1.1.1 Introduction
The goal of this chapter is to examine recurrent themes emerging from quantitative research on grammatical variation and change. These range from methodological considerations to more abstract issues such as the identification of a particular communityâs linguistic standard. I refer both to historical sociolinguistic investigations and to research in contemporary communities. In subsequent sections, I look first at historical sociolinguistic research, moving on to analyses in contemporary communities of a specific variable: subject-verb agreement. Finally, we come to the standard language question, which has regularly emerged as interesting and problematic but has seldom been systematically addressed.
For a variety of reasons widely discussed in the sociolinguistic literature, some morphological variables are viewed as relatively amenable to quantitative analysis. Cheshire (1999) has observed that this same set of apparently simple variables has tended to get analysed repeatedly, i.e., various non-standard verb forms, subject-verb agreement, multiple negatives and non-standard negative forms such as ainât. She also notes that the saliency of such non-standard forms to laypersons and linguists alike makes them a target for prescriptive comment. The analytic procedure is generally to relate speakersâ choices to social variation with a view to inferring patterns of change. However, the extension of these basic quantitative methods to high-level grammatical variation raises methodological questions which regularly impinge on broader theoretical concerns. For this reason, sociolinguists have often argued that the concept of the variable can be usefully applied only to low-level morphological variation (see further Milroy and Gordon 2003:169â172).
One intrinsic difference between phonological and grammatical systems has far-reaching implications for quantitative analysis. Since speakers make use of a sharply limited and therefore frequently recurring, inventory of phonological contrasts, realisations of any given variable are likely to show up frequently in even a short sample of speech. This is not the case for grammatical variables, since a sufficient quantity of tokens of a given type of construction cannot usually be guaranteed to appear in a piece of spontaneous discourse. This difficulty is partly a consequence of the non-finite or âleakyâ nature of syntactic systems, which in turn is associated with the susceptibility of syntactic choices to pragmatic and semantic constraints. Speakers can exercise considerable choice in the way they use grammatical resources to encode meanings since there is no isomorphic relationship between function and form. For example, questions are not always realised syntactically as interrogatives, and interrogative forms may realise many different functions (Coveney 1996:123).
These more âdifficultâ types of variable often present interesting issues of analysis and interpretation. Conversely, apparently low-level morphological variation is not always as simple as it seems, and analyses of grammatical variables have often focussed on internal linguistic constraints rather than on relationships between linguistic and social variation. Indeed, grammatical variation has regularly been found to benefit from the insights of formal syntactic analysis (see further Milroy and Gordon 2003:190ff). The social distribution of a grammatical variable on a vernacular/standard continuum is often far from straightforward; contrary to what we might expect from reports in the variationist literature, clearly non-standard grammatical variants are used by high-status local speakers in some communities, as noted by Beal (1993) in Tyneside and by J. Milroy (1981) and Harris (1984) in Belfast. This pattern of use raises ideological issues of how a spoken standard English is imagined in dialect areas distant from the central and southern regions of the UK. Generally speaking, the problems summarised above are less relevant to historical sociolinguists than to those who work in contemporary communities. Accordingly, we look now at some particularly illuminating historical analyses of two frequently studied morphological variables.
1.1.2 Historical Sociolinguistics
Sociolinguists have regularly followed Labovâs (1972:275) excellent advice to âobtain at least one measurement at some contrasting point in real timeâ. Thus, for example, our understanding of the direction of change around 1980 of the complex (a) variable in Belfast was greatly enhanced by evidence provided by the elocutionist David Patterson in 1860. But variationist work on earlier stages of the language has gone far beyond the methodologically motivated search for a linguistic anchor in real-time; investigators have regularly used older written texts to add historical depth to their analyses. Sankoff and Vincent (1980) report that stylistically stratified patterns of variable deletion of the French negative particle ne have hardly changed since the sixteenth century, when deletion was associated with informal styles. They note that ne now appears only rarely in conversational contexts but is favoured in certain formal (particularly written) styles. Romaine (1982) reports a similar stability over time of the relative pronouns system of Middle Scots, where the ranking of stylistic and syntactic constraints on choice of relative pronoun variant appears to have changed little in 450 years; zero marking in subject position was preferred in written Scots in less formal styles and continues to be a characteristic of the contemporary dialect. Trudgill (1996) provides details of a language contact situation in sixteenth-century Norwich to account for a contemporary pattern of alternation between zero and âs present-tense, third-person singular verb forms in contemporary Norwich vernacular.
Thus, the relationship between sociolinguistics and historical linguistics has always been close, but the influence has not all been in one direction. Historical linguistics in general has been quite extensively influenced by the methods and theories of variationist sociolinguistics, giving rise to a fruitful two-way exchange of findings and ideas. Twenty years ago, Pratt and Denison (2000) noted that sociolinguistic models were attracting increasing interest among historians of language who have previously concentrated on language internal accounts of change. More recently, Auer and Voesteâs (2012) review of the methods and findings of historical sociolinguistic work on grammatical variables reveals the development of a subfield employing variationist methods to investigate trajectories of changes which have often been completed at early stages of the language.
Historical linguists work with the uniformitarian principle, which holds that patterns of variation in the past are similar to those observed in contemporary speech communities (see Lass 1997:26ff), but the methodological challenges encountered in applying this principle have led Labov (1994:11) to describe historical linguistics generally as âthe art of making the best use of bad dataâ. The âbad-dataâ problem has several dimensions: data are often patchy as a consequence of the random preservation of some texts and the equally random loss of others; the relationship between data derived from various kinds of written source and the data of spoken interaction which forms the basis of much contemporary sociolinguistic work is unclear; reconstructing the social information needed to interpret patterns of variation in written texts is not always straightforward. Given, however, the difficulties of obtaining sufficient tokens of grammatical variables in contemporary speech communities, the bad data problem is surely of limited relevance to accounts which draw on substantial computerised corpora. Historical researchers also have some benefit of hindsight in assessing the social significance of particular changes. In a number of publications (e.g., Nevalainen and Raumolin-Brunberg 1996, 2017; Nevalainen 2000a, 2000b; Nevalainen et al. 2011) Nevalainen and her colleagues report on a series of analyses which draw on the University of Helsinkiâs Corpus of Early English Correspondence (CEEC). This extensive corpus spans the years 1410â1681 and includes the letters of 778 writers. The researchers show that earlier social worlds can be reconstructed from the detailed findings of social historians, some of which are of considerable sociolinguistic relevance (see, for example, Keene 2000). Nevalainen and Raumolin-Brunberg (2012:32) suggest that
personal correspondence provides the ânext best thingâ to authentic spoken language, and even with its obvious limitations makes it possible to extend the variationist paradigm into the more distant past. It enables the researcher to combine macro- and micro-level approaches and place individuals within their language communities.
Researchers are able to present well-motivated accounts of the social trajectories of particular grammatical changes associated with the sixteenth and seventeenth century, a period of particularly rapid social change (Nevalainen 2000b). Beal (2020) similarly shows that the personal letters of the naturalist Thomas Bewick, along with other âego-documentsâ such as notes and diaries, are akin to spoken language, showing a clear contrast with Bewickâs professional published work (see comments in Section 1.1.3 below).
The Helsinki group has examined a wide range of grammatical variables, and I comment here on just two of them. The first is the change from the older third-person present singular verb form -eth as opposed to the innovatory northern dialect form -(e)s in the sixteenth century. These variants were in competition for over 200 years before -es, which had emerged as an alternating variant in the fifteenth century, took over from -eth, becoming the norm by about 1600 in all but âhigh registersâ (Lass 1999:162â165). The following example (dated 1585) shows Queen Elizabeth I of England using both variants in a single letter:
He knoweth not the pryse of my bloude, wiche shuld be spilt by the bloudy hande of a murtherarâŠâŠI am assured he knowes and therefor I hope he wil not dare deny you a truthe.
This extract gives a flavour of the very rich data collecte...