words—did not.

Another interesting findingfrom EAC studies is that age does not appear to limit learning. Learners as old as 60 years of age have participated in EAC research with encouraging results. While older learners tend to have the lowest scores at the outset, EAC facilitates rapid improvement. In fact, the oldest learner in the 2011 Thomson study, a 50-year-old female, demonstrated one of the greatest improvements.

Using EAC to Inform and Complement Traditional Classroom Instruction

Results of EAC studies (e.g., Thomson, 2011; Thomson & Derwing, 2016) can be used to inform classroom pronunciation instruction. For example, instead of assuming that using the minimal pair ship/sheep will result in the learning of these contrastive vowel sounds in all word pairs, teachers likely need to present many more pairs in order for learning to occur (e.g., hit/heat, sip/seep, fit/feet). Teachers should also be sure to incorporate some practice of target sounds in nonsense syllables, because this appears to lead to larger gains than training that contains only real words.

EAC cannot replace the teacher, but it provides an important complement to any course of pronunciation instruction. Learners can create their own user accounts, which will allow them and their teachers to track progress over time. In addition to a progress tool, learners can also download and print report cards for individual sessions. This functionality allows teachers to assign EAC activities as homework and gives learners the ability to demonstrate that specific training sessions have been completed.

References

Derwing, T. M., & Munro, M. J. (2015). Pronunciation fundamentals: evidence-based perspectives for L2 teaching and research. Philadelphia: John Benjamins Publishing Company.

Murphy, J. (2014). Myth 7: Teacher training programs provide adequate preparation in how to teach pronunciation, In L. Grant, L. (Ed.). Pronunciation myths: Applying second language research to classroom teaching. Ann Arbor: University of Michigan Press.

Thomson, R. I. (2011). Computer assisted pronunciation training: Targeting second language vowel perception improves pronunciation. CALICO Journal, 28, 744-765.

Isaacs, T. (2009). Integrating form and meaning in L2 pronunciation instruction. TESL Canada Journal, 27, 1-12.

Thomson, R. I., & Derwing, T. M. (2016). Is phonemic training using nonsense or real words more effective? In J. Levis, H. Le., I. Lucic, E. Simpson, & S. Vo (Eds.). Proceedings of the 7th Pronunciation in Second Language Learning and Teaching Conference. Oct. 2015. (pp. 88-97). Ames, IA: Iowa State University.

Ron Thomson is associate professor of applied linguistics at Brock University. His research focuses on the development of second language pronunciation and oral fluency.