Article title: An insight into listeners’ problems: too much bottom-up or too much top-down?
Journal: System 32 (2004)
Author: John Field, professor at University of Leeds, UK. Teaches psycholinguistics, child language, and English grammar. His widely-used textbook, Listening in the Language Classroom, won the Ben Warren International House Trust Prize. A teacher trainer, materials writer, and syllabus designer as well as practitioner, Field travels widely lecturing on L2 listening.
Type of Study: Empirical study based on observations gleaned from an analysis of three related experiments.
Purpose: To clarify the relationship between L2 listeners’ use of top-down and bottom-up listening strategies.
Research questions: 1) If top-down and bottom-up information are in apparent conflict, which one prevails? and 2) How do learners deal with new items of vocabulary when they crop up in a listening passage?
Procedure: 47 NNS students from a leading British EFL school were given listening tests in a classroom with good acoustics. The tests were designed to reveal learners’ listening strategies by presenting them with a series of problematic items, forcing them to choose between semantics (representing top-down listening strategy) or phonology (representing bottom-up). Each test was slightly different, and designed to explore different aspects of the research questions.
Results: The experiments produced both expected and unexpected results. Fields found that L2 learners often do re-interpret or misinterpret words to fit their own schema (relying on top-down, rather than bottom up processing), but also that they accurately perceive other words by relying on the onset sound. Lastly, he found that learners rely on a third strategy as well, which he calls a “lexical strategy”; in this case, learners bypass the top-down strategy in favor of matching an unknown word with a similar-sounding familiar word, regardless of the word’s semantic appropriateness.
My thoughts: First, let me pat myself on the back for choosing such a brilliant and easily comprehensible article to review. Now, let me see if I can elaborate a bit on the design, and its simple elegance.
Field begins with a history of the opposing views of bottom-up-influence versus top-down-influence proponents; midway into the discussion, in a section labeled “The legacy of scripted materials“, he points out an interesting connection that I hadn’t realized before: many L2 learners have developed an expectation of understanding everything in the text, since listening materials have traditionally been heavily scripted (contrived for the learner’s benefit, rather than reflecting natural spoken language) and graded according to level. Think about that: understanding everything in the text. That means doing the kind of precise and accurate bottom-up processing that is almost impossible for L2 listeners to do in a real-world context. Is it any wonder, then, that some learners panic when they’re exposed to the unpredictability of real conversation occurring at normal speed? Since the EFL world has become more “communicative”, learners must now rely more heavily on their top-down processing skills to hypothesize and compensate for the bits of language that they cannot fully process in the speech stream in real time. So while many scholars insist that L2 learners cannot focus on semantics when they’re unable to catch sounds and segment them into words, other scholars–such as Long (1989) and Field himself (1997)– counter that learners’ top-down processing ability is exactly what enables them to make sense of what they hear. This ability (they say) is what provides the support necessary to compensate for L2 learners’ imperfect decoding skills.
Moving on to the experiments themselves, I’ll briefly explain the design:
The first experiment: This test consisted of groups of four to six high-frequency words likely to be familiar to the learners ( all high elementary or low intermediate level). Sometimes the words all belonged to the same lexical field (for example, desk, chair, lamp, computer), and sometimes only the two last words had a semantic connection (e.g. sunny, excited, bumpy, hot, cold). For the target items, Field changed the onset of the last word only, making a similar word which “didn’t fit” semantically (i.e. sunny, elegant, bumpy, hot, bold). Foils, or examples where the last words were not changed, were mixed in, giving learners examples of target patterns. Fields wanted to see whether what learners expected to hear would override what they actually heard. And what he found was…….the opposite! Out of 18 example sets, listeners only “re-interpreted” one answer to fit their expectations. Well, now, that’s discouraging if you thought you had a good case in favor of top-down processing skills. Field, however, re-assessed his test design and found an important flaw: he had chosen to change the onset–rather than middle or the offset of the target words. Native speakers attach great importance to the initial sound of words, and Field (cheerfully?) proclaims that this in itself is an interesting result, since it clearly shows that L2 speakers also pay great attention to the onset of words, to their obvious advantage.
As an aside, I do hypothesize that John Field is a cheerful person. I listened to a podcast of him speaking about his reception of a prestigious award; he describes himself as “gobsmacked” upon hearing the news, which I take to mean pleased and surprised. The surprise implies a very appealing modesty and genuineness. I wouldn’t mind having tea (or a beer?) with him.
So, on to the second experiment, which produced more expected, but this time positive results. This next test consisted of semantically constrained sentences marked by “acceptable but unpredictable” final words. For instance, “I couldn’t listen to the radio because of the XXX”. One might expect an appropriate answer to be noise, but the actual spoken answer is boys. Again, Field was curious to see whether learners would choose a more predictable answer over the what they actually heard, which was less appropriate or expected. This time, results showed that words in seven out of the twenty items were substitutions of “expected” words rather than the words listeners actually heard. But once again, there were unexpected results as well: an analysis of the words chosen showed that test takers tended to change to a word whose onset was similar to the target word (for instance, Field predicted that noise would be the preferred choice in the previous example about listening to the radio; learners, however, preferred to substitute voice, which shares a similar labial onset sound). Aha–further evidence of the importance of onset sounds, concluded Field.
Finally, on to the last and most interesting experiment. This time, subjects were presented with a sentence designed to provide a meaningful context for the very last word….which was a potentially unknown low-frequency lexical item. For instance….”They’re lazy in that office; they like to shirk“. Field hypothesized that learners might be tempted to substitute a more familiar and phonetically similar word, even if the context didn’t fit. In the case of the office question, learners might choose work, a word that didn’t necessarily make sense, but that sounded similar and was associated with offices.
So what happened? Well, as Field reports, “results were striking”. He found that 33.31% of the listeners did not accept the acoustic evidence of unfamiliar words, instead deciding on familiar and phonetically similar ones that often had very little connection with the sentence context. Eliminating those listeners who left questions blank, the number went up to 42.39%. Again, words that learners chose were often not those that Field had predicted. And not only did learners often choose semantically inappropriate words, but they sometimes chose words from different word classes! The trend, he found, did not vary according to individual either, as only one of the forty-seven subjects did not use this strategy at least once. And learners chose to match heard words with known words in each of the 20 test questions. There was no case of a question where a learner had not used this particular strategy. A significant result, right?
It all depends on how you look at it. Field admits that if the data for the third test (not adjusting for those who left questions blank) were to be calculated, it would show that learners’ choice of inappropriate words was not statistically significant (i.e. it was little better than chance) by quantitative standards. But, as he makes clear, such a pronouncement does not take into account “the most striking fact about the figure–that it was achieved despite the evidence of the listener’s ears and, in many cases, to the evidence of the contrary”. Learners, it seems, can sometimes ignore both top-down (the logical contextual choice) and bottom-up (the word as heard phonetically) evidence when faced with a difficult unknown lexical item. The tendency to substitute a known for an unknown word, then, is an entirely different phenomenon which he calls a lexical strategy.
In the end, Field provides insights rather than answers, and that, for me, is the beauty of this article. He hypothesizes, considers his results accurately and honestly, revises and expands his ideas, and finally sheds some real light on the process of listening for L2 learners. A must-read for all TESOL students and language EFL/ESL practitioners, Field’s article is also an interesting and accessible read for language lovers in general. Treat yourself to an enjoyable afternoon with John Field, and you’ll come away a little wiser.