Ariana Avila and Dr. Mark Tanner, Linguistics and English Language Department
The purpose of this study was to gather acoustic data aimed at empirically assessing the degree to which non-native English speakers can approximate the linguistic features of native speech when engaged in imitative interaction. The focus of my study was to investigate the degree to which ESL learners living in an English speaking environment can produce linguistic features (variation in intonation, appropriate word and sentence stress, and appropriate use of pausing to delineate meaning units in discourse) in their speech similar to those produced by native English speakers when the ESL learner is trying to imitate what a native speaker of English sounds like. In order to measure this, we collected data from both native and nonnative English speakers across a range of proficiency levels.
We hoped that through this research, we could uncover the degree to which imitation effected non-native English speaker’s ability to produce prosodic features similar to that of a native speaker; features that can actually help them be more comprehensible. For more than 20 years now, empirical research (Anderson-Hsieh, Johnson, & Koehler, 1992; Derwing, Munro, and Wiebe, 1998; Derwing and Rossiter, 2003, Hahn, 2004, Tanner and Landon, 2009) has shown the importance of suprasegmentals in helping ESL learners be more comprehensible in their speech. Pronunciation instruction that incorporates attention to prosodic features such as pausing, stress, and pitch has been shown to significantly impact ESL learners overall measures of comprehensibility, even in as little as 12 weeks. We knew that if empirical research could identify a relative weighting for prosodic features, pronunciation instruction could then be strategically designed to help ESL learners boost their overall level of comprehensibility.
In order to capture the desired data, we created seven speaking tasks that included both oral readings and spontaneous speech prompts. The prompts were written in such a way as to encourage non-native English speakers to do half of the tasks speaking normally and then the other half trying to mimic an American accent. We collected this acoustic data from 13 ESL subjects using the software program Audacity. The data was then analyzed using the software program Praat for patterns of pausing.
In addition to the 13 ESL subjects, we also recorded 13 native English subjects in order to gather data against which we could compare the data from the non-native speakers.
From the seven tasks recorded, we chose to focus our analysis on tasks 1, 5, and 6 because each provided a different speaking scenario. Task 1 was a brief dialogue, task 5 was mimicked a phone message, and task 6 was a short story. Within these three tasks, we marked pausing patterns and then compared the patterns between the native and non-native speakers. Pauses were categorized in three groups as demonstrated by native speakers- a full pause (250 milliseconds or longer), a mix between a full pause and micro pause, and a micro pause (less than 250 milliseconds). In addition, we compared the “normal speaking” tasks against the “American accent” tasks in the non-native English speaker data that was collected.
The initial data comparison that we did was that of non-native English subject “normal speaking” data against “American accent” data. Unfortunately, there was not a significant difference between these two groups. Many of the prosodic features were the same between the two groups, thus suggesting that in reality mimicking did not have an effect on non-native English speaker’s level of comprehensibility. Our hypothesis was that there would be a significant difference as a result of non-native English speakers imitating what they thought to be an American accent, but according to the data, few non-native English speakers could effectively imitate the prosodic features of a native English speaker. The only difference that we saw between the two sample groups was a heightened level of intonation or animation when non-native English speakers were attempting to mimic a native English speaker.
This discovery closely relates to what we discovered when we compared data from native speakers with non-native speakers. The data collected from the native speakers shows a common pattern of pausing at written in punctuation marks as well as pausing according to phrasal boundaries. The pausing at punctuation marks was typically a full pause and the pauses after a phrase were micro pauses. In addition, there were occasional micro pauses throughout the sentence or paragraph according to the speaker’s emphasis on given words or concepts. The micropauses typically ranged from 50-150 milliseconds long.
Similarly, non-native speakers paused on punctuation marks, but spoke with many additional pauses due to added emphasis, a need to breathe, or other unknown factors. There was not one clear pattern among the non-native speaker data with regards to the additional pausing. Some of the possible contributing factors could have been a result of difficulty with contractions, comprehension process time, or difficulty with the variety of grammatical category inflections. These additional pauses led to incorrect division or pausing in phrases and thought groups. One additional insight is that non-native speakers had a tendency to use full pauses more often than micro pauses.
These results have provided an excellent base for further research. We originally intended to study all prosodic features, but in the end we only studied pausing patterns between native and non-native English speakers. Our research shows that non-native speakers struggled to accurately produce correct pausing patterns despite imitating and American. However, they did have a higher level of animation and a more diverse intonation when imitating and American. It would be beneficial to study other prosodic features to see if the results would be similar or different. In addition, we would like to gather more data over a larger variety of skill levels in order to root out possible factors that inadvertently are skewing the data.
According to the data collected, we have concluded that there is not a clear correlation between non-native English speaker’s accurate production of prosodic features and the pronunciation technique of imitation. There was not a significant difference between similar speaking tasks that were performed in a “normal speaking” form and in an “American accent” form. In addition, we concluded that imitation does not help non-native English speakers produce prosodic features similar to a native English speaker. While non-native speakers could match native speaker’s pausing patterns when following punctuation rules, they couldn’t accurately match pausing pattern of phrasal boundaries.
Anderson-Hsieh, J., Johnson, R. & Koehler, K. (1992). The relationship between native speaker judgments of nonnative pronunciation and deviance in segmentals, prosody, and syllable structure. Language Learning, 42, 529-555.
Derwing, T.M. & Munro, M.J. & Wiebe, G. (1998). Evidence in favor of a broad framework for pronunciation instruction. Language Learning, 48, 393-410.
Derwing, T. & Rossiter, M. (2003). The effects of pronunciation instruction on the accuracy, fluency, and complexity of L2 accented speech. Applied Language Learning, 13, 1-17.
Hahn, L. D. (2004). Primary Stress and Intelligibility: Research to Motivate the Teaching of Suprasegmentals. TESOL Quarterly, 38, 201-223. Open Doors. Report on International Educational Exchange. (2015). Washington, D.C.: Institute for International Education
Tanner, M. W. & Landon, M. M. (2009). The effects of Computer-Assisted Pronunciation Readings on ESL learners’ use of pausing, stress, intonation, and overall comprehensibility. Language Learning & Technology, 13, 57-71.