• Tidak ada hasil yang ditemukan

Finding matches for inconsistent input

Dalam dokumen Listening in the (Halaman 175-178)

10 Decoding: sounds, syllables and words

10.1 Finding matches for inconsistent input

Let us first consider a major question left unanswered in Chapter 9.

If the input to listeners is so variable, how on earth do they succeed in matching it to their knowledge of the sounds and words of the language?

10.1.1 Listening as a tentative process

One possible conclusion from the evidence is that much listening, whether in L1 or L2, is approximate. If the input is so variable, then listeners may need to form tentative matches on the basis of the avail-able evidence and to confirm or change them as they hear more and more of the utterance. The idea that listening is based upon forming and testing hypotheses will recur at many points throughout this chapter and the next.

It accords well with the interactive account given in Chapter 8, which holds that the matching process does not rely upon a single source of evidence (e.g. just a series of phonemes). Instead, the listener is seen as constantly weighing many different pieces of information against each other in order to establish what has been heard. To repeat the veshtables example, the listener might draw not only upon matches made at the level of the phoneme but also upon matches made at the level of the syllable (vesh?) and the word (vegetables) and upon clues provided by the surrounding text (fruit and . . . ). The goal is to find the ‘best fit’

rather than a fit that necessarily forms a 100% match with what was heard. In this way, the listener is able to overcome the many variations acquired by the citation forms of words when they occur in connected speech.

Psychologists often portray this weighing of evidence in terms of acti-vation. Think of activation as a kind of electric current that has to reach a certain strength in order to light up a light bulb. Let us assume that a

listener hears the sequence [matren] and is unsure whether to match it to my train or might rain or even might train. All three would be activated in her mind, i.e. would receive a certain level of electrical cur-rent. As the sentence proceeds, the current for one or more might be boosted by subsequent evidence (e.g. if the next word was came, heavily or students) and the current for others would weaken. When there was sufficient evidence for one interpretation, its light bulb would turn on and the listener would be satisfied that a match had been achieved.

If listening is such a tentative operation, it requires us to keep quite a precise trace of the sounds that reach our ears, in case we have to go back and reinterpret them. It also means that we have to be very flexible in our ongoing interpretation of what we hear. There are difficulties here for the L2 listener. She may find it harder to store a temporary record of the sounds that were uttered because her knowledge of the L2 sound system is so much more limited. She also has to lend much greater attention to decoding, which must limit her scope for rethinking her early matches.

Furthermore, the process of weighing evidence is complicated by her limited vocabulary, which may mean that she is not aware of all the possibilities that need to be considered, and by her uncertainty about how accurately she can apply the phoneme values of L2.

Small wonder if, instead of forming several hypotheses, a less experi-enced L2 listener often chooses one and clings to it, ignoring subsequent evidence that is inconsistent. The light bulb for might rain might light up prematurely and cause the listener to ignore the disambiguation provided by leaves at six.

10.1.2 The question of how we store language

Another explanation considers not what the listener does but how she stores sounds, words and grammar patterns in her mind. It should be noted that this explanation is not necessarily an alternative to the one just given; we might choose to treat it as complementary.

It seems to be taken for granted by some commentators on second language learning that a listener carries a ‘pure’ version of a phoneme or a citation form of a word in her mind, against which she matches what she hears. The problem with this assumption is that one then has to assume that the listener possesses some means of editing out all the variation that occurs in natural speech, in order to make the match. She needs to edit out not only the voice and accent of the speaker but also features described in Chapter 9 such as assimilation, reduction, etc. This would appear to be a very complex operation.

A more recent view is that we retain multiple records of the sounds and words of a language in our memory, and match new instances against

them. To give the example of the word actually, a native English listener would have at least the seven different versions of it shown in Table 9.6, all linked together in the mind as representing the same word. The listener would also have records of the word said in a range of different voices and different accents. Hearing a sequence of sounds, she would make a direct link to one of these stored memories without having to edit anything.

This account of how we store sounds and words might seem implau-sible and wasteful. But it is a solution to which speech scientists are increasingly turning.1 It agrees with evidence that the brain is capable of holding much more information than we ever suspected. It explains how a listener is able to make allowances for the enormous variations between speakers’ voices. It also suggests how, over a period of time, we come to understand those who speak with different accents from our own. We do so by a process of laying down more and more traces of voices speaking in those accents until we have sufficient points of reference to be able to understand them.

If an exemplar view of this kind becomes widely accepted, there are implications for L2 listening instruction. Teachers will need to rethink the received wisdom that learners must spend a lot of time internalising the sound system of the target language in some kind of abstract form. They will also need to recognise the limitations of learning and storing words in their citation forms. A primary requirement will be for the listener to encounter the same words in a wide range of contexts and voices.

We can then conclude that exposure to lengthy listening passages (as in the comprehension approach) does assist the listener in that it establishes multiple traces of particular words and patterns. But, given the limited amount of time available to the instructor, we should also conclude that it is important to support this extended approach with more focused practice activities. One useful innovation in the early stages of L2 listening might be to assemble examples of the same groups of words uttered in different circumstances and at different speeds by a number of different speakers.

Having discussed some general principles in this section, we now move on to look more concretely at how listeners make matches at several different levels. We will consider matching at phoneme level (Section 10.2), syllable level (Section 10.3) and word level (Section 10.4). The format in each case will be the same. There will first be an outline of some current ideas about the processes that a skilled listener employs. We will then consider the ways in which they might pose a challenge for the second language listener. There will be suggestions as to the implications

1 Usually in conjunction with an interactive view of how a final match is made.

for the classroom and concrete examples of exercise types that a teacher might use.

Dalam dokumen Listening in the (Halaman 175-178)