Listening presents significant challenges to second language learners (e.g., Bloomfield et al., 2011), and teachers can benefit from an increased understanding of the ways students succeed and fail in converting a stream of sound into meaning. This article summarizes a study focusing on aural decoding, the process of recognizing words from speech, and recommends the data collection and analysis method used in the study—a method known as paused transcription (Field, 2008)—as a classroom activity. (For a fuller description of the study and suggested pedagogical interventions, see Sheppard & Butler, in press.)

During paused transcription, participants listen to recorded texts in which pauses have been inserted at irregular intervals. When participants encounter a pause, they write down the last phrase they have heard in the recorded text. Transcriptions can then be analyzed to identify errors. 

Participants

Participants in our study included two groups: an advanced group composed of 48 students in Level 6 of a 6-level intensive English program (IEP) at an American university, and a middle-level group consisting of 29 students in Level 3 of the same program. These students included native speakers of Chinese (65.4%), Japanese (10.2%), and Arabic (24.4%). The two groups had a similar length of study in the United States. 

Texts

The recordings included three audio texts, each 3 minutes in length. The texts were graded easy (selected from unused audio materials associated with the textbook used in our IEP Level 2 course), medium (selected from unused audio materials associated with the textbook used in our IEP Level 5 course), and hard (an introductory excerpt from an authentic recorded undergraduate university course). In each text, we identified 12 target phrases, with 4 target words in each phrase, for a total of 144 target words. All of the target words were estimated to be familiar to all of the participants. (Of the words, 141 were from the first 1,000 most common English words, and the other three were from the second 1,000: dance, repeat, and probably.) Data analysis consisted of analyzing differences between participant transcriptions and actual words in the recording.

Analysis 

The analysis focused on four areas of interest: (1) how accurately participants decoded the words of the text overall, (2) whether participants decoded more content words than function words, (3) whether speed of articulation in the text affected the accuracy of decoding, and (4) what insights about participants’ listening errors could be gained from their transcriptions. Following, we discuss the findings from each of these areas of interest.

1. Overall Transcription Accuracy

The overall average transcription accuracy was 67%, with accuracy for the upper level students at 73% and lower level students at 54%. In the most advanced text, the advanced students could accurately decode only about 60% of the target words. Because a lexical decoding accuracy of 90–95% is estimated to be necessary for adequate listening comprehension (Van Zeeland & Schmidt, 2012) and these students were only weeks away from entering the university, this finding suggests that such students may face significant challenges in lecture comprehension as a result of undecoded words.

2. Accuracy in Decoding Content vs. Function Words

Participants transcribed content words more accurately than function words (a finding in line with previous studies, e.g. Field, 2008) at a rate of 76% vs. 54%, respectively. Because function words can have a significant effect on meaning (e.g., compare “He bought it for them,” “He bought it from them,” and “She bought it for him”) and because they are often reduced in speech, this result suggests that students could benefit from improved listening training aimed at recognizing function words.

3. Effect of Speech Articulation Rate on Accuracy of Listeners’ Decoding

A simple measure of articulation rate (pronounced syllables per second within the target phrase) was compared to a measure of phrase difficulty (average percent of participants who correctly transcribed each of the four words). Results show no significant correlation between articulation speed in the target phrases and participant accuracy in decoding these phrases (r = –0.253, n = 36, p = 0.137). Thus, contrary to many language students’ stated beliefs (e.g., Bloomfield et al., 2011), slower speech in the target phrases did not result in better listening accuracy. Informing students of such findings could help them adjust their expectations and emotions regarding speech rate when they are listening. 

4. Insights About Listening Processes From Participants’ Transcriptions

Qualitative analysis of the transcription data revealed several types of decoding errors involved in the listening processes of participants. The most common error types included

• misidentification of word boundaries (often with a stressed syllable misidentified as the onset of a new content word),
• misidentification of phonemes (seen in vowels and consonants at all positions in words),
• various approaches to transcribing words that were not recognized (e.g., attempting a phonological transcription, or filling in with a different word that shares some sounds with the target word), and
• top-down fabrications (writing words or phrases related to the topic at hand but not phonologically similar to the words actually spoken).
  
  

Details and examples of each of these error types, along with suggested classroom activities to develop learner skills, can be found in Sheppard and Butler (in press). 

Using Paused Transcription in the Classroom

Just as the paused transcription methodology was useful for identifying patterns of student errors in this study, it can also be easily adapted for use as a tool for identifying listening problems in the classroom. This method can allow teachers and learners to become aware of specific errors in aural decoding and lead to opportunities for targeted interventions. For example, if teachers or students identify a pattern of errors in word segmentation, teachers can give dictations delivered with natural connected speech, specifically including challenging word boundaries. If problems identifying certain phonemes are noticed, students can practice with minimal pairs or use software providing high variability perception practice with phonemes (e.g., http://www.englishaccentcoach.com/). If many words are going unrecognized, and especially if teachers and students identify these as known words, aural recognition should be integrated into all vocabulary instruction using simple techniques such as dictation, elicited imitation, and aural analysis of new words (“How many syllables?” “What vowel?”). If learners are found to rely too heavily on top-down fabrication, it may be valuable to spend more time discussing listening strategies such as comprehension monitoring (e.g., ask “How sure are you?” along with all other comprehension questions) and making and checking predictions.  

It is not difficult to apply paused transcription in the classroom, and it can be done in just a few minutes. Teachers can choose a listening text and prepare it for in-class paused transcription by inserting beeps and pauses into the recording using audio manipulation software or by planning where to pause the recording manually. After students transcribe the target phrases, teachers can show the full text and ask students to correct their own errors. Teachers can then either analyze the errors or guide students in analyzing their own errors and design activities aimed at overcoming problems with word segmentation, phoneme recognition, word recognition, or other issues. Along with allowing for focused assessment and discussion of strategies to address listening problems, such paused transcription activities can build curiosity and autonomy in students’ approach to learning, increase student engagement, and reduce student anxiety, creating in the process a more effective approach to improving student listening.

References

Bloomfield, A., Wayland, S., Rhoades, E., Blodgett, A., Linck, J., & Ross, S. (2011). What makes listening difficult? Factors affecting second language listening comprehension. Technical report. College Park, MD: University of Maryland Center for Advanced Study of Language.

Field, J. (2008). Bricks or mortar: Which part of the input does a second language listener rely on? TESOL Quarterly, 42, 411–432. 

Sheppard, B., & Butler, B. (in press). Insights into student listening from paused transcription. CATESOL Journal.

Van Zeeland, H., & Schmitt, N. (2012). Lexical coverage and L1 and L2 listening comprehension: The same or different from reading comprehension? Applied Linguistics, 34(4), 457–479.

Brian Butler is the associate director of the Center for Language Education at Southern University of Science and Technology in Shenzhen, China. He teaches academic reading and writing and uses experimental research methodologies to explore and explain the functions of the English article system. 

Beth Sheppard teaches and develops curriculum for ESL listening and speaking at the University of Oregon. She is also involved in teacher training.