2. Literature Review 17
2.2 Aspects of Production and Decoding of Spoken English 17
2.2.2 Intonation Units and Communicative Features 24
All spoken language is composed of different sized ‘chunks’. These speech units have been focused on by different researchers and are amenable to different types of linguistic analysis. The investigation in this section starts with some short intonational units which are characterised by having a tonic stress within each unit, i.e., the tone unit, the information unit, the intonation group and the intonational phrase. In addition, some other longer units which consist of more than one prosodic pattern, such as the paratone and the TM intonation model, are also outlined.
A. Units characterised by presence of tonic stress a) Tone Unit
For analysing spoken language and its communicative value, the minimal intonational unit – the tone unit, is proposed and systematically analysed by Brazil (1997). The tone unit is ‘the stretch of language that carries the systemically-opposed features of intonation’ (p.3), which is similar to ‘sense groups’, ‘breath groups’ and ‘tone groups’ (p.5). According to Brazil, there is only one indispensable tonic syllable (which carries the operation of salient pitch movement) within each tone unit. This is located on the last prominent syllable, although the criteria for assigning the tonic syllable ultimately depend on speaker decisions.
Depending on the communicative event, either a P tone (proclaiming – providing new information) or an R tone (referring – indicating shared information) is assigned by the
interlocutors (Brazil, 1997, pp.68-73). Brazil defines the values of the P/R tone choices, from the social interactive point of view. In general, a proclaiming tone shows a distance between the speaker and the listener, and a referring tone normally expresses a co-operation between the interlocutors (p.82). Brazil emphasises that the P/R tone choices depend, among other things, on whether the speaker wants the speech to benefit himself or his listener (p.95).
Pause phenomena also play an important role in orientating a discourse and are defined as ‘dummy carriers’ of the tone (Brazil, 1997, p.139). Given the imperfect nature of tone units (i.e., the ‘incomplete tone unit’), Brazil acknowledges that ‘pauses are always treated as tone unit boundaries’ (pp.147-48), and the boundary of the tone unit often coincides with a grammatical unit, i.e., the sentence or the clause (p.150).
The tone unit is regarded as a ‘communicatively functioning whole’ (Brazil, 1997, pp.18-19), and the communicative value of intonation, as Brazil emphasises (1995, p.240), is associated with the purpose in which it occurs in a certain here-and-now context of interaction.
This stretch of sound continuum – the tone group – is also analysed by Brown (1990). She tends to break a stretch of speech by ‘immediate constituent[s]’ (p.93). The boundary most likely occurs between subject and predicate, and might be found within a long subject or a long predicate. The criteria for this division are not only pitch, but also lengthening of the final syllable and pause. In general, in spontaneous speech, the division into tone groups is less clear, but at least the tone group delimits major syntactic constituents (p.102). While Brown also locates the tonic syllable on the last
lexical word, she points out the phenomenon of tonic shift to the left. This would suggest a speaker-dictated prioritisation of discourse function over syntax.
Roach (2000) also discusses tone units from the perspective of the analysis of intonation patterns. He points out that the smallest tone unit can be of only one syllable. The tonic syllable carries the pitch movement of the tone unit, yet, when there is a tail following, the pitch movement will be extended (pp.167-73). In addition to silent pauses, ‘sudden change[s] of pitch level’ and rhythm are also defined for the identification of tone unit boundaries (p.178). He also agrees that usually tone units accord with grammatical units. A more generalised view on the position of the tonic syllable is given by Roach (ibid.) who claims that it ‘tends’ to occur on the last lexical item; on a few occasions it can be earlier due to the speaker’s decision on prominence (pp.194-95).
b) Information Unit
As mentioned above, an important function of intonation is to mark off new information from given information, thus an ‘information unit’ is proposed by Halliday (1967). The information unit is directly associated with the realisation of a ‘tone group’. Each tone group ‘represents what the speaker decides to make into one unit of information’ (Halliday, 1970, p.162). The prominent syllable in this group, i.e., that which carries the ‘main burden of the pitch movement’, is called the tonic syllable (ibid., p.4). There is one and only one tonic syllable within each tone group. According to Halliday, each tone group must begin with a stressed syllable, or a ‘silent ictus’ when the initial foot is unstressed. Although no criteria are given for the division of tone groups, Halliday (1967, p.203) emphasises that the tone group, due to its phonological characteristics, must occur within the limitations of rhythm. Since information is mostly ‘mapped on to
the clause’ (1970, p.127), the clause is posited as the common unit of the information group.
