The data problem

Pragmatics is predominantly concerned with spoken language. The most prominent problem for historical pragmatics, therefore, concerns the availability of historical natural language data (cf. e.g. Bax 1983: 3). Since electronic recording techniques were not available until the twentieth century, spoken language of the past cannot be investigated through direct observation. Contrastive studies of present-day languages and language varieties can rely on the whole range of data-gathering techniques developed for the description of living languages, such as recordings and transcriptions of spoken conversations, native speaker intuitions, questionnaires and the like. Except for the very immediate past, historical-pragmatic hypotheses can never be empirically supported (cf. Bax 1983: 18). The pragmatician has to rely on written records as approximate evidence for his or her claims on spoken language. Furthermore, if the linguistic researcher accepts that for a historical pragmatic analysis he or she needs data on the socio-historical environment, he or she has to face the problem that texts from earlier periods inform very sparsely about their situational context (cf. Sitta 1980: 32).

However, historical linguistics has recently refined its methodology to a considerable extent, and it is now possible to get an approximate picture of the spoken language of past centuries. For example, sociolinguistic methodologies have provided the means to compare and measure the styles of different texts in terms of their formality.

Moreover, there are some arguments which show that results obtained through the analysis of written records may support claims on spoken language as well (cf. Bax 1983: 18f; Bax 1991: 212f). First, types of verbal interaction may be recorded not only in literary but also in other contexts such as juridical or clerical. Analogy may thus help to support a hypothesis which otherwise would be based on a single data source only. Second, literary history has shown that literary texts from the Middle Ages tend to be more realistic than today's fictional works. Third, from a functionalist point of view the question arises whether specific types of verbal interaction may be imaginable not only in fictional but also under real life circumstances. Fourth, there are many types of verbal interaction, such as quarrels, children's games, non-verbal signals, routines, which can be readily understood on the basis of underlying behavioral patterns. Thus, if a specific type of verbal interaction appears in historical texts and is also easily understood, the text may be taken to reflect a real life situation.

It is plausible to suggest that written records of spoken language are closer to the actual spoken language of the time than written language not based on spoken language. Features that are consistently more frequent in records of spoken language than in written language proper can fairly safely be hypothesized to be even more frequent in the spoken language of that period (cf. Rissanen 1986). These points have also been treated by Biber and Finegan (1992) in their comparative diachronic analysis of three written genres - essays, fiction and personal letters - and two speech-based genres - dialogue from plays and from fiction. Speech-based genres are defined as varieties originating in speech that have been permanently preserved in writing. These include various kinds of transcribed speech, such as court proceedings, political debates, town meetings, and some public speeches and sermons, as well as various literary representations of speech (Biber and Finegan 1992: 689)

The analysis takes place on three dimensions. Each one of them displays linguistic features that interact directly with the contrasts among 'oral' and 'literate' genres in English. Dimension A matches 'Informational vs. Involved Production'. Positive factors for this dimension include, for example, the type token ratio and attributive adjectives, while negative factors cover, for example, contractions and discourse particles. On Dimension B they distinguish 'Elaborated vs. Situation-Dependent Reference'. Here positive factors involve WH-relative clauses on object positions and nominalizations. Time and place adverbials belong to the negative factors. Finally, Dimension C is concerned with 'Abstract style', which comprises conjuncts, by-passives or agentless passives. The analysis comprises 163 texts representing four periods from the 17th century up to 1950.

As a result, texts from the 17th and 18th century tend to be more literate while from the 19th century onwards there is a transition to more oral styles (Biber and Finegan 1992: 695ff.). As far as abstract style and situation dependence are concerned, there is a drift to more oral styles and to less variation between oral and written styles. This might be taken to indicate that authors capture the linguistic characteristics of conversation fairly accurately with respect to Dimension B [...] and Dimension C [...] and further that conversation itself has not changed much over these centuries (Biber and Finegan 1992: 699).

