In attempting to develop interpretive, sociolinguistic approaches to the analysis of verbal strategies, this book has touched on a number of recurrent themes. The objects of study are automatic, context and time bound inferential processes, not readily subject to conscious recall, embedded in oral exchanges which until the advent of modern electronic technology were not accessible to detailed investigation. In order to clarify precisely what it is that is being investigated concrete examples of situated talk have been transcribed and analyzed in such a way as to reveal the working of phonetic, prosodic, formulaic and other contextualization cues in generating the perceptions of discourse coherence on which interpretation must rest. In this Postscript I will review some of the theoretical issues raised by these examples and attempt to show how, when seen in interactional perspective, they can be integrated to lay the foundations for a unified program of research on human understanding.

The study began with a brief historical outline of developments in linguistics that led to the recognition that linguistic processes are basically cognitive in nature. The notion of cognitive processing, which argues that human understanding rests on meaning assessments in which physical reality is selectively perceived, transformed and reintegrated with reference to pre-existing background knowledge, is by now generally accepted. First illustrated in Saussure's concepts of opposition and relationship and in Sapir's phonemic principle, it has been generalized to apply to grammatical, interpretive and cultural phenomena of all kinds.

Both Labov and Bailey agree that it is the ‘facts of variation’ which are central to linguistic theory. If a reasonable account of language change depends on the possibility of describing orderly differentiation within language, as they maintain, the problem is to decide which of the competing variationist theories, i.e. the quantitative or dynamic paradigm, is best able to incorporate the facts of variability within a linguistic description. There is, however, disagreement between the two over the set of facts a linguistic theory can properly choose to address, and what a formal model of grammar which incorporates the observation of variability looks like.

Bailey has emphasized in a number of places that linguistics must make dialectology central, but that ‘the job of linguists is linguistics’ (Bailey 1969a: 118).

Brief description of the proposed study

Thus far, sociolinguistics has concerned itself primarily with the analysis of synchronic variable speech data. If, however, sociolinguistics is to reach the final goal set for it by Hymes (1974a: 206), namely, that it should preside over its own liquidation, then clearly sociolinguistics should itself be an integrative mode of description. I take Hymes' comment to mean that the subject matter of linguistics should not be confined to the study of the conceptual function of language, but should also include its social function or communicative use. What he argues for, then, is a general theory of language which comprises both aspects. Such a theory would dispense with the need for a separate sociolinguistic theory; in other words, sociolinguistics would be ‘redundant’ (Labov 1972a: 183).

The chances of sociolinguistics becoming such an integrative discipline are, in Hymes' opinion as well as my own, uncertain. Furthermore, I question whether a sociolinguistic theory in these terms is possible. My starting point is the assumption that if sociolinguistics is to meet the challenge given by Hymes, it must move beyond the treatment of synchronic phonetic and phonological data to a more general body of linguistic data. My study attempts to extend the application of variation theory from the domain of synchronic phonological variation to the study of a problem in historical syntax.

The origin of the relative clause in the Germanic languages: a problem of general syntax

In his discussion of the relative clause as a problem of general syntax, Benveniste (1971) has commented that a comparison of relativization in languages cannot be based on formal elements alone since there are not always any comparable units; it must instead be approached from a functional viewpoint. He observes that the typical Latin construction with the relative pronoun qui governing a verbal clause has been taken to be the model for all relative clause constructions.

Nowadays, we are well aware of the dangers of adopting a Graeco-Latin point of view when dealing with historical syntax, and we avoid ‘forcing’ linguistic data into already existing models of description. However, the influence of Greek and Latin models on some of the older scholars of comparative syntax has been so pervasive that it deserves some mention. For example, quite a number of scholars (e.g. Curme) have commented that in early Germanic two basic types of sentence structure or relations between structures can be distinguished: parataxis and hypotaxis. These terms represent relative notions which are often somewhat vaguely used in the literature. In both parataxis and hypotaxis successive clauses may occur in sequence with no formal connecting link between them. Some have claimed that parataxis is a property of ‘primitive’ languages and is the simpler construction of the two because of its presence in the early stages of many languages.