The tone group is also analysed by Laver (1970, pp.68-69), from the speech production point of view, in which it is stated that it is ‘handled in the central nervous system as a unitary behavioural act’. The average length for a tone group is about seven or eight syllables. The tonic syllable, according to Laver, is usually located ‘at or near the end of the tone group’. The pause is seen as the primary tone group boundary. The syntactic clause often coincides with the tone group. In agreement with Halliday, Laver also states that the tone group corresponds with a rhythm unit.
c) Intonation Group
Another type of unit – the ‘intonation group’ – is proposed by Cruttenden (1997). In his analysis, four-levels of stress patterns are present within an intonation group: primary stress, secondary stress, tertiary stress and unstressed (p.18). Given the unclear delineation of an intonation group (apart from pauses), three other external criteria for the identification of intonation groups are proposed by Cruttenden, i.e., an anacrusis, a lengthening of the final syllable, and the change of pitch contour on an unaccented syllable (pp.32-34). Cruttenden points out the problems encountered when using pauses as the criterion, and when there is no clear pitch movement to indicate tonic stress. This indicates that the intonation group is in fact an abstraction, ‘a theoretical construct’ (p.37). While Cruttenden uses ‘nucleus’ to describe the tonic stress within an intonation group, he also argues that the contour following the nuclear accent always expresses the most salient intonational meaning of an intonation group (pp.44-49).
Intonation groups more commonly align with large syntactic constituents – more commonly clauses than any other grammatical unit (Cruttenden, 1997, pp.69-73). However, due to speaker-determined performance, the division of an intonation group is unpredictable. According to Cruttenden (ibid.), normally the average length for an intonation group is about five words, with very few groups over seven words (p.72). Given that the focus of the intonation group is the tonic syllable, Cruttenden argues that there are quite a lot of exceptions to its final lexical item placement (p.75). The reason for assigning and for the interpretation of different tone choices depends firstly on the syntactic type, and also it depends on the relationship between the speaker and the listener (p.104). Cultural factors are also emphasised, in that a high pitch register seems to be associated with certain cultural conventions (p.124). Like Brazil, Cruttenden labels the contrastive, confirmative information produced by falling intonation as ‘closed’, and the non-assertive, continuative associated with rising tone patterns as ‘open’ (p.163).
d) Intonational Phrase
Another term, defined by Levelt (1989) from the melodic point of view, is the intonational phrase. There is only one nucleus in each intonational phrase, and when there is only one pitch accent (there could be more), then the main pitch contour rests on the last lexical head. According to Levelt, an intonational phrase consists of ‘one or more phonological phrases or metrical groups’; ideally it spans about 2 seconds and ranges between 1 and 3 seconds (p.386). An intonational phrase is usually a sentence unit surrounded by grammatical pauses (usually more than 200 milliseconds). It can also be isolated by syntactic, semantic, or ‘operational’ definitions, as long as it is under the speaker’s control and carries one of a set of tones. Two different functions in one
intonational phrase are posited by Levelt – a qualifying and modifying function on the pre-nuclear part, and an intonational focus function (expressing meaning) on the nuclear pitch movement of an intonational phrase.
B. Units with more than one prosodic pattern
All the analysis given above is focused on the identification of the pitch movement and intonation functions within one intonational unit. Given the review earlier, many factors contribute to the isolation of intonational units, and the criteria are so complex and variable that in no case can the boundaries of intonational units easily be identified. Therefore, not all linguists agree how natural flowing speech can necessarily be divided into small intonational units, especially in the rapid continuum of informal speech. Some researchers abandon intonational units, and adopt different units for their different analytical purposes, as examined below.
a) Paratone
The ‘paratone’ is proposed by Brown and Yule (1983b). In order to analyse how speakers organise larger chunks of discourse and how the topic is smoothly changed between the interlocutors in natural conversations, Brown and Yule divide large pieces of discourse up into a series of small structural units, each on a separate topic, called paratones (p.100). Paratones, as Brown and Yule state, function as ‘spoken paragraph[s]’, and mark the beginning of new topics. The tone patterns in these units tend to display similar contours. A high pitch value usually occurs at the onset in order to draw the listener’s attention and mark the start of the turn. Brief pauses tend to be embedded in the middle of the paratone, and a gradual declination in pitch level or a long pause towards the end of the paratone indicate the closing remarks and a readiness
to hand over the turn. The paratone is more appropriately associated with topic structure, rather than with small individual pitch features.
b) TM Intonation Model
Another unit, called the TM intonation model, is employed by Tatham and Morton (2005) in the analysis of speech synthesis. In this unit four aspects are marked: lexical stress, a syntactic phrase within an intonational phrase, sentence and intonational phrase boundaries, and sentence focus on a single syllable. Only two levels of each prosodic characteristic are identified in the TM intonational model, which is different from Cruttenden’s (1997) four levels of stress, as mentioned earlier. Based on Tatham and Morton’s speech production theory, expression is a ‘central’ or ‘enveloping’ characteristic of utterances (p.112). In this model, they focus on identifying intonational functions by fitting the segments into a prosodic wrapper (p.287).
All the above listed different units identified by various researchers are based on the specific purposes of their linguistic analysis, and all of them are related to pitch movement and fit the prosodic intonation contours of English. However, there is a lack of agreement among linguists on the most useful unit of analysis in speech. In order to analyse the phonological realisation of natural speech, a minimal production unit – ‘flow unit’ (Campbell, et al., 2006) – is proposed and adopted in the current study, as detailed in Section 7.2.2.