In order to test this hypothesis Biber and Finegan used present-day conversation data from the London-Lund corpus. They come to the conclusion that "literary dialogue is remarkably similar to actual conversation with respect to these two dimensions" (Biber and Finegan 1992: 699f). In contrast, as far as Dimension A (Informational vs. Involved Production) is concerned, literary dialogue is markedly unlike modern face-to-face conversation. This is interpreted to be due to the high information load in literary dialogues. Nevertheless, Biber's and Finegan's study offers a way to determine the oral character of speech-based genres.

Letters, and in particular private letters, are a rich source of data for historical pragmatics. They may contain more intimate and more colloquial language than other text types. It is an empirical question whether they are therefore closer to the spoken language than other more formal texttypes, but they contain many interactional features such as address terms, directives, politeness markers, apologies, and so on. Kryk-Kastovsky (this volume) uses Early Modern English letters to analyze the demonstratives this and that. Nevalainen and Raumolin-Brunberg (this volume) use the same type of data for their investigation of address formulae.

Many researchers have come to see written artefacts not just as imperfect renderings of the real thing. Written texts can be understood as communicative manifestations in their own right, and as such they are amenable to pragmatic analyses. Danet and Bogoch (1992a, 1992b), for instance, analyze the pragmatics of Anglo-Saxon wills; Bach (this volume) investigates English wills from the 16th and 17th century; and Fritz (1993b) explores the communicative strategies of early German newspapers. In all these cases there are no claims about the spoken language of that particular period of English or German respectively. Wills as well as newspapers are analyzed as independent forms of communication that warrant a pragmatic analysis.

Literary texts, too, have been analyzed as communicative acts. Watts (1981) and Brönnimann-Egger (1991) both concentrate on the cooperation between the author and the reader of literary texts. Watts develops his analysis on the basis of Charles Dickens' Hard Times, while Brönnimann-Egger uses eighteenth-century English poets. Sell (1985a and 1985b) analyses Chaucer and his relationship to his audience (cf. also several articles in Sell 1991), and Navarro-Errasti (this volume) studies the Middle English poem Sir Gawain and the Green Knight.

These approaches contrast with those that use literary texts as a source for simulated spoken interactions. Bax (1981), for instance, explores the ritual challenges among medieval knights on the basis of Middle Dutch romances, while Breuer (1983) and Kopytko (1993a and this volume) use the verbal interactions in Shakespeare's plays to analyze the use of address titles (such as master, madam, your honour or goodman) and the use of politeness formulae respectively.

It is also possible to deduce information about the conversational behavior of earlier generations by closely studying what contemporary authors had to say about conversations of their time. Gloning (1993), for instance, analyzed a whole range of sources including manuals on good behavior, language teaching books, and literary references to conversational behavior, and Fritz (this volume) uses extracts from Albrecht von Eyb's so-called Ehebüchlein - published in 1472 - which is a treatise on legal proof procedures on matters of marriage.

Burke (1993: 89-122), in a study of conversational behavior in Early Modern Europe, looked at a large number of manuals which appeared between the 17th and 19th centuries in England, France and elsewhere. They typically have titles such as The Art of Conversation and tell their readers how to behave in conversations either generally or on particular occasions. They warn their readers, for instance, not to interrupt their interlocutors and not to talk about themselves.

Historical pragmatics will always have to rely on written material. However, this should no longer be seen as detrimental. There are many ways in which the written data can be used. Modern sociolinguistic methodologies can help to establish which texts may be used as rough approximations to the spoken language of that time. Literary texts may contain simulated spoken interactions. In the case of drama they actually consist almost entirely of simulated spoken interactions. And finally, written texts can be analyzed as communicative acts in their own right.

In all cases it is of course crucial to use the most reliable editions available. Both Ronberg (this volume) and Lennard (this volume) argue very strongly for editions that stay as close to the original texts as possible. They show that the modernization of punctuation has far-reaching consequences for the pragmatic interpretation of historical texts.