In the preceding chapter I attempted to apply variable rule analysis to a specific test case, variation in the relative marker system, to see if there was any justification for the claim that the Cedergren–Sankoff program can demonstrate a convergence between formal rule schemata and quantitative data. The results of the analysis were indeterminate and difficult to interpret, but they were indicative of problems in Labov's theory of variation and change. My data might be thought to be in some sense atypical and consequently my findings to be atypical also, but I will argue that this is not the case. In this chapter I focus on the empirical and ontological status of some of the so-called decisive solutions which have been proposed in sociolinguistic theory to show that there are indeterminacies as well as inconsistencies in a number of these analyses. I take up first Labov's (1969) analysis of the copula in BEV as a case in point. I mentioned in Chapter 2 that it was in this paper that Labov argued most forcefully for a program of empirical research which would produce decisive and correct solutions. I will demonstrate that the analysis Labov presents is not necessarily the ‘right’ one based on the quantitative data.

Labor's analysis of contraction and deletion of the copula in BEV

Labov decided to treat contraction and deletion of the BEV copula as two separate rules which were ordered so that contraction preceded deletion.

In this chapter I discuss the contribution which variable rule analysis makes to my study of the relative system, and show how some hypotheses about the types of rules needed to describe the variation can be tested using the Cedergren–Sankoff variable rule program. I will be particularly concerned with examining the kinds of arguments which have been advanced in favor of the claim that the program (and variable rule analysis) gives new and reliable evidence for judging certain linguistic issues.

Cedergren and Sankoff's (1974) work on the probabilistic component of variable rules has had a major impact on quantitative analysis in sociolinguistics. The ramifications of this shift in emphasis from frequencies to probabilities are enormous; yet, they have scarcely been treated cogently in their relation to the construction of a sociolinguistic theory which claims to have an empirical base. Labov (1975a: 228), for example, has maintained that the ‘new tools of probability theory’ give us a ‘dramatic increase in power and perception’. The claim about increased power is true, though, ironically perhaps, not in the way Labov seems to think (cf. Romaine 1981a).

Probability theories might be considered more powerful with respect to their domain of application because, strictly speaking, they cannot be confirmed or disconfirmed by empirical findings.

A few years ago I became interested in claims made about the epistemological status of sociolinguistic methodology and, in particular, the so-called empirical foundations of a sociolinguistic theory. My concern with the nature of sociolinguistic methods and data grew out of some of the difficulties I encountered in trying to present a sociolinguistic description of some variables in Scottish English (cf. Romaine 1975). Some of the problems (e.g. continuous vs. discrete variation, levels of abstraction in the construction of sociolinguistic grammars, probabilistic rules) still bother me. I deal with them again here, but this time with reference to another descriptive problem, namely, variation in the relative marker in Middle Scots. This also leads to consideration of some new issues, e.g. the scope of sociolinguistic theory and the relevance of sociolinguistic methods to problems in historical syntax.

It will become apparent that I am using the term ‘sociolinguistic’ primarily in a narrow sense, i.e. to refer to the work which has derived from Labov (1966). I have concentrated on Labov's research program because it has been so influential; supported by a substantial body of empirical research, it represents one of the most concrete proposals yet made for a sociolinguistic theory. However, I also discuss Bailey's work; and I attempt to show that much of the controversy between the so-called quantitative (Labovian) and dynamic (Baileyan) paradigms results from a misunderstanding of the ontological status of some of the arguments and explanations which can be supported on the basis of sociolinguistic or variable data.

The relevance of sociolinguistics to syntax

In this chapter I examine the contribution which variable data in historical syntax can be expected to make to descriptive and theoretical issues, with special reference to my study of the relative system in Middle Scots. A number of linguists have argued that variable data can be used in support of abstract linguistic analyses, and that quantitative relations bear directly on, and provide evidence for, ‘deciding’ some important theoretical issues. If such claims are tenable, then the argument put forward by both the quantitative and dynamic paradigms, that variation should form the core of linguistic theory, seems well justified.

Carden (1972), for example, has claimed that native speakers' judgements about the grammaticality or acceptability of sentences are crucial for determining the nature and operation of syntactic systems. He argues that by seeking out the patterns of variation in acceptability in and among speakers, we can confirm postulated abstract structures, since variation in the data results from rule differences and differences in rule ordering. Likewise, Bailey (1970: 77) has maintained that an analysis of variable data allows us ‘to confirm fairly abstract syntactic structures, rather than cast them into doubt’.

