Gentlemen, listen to me slowly.
Samuel Goldwyn (1882–1974), Polish-American film mogul
We now turn our attention to larger units of speech. We consider how an expert listener manages to identify grammatical patterns in the words that have been decoded. We also consider what information a listener obtains from the way in which intonation shapes the speech signal.
There is logic in discussing syntax and intonation together, because, as will become clear, they are quite closely interconnected.
These two systems play a part in decoding but also contribute to meaning building. To give a simple example, if a speaker says we’re leav-ing, the listener recognises the grammatical pattern that has been used (a matter of decoding) but can only fully understand it by relating it to con-text and co-con-text (does it refer to now or to the future?). Similarly, features connected with intonation often assist decoding by marking the end of an ‘idea unit’, but they also add to overall meaning by highlighting the most important element in the unit or by indicating the mood or attitude of the speaker. We will consider both types of function in the present chapter.
11.1 Syntax and decoding
11.1.1 Recognising syntactic units First language processing
Having found matches for a group of words, a listener has to trace the grammatical structure that binds them together. This operation is known as parsing. It depends partly upon the listener’s ability to monitor a structure as it develops and partly upon her ability to recognise when it is complete. Once she realises that the structure has come to an end, she turns the group of words into an abstract idea. From this point onwards, she may or may not be able to recall the exact words that the speaker used, but she can still report the information that the speaker conveyed.
How does a listener recognise the point where a syntactic unit (a clause or sentence) finishes? Clearly, L1 listeners draw upon their knowledge of the grammar of the language and of the most frequent syntactic patterns.
They know, for example, that a particular verb requires an object and that the sentence will not be complete until the object has been men-tioned. However, equally important are the cues that the speech signal provides.
One such cue is found at the points where a speaker pauses to plan what to say next. We tend to assume that informal everyday speech is totally unplanned, but that is not the case. A sentence has to be assembled, using the speaker’s knowledge of vocabulary and organis-ing the words to conform to the rules of grammar. Speech thus consists of alternate spells of planning and uttering. Speakers usually plan their utterances as complete phrases or clauses, so a short pause in connected speech often signals the beginning of a new grammar structure. Alterna-tively, the speaker may use a meaningless filler (I mean, as I was saying, so to speak) in order to buy time for planning. Pausing and these fillers thus offer the kind of marker that commas, semicolons and full stops provide for the reader. Both carry an advantage for the listener in that they provide a brief remission from decoding, during which she can rapidly review what has been said.
However, some caution is needed here. Pauses for planning need to be distinguished from hesitation pauses where the speaker has forgotten or changed his mind about what he intended to say, or is having trouble in retrieving a word. Hesitation pauses are quite distinct. They are much less consistent in length than planning pauses, often occur in the middle of an incomplete intonation pattern and sometimes include fillers such as erm or er. A skilled listener distinguishes them from pauses that mark syntactic boundaries and edits them out.
Assistance in locating syntactic boundaries also comes from intona-tion. Natural speech falls into small groups of words which are linked together by the movement of the speaker’s voice.1An intonation group of this kind often corresponds to a grammatical unit – a phrase, a clause or a sentence. It thus indicates to the listener that certain words are connected syntactically. One count (reported in Cruttenden, 1986: 75) suggested that about 40% of groups correspond to clauses, though this may depend very much on the speaker and the level of formality. Into-nation groups also serve to mark out longer adverbial expressions (at ten o’clock on MONday, in front of the SUpermarket) or to divide a sentence into subject and predicate.
Sometimes intonation groups are clearly demarcated. For example, the end of the group might be indicated by a fall in pitch, a downwards movement of the speaker’s voice. But when these cues are not present,
1
the listener can usually identify the group by reference to its focal point, a stressed syllable which often occurs at or near the end.
Second language practice
It is helpful to give learners practice in identifying syntactic patterns that are characteristic of the language they are acquiring. Table 11.1 gives examples of possible exercise types. Learners can be told that pausing assists them to decide where one syntactic pattern ends and another begins, though it should be pointed out that pausing may not always be present. Intonation is more problematic to demonstrate, as intonation groups vary greatly both in their length and in the ‘tunes’ they adopt.
Table 11.1 Exercises in recognising syntactic units
Awareness raising. Class listens for focal syllables and for pauses in a short authentic recording (30 secs). They listen again, and mark the stressed words and the pauses on a transcript. Class discusses the role of stress and pausing in marking out phrases and clauses.
Hesitation and juncture pauses. ‘Mark the pauses.’ Class listens to a short authentic recording, then marks pauses on a
transcript. Class decides which are hesitation pauses and which indicate syntactic boundaries.
Phrase patterns. Class listens to a short recording for general understanding. Teacher plays a second version with pauses or bleeps inserted at phrase boundaries (i.e. after subject, after verb, after object, etc.). Class reports on improvement in understanding.
Dividing into phrases. Class listens to a short recording. They listen again with a transcript and divide the sentences into phrase and clause groups.
11.1.2 Online parsing First language processing
The parsing operation is rather more complicated than the account so far has suggested. There is evidence that listeners do not wait until the end of a sentence in order to start working out what a speaker is saying, but decode speech at a delay of as little as a quarter of a second after it has been uttered. (A quarter of a second is about the length of a syllable in many languages, suggesting a special role for the syllable in decoding.)
This means that listeners have to trace grammatical patterns in running speech while the patterns are still unfolding. They can form expectations about the shape a sentence will take; but they may have to revise them as they hear more and more, and they cannot be satisfied that they have achieved a final version until the sentence is complete.
So, for example, a listener might start off with the assumption that the first noun phrase she hears (e.g. the hardened criminal) is the ini-tiator of an action; but when she goes on to encounter a passive verb (was arrested), she has to reverse this assumption. This view of parsing conforms very much to the picture of listening painted in Chapter 10:
a process in which listeners constantly form and revise hypotheses until sufficient evidence has accumulated for them to be certain about what the speaker intends.
When parsing in this online fashion, an expert listener makes use of certain cues in the input, which help her to narrow down the possible options. Among them are:
Probability. Long experience of listening to the language makes an L1 listener highly sensitive to the way in which certain word forms occur together or signal what is to come. Thus, the presence of the or a indicates that the speaker is beginning a noun phrase like the fast train, and that either an adjective or a noun is likely to come next. If that noun phrase occurs at the beginning of an utterance, experience dictates that it is probably the subject of the sentence; following the verb, it is most probably a direct object. While listening, the L1 listener is thus capable of forming cautious hypotheses as to what will come next and as to what the overall structure of the utterance will prove to be.
Chunks. Also of assistance are clusters which speakers store and produce as pre-assembled chunks. These include collocations (heavy smoker, light rain) or short question initiators such as do you ([ə]) or is it ([zt]); but they also consist of complete syntactic structures ( . . . can’t have done, I don’t mind, I couldn’t agree more, what I’m trying to say is) which the listener does not need to divide into constituents because they can be recognised as a single unit.
The verb. The structure of a sentence is often decided by the verb that is chosen. If a speaker elects to construct a sentence around the word PUT, then she has to follow the pattern
PUT+ [thing that is placed] + [where the thing is placed]
The last element cannot be omitted (∗Please put the money is unac-ceptable). From the point of view of the listener, knowing the pattern