1 IntroductionFootnote 1
Historical linguistics is a field that, perhaps more than other branches of linguistics, can be said to exhibit a certain conservatism. