Some neglected aspects of intonation
V. J. Cook
In H. Dechert & M. Raupach, Temporal Variables in Speech, 1980
The majority of the papers presented to this workshop have been within the discipline of psychology. This contribution, however, tries to relate other areas to the study of speech production, namely work on describing and teaching English intonation. The starting point in terms of intonation is the kind of analysis suggested in Halliday (1967a). Halliday described three systematic choices that confront the speaker. One is how divide the utterance up into tone-groups; a second is where to put the major change of pitch, the "tone"; the third is which tone to use. This analysis is useful for the study of speech even if it does nothing more than indicate that intonation is a complex of several factors, not just the rise and fall of the voice.
For most of the approaches that have been used to intonation in the literature of speech production have been one-sided. Commonly, in fact intonation has been ignored. Equally commonly people have relied on the concept of "normal English intonation" as, for instance, in O'Connell, Turner & Onuska (1968). The parallel concept of normal stress has attacked by Schmerling (1974) and the existence of normal English intonation can also be challenged. One reason is the question of regional intonation.
While the neutral intonation for decontextualised declarative sentences in Southern British is usually claimed to be a fall, without leaving the British Isles, in Belfast the neutral intonation is said to be a rise (Jarman & Cruttenden, 1976). Specifying "normal intonation" in an experiment is of no use if we do not know the background of the speaker. A second reason for distrusting "normal intonation" is concerned with the location of the tone. The reason for putting the tone on one syllable rather than another is usually felt to reflect the structure of discourse. Within Halliday's framework, the location of the tone is seen as contributing to the cohesion of the discourse, particularly by drawing attention to new information, though this is queried by Goodenough-Trepagnier & Smith (1977). A similar point is to claim that the location tone depends on semantic focus (Cutler & Foss, 1977). The location of the tone then depends upon the context - what the participants already know about the subject and what they have already said. Mostpsychological work with speech has dealt with sentences and has deliberately ignored the effects of context, including intonation. Perhaps this is the right approach and intonation is psychologically irrelevant.
However, there is a body of evidence from other types of experiments that intonation is far from irrelevant. This starts with Neisser's of rhythmic groups in active verbal memory (Neisser, 1967) and includes Glanzer (1976) who showed that tone-group division was connected with memory for related nouns in word-lists, O'Connellet al (1968) who found that intonation was more important with structured than with unstructured material, and Bower & Springston (1970) who found that grouping by pauses and "vocal intonation" aided recall. Certainly division into tone-groups, and probably also location of the tone play an important part in speech processing. Any experiment with speech, whatever it is ostensibly about, needs to control and specify some aspects of intonation.
But intonation also needs investigation in its own right. It may be that intonation is central to the planning of speech production (Laver, 1970); it may also be important to language acquisition (Du Preez, 1974). As the contributions to this workshop show, some aspects of intonation are indeed being studied; but the link between intonation and context is under-explored. From a linguistic point of view Brazil (1975, 1978) has been looking at the intonation of discourse. One of his ideas is called key. This adds another choice to the speaker, whether to use a high key, medium key, or low key. Usually the progression through an utterance is from high to low, each new "paragraph" starting the cycle up again. Thus a switch to high key marks the tone-group as contrastive, a switch to low key as an equivalence between items in successive tone-groups. Each tone-group is measured against the one preceding to see whether it is higher or lower in key. Aspects of intonation and discourse must be taken into consideration in studying speech; the place of the sentence in the utterance may play a crucial part in its production.
The rest of this paper looks at the place of intonation in the structure of one type of discourse, ordinary conversation, particularly in terms of tone. Conversation is a process of give-and-take in which the participants have certain moves that they are allowed to make. In a scientific sense, little is yet known of this. However those who teach foreign languages often have to make leaps beyond our present state of knowledge if they are to do justice to their students' needs. The scheme that is used here is a leap of this kind embodied in a book called Using Intonation (Cook, 1979). This is a teaching scheme for intonation that brings out some of the aspects of intonation that the psychological experimenter will eventually have to take into account. The system of description employs five tones: high and low rises, high and low falls, and a fall-rise. More sophisticated versions of the same type of analysis can be found in O'Connor & Arnold (1961) and Cook (1968). The structure of conversation is analysed into seven categories and the most common tones for each of these is given. Obviously this scheme covers only certain intonation patterns found in Southern British English but at least it shows what kind of analysis of intonation is going to be necessary.
The first category is checking what people say. It is vital in conversation that the participants are talking about the same topic. Checking usually involves a rising tone, either with repetition of part of the other speaker's remark - "I went to Bristol yesterday", "You went to Bristol?" - or with question words or special checking forms - "You went where? Where did you go?", "Sorry?". The second category is asking for information. This implies a potential shift of roles in that the listener can take over the active part of the conversation. It is usually believed that tag questions and Yes/No questions have low rises - "Have you been to Germany before?", "It's nice, isn't it?" - while question-word questions have falls - "How much is it?" - except perhaps when introducing a new topic of conversation - "What's the time?" The third category is answering questions with short answers. Falls may be used, the low fall suggesting the speaker doesn't want to go on talking about that topic - "Do you like beer?", "Yes." - the high fall sounding more interested in the topic - "Can you speak Spanish?", "Yes." The fall-rise suggests the speaker is doubtful - "Can you dance?", "Well".
The fourth and fifth categories are stating things more or less positively. Saying something positively involves a fall - "I think he's wrong" and a tag question with a fall expects brief agreement from the listener rather than allowing him to take over the active role - "It's hot today, isn't it?" Saying things less positively uses either a fall-rise to show doubts or reservations - "I like the "breakfasts in the hotel." - or a low fall preceded by low syllables suggesting reluctance to continue speaking "That's enough." - or a low rise suggesting the speaker is not committing himself - "The food's all right I suppose." Category six is reacting; listeners are expected to show that they are paying attention by showing a reaction of one kind or another. Usually this involves falls, the height of the fall showing the degree of enthusiasm or astonishment - " Did you?", "Oh." - or a low rise urging the speaker to continue - "Were you?" Finally category seven is getting things done. The different tones appear to relate to the speaker's role. If he is authoritative he may use a low rise - "Put your hands up."; definite, a low fall - "Don't move."; friendly, a high fall - "Give me the money."
This simplified model begs certain questions; whether intonation goes with grammatical structure, which has been queried in Kenworthy (1978); and whether it conveys attitudes. Nevertheless it highlights certain aspects that are relevant to the study of speech. One is simply that it is dangerous not to control intonation in any language experiment; probably most of the classic psycholinguistic experiments would fail to replicate if their intonation patterns were sufficiently distorted. The second aspect is that intonation shows how sentences form part of discourse. There are relationships between sentences in terms of many aspects of cohesion, lexical, grammatical, and intonational. The study of speech will have to take into account higher levels of organisation than the sentence.
Arising from this, perhaps the most important aspect is that speech is a meaningful activity that involves at least two people. Language has a purpose. Except in psychological experiments we are not forced to speak but talk when we want to. If we cut speech off from discourse and from purpose, our experiments tell us as much about speech production as would experiments into breathing which studied the behaviour of subjects in a vacuum. The disciplines of linguistics, applied linguistics, and developmental psycholinguistics have taken this crucial step towards looking at language as a process of interaction between human beings. If it is to progress, the study of speech production and comprehension must follow their lead.