It was, however, Labov who first programmatically outlined the questions which the study of variation (in the form of variable rules) could address, thus suggesting contributions it might make to linguistic theory.

In this chapter I will examine a number of possible linguistic factors which affect the realization of the relative marker (i.e. either as WH, TH or ∅):

1. type of clause – restrictive or non-restrictive

2. features or characteristics of the antecedent/head NP – animateness, definiteness, type of noun modification structure, e.g. determiner, quantifier, superlative

3. syntactic position/grammatical function of the relative in S2, the relative clause – subject, object, indirect object, predicate nominative, temporal, locative, stranded and shifted prepositions and genitive

Type of clause

Most grammars of English, whether prescriptive or descriptive, recognize at least two types of relative clauses: restrictive and non-restrictive. This distinction is made on the basis of the way in which the head NP or antecedent is modified by the relative clause. A restrictive clause further limits the head NP's reference, while a non-restrictive clause adds only additional information to a head which is already independently identified, or is unique in its reference and has no need of further modification to identify its referent. Classic examples of each type are:

1. (1) The girl who lives next door to me. (restrictive)

2. (2) Mary Smith, who lives next door to me. (non-restrictive)

Proper names constitute a class of unique referents because their identity is the same no matter what else may follow after. In the case of possible ‘mistaken identity’ though, proper names may occur with restrictive clauses, e.g. where there is clearly more than one person with the same name. In the following example, the use of the definite article also adds to this interpretation (cf. also Lyons 1977: ch. 7).

1. (3) The Mary Smith that I know lives next door.

In this chapter I discuss the extralinguistic factors which may affect the choice of relative markers, and a method for taking such factors into account in my study. I take it that it is also within the scope of this discussion to provide a critical review of some current methodological principles which relate to my investigation of the relative markers in Middle Scots, for example, sampling procedures and the problem of defining style and isolating contextual styles within texts. I will be particularly concerned with examining the extent to which the nature of my data allows the transfer or necessitates modification of sociolinguistic methods which have been used in dealing with synchronic speech data (cf. also Chapter 2). Although I am treating specifically one particular set of Middle Scots data, the problem I face is a more general one, which might be called the ‘reconstruction of language in its social context’. The question of whether such reconstruction is in principle possible is fundamental to the larger issue my work raises, namely, the status of socio-historical linguistic theory.

The problem of sampling

There are a number of ways in which a sample of data suitable for sociolinguistic analysis might be obtained. Currently the most fashionable one consists of interviewing individuals and groups both formally and informally.

The purpose of this Appendix is to illustrate the range of characters portrayed in Western Apache joking imitations of Anglo-Americans, and to provide supplementary information about the kinds of cultural contrasts these performances serve to highlight and interpret. Besides the four imitations described in the body of the essay, I witnessed eight others. Accounts of seven of these are presented below, together with reports of seven additional performances (selected from among a total of twenty-nine that were witnessed by Apache consultants. I cannot claim that my accounts are entirely faithful or fully complete. Western Apache jokers ply their trade on the spur of the moment, and, in most cases, conclude their imitations less than a minute after beginning them. Usually, it is possible to take note of most that has happened – but never all of it.

1. A ‘physician’ (izee' nant' án).

Setting: A ramada at an Apache camp in Cibecue.

Participants: A (male, age 29); B, A's brother-in-law (age33);and KHB.

Scene: A is sitting in the shade of the ramada, leafing idly through the pages of a comic book. B, who has recently injured his hand in a fall from a horse, enters the ramada and sits down. He examines his hand, which is wrapped in a bandage that has become soiled and dirty.

A: gostood' né? (‘It's hot, isn't it’?)

B: ' aa. gostood'. (‘Yes. It's hot’.)

A: nigan ndiih né? (‘Does your hand hurt’?)

B: 'aa. (‘Yes’.)

A: Why you go let that cloth get dirty? No good! You sit right there, my friend. I take care of you. I know everything! What happen? You been drinking too much?

B: 'ąął (‘That's enough’.)

A: [Ignoring B's request to stop joking] Maybe you been fighting – that's why you get all dirty. Maybe you got family problem. Maybe your wife get after you. Come here. I fix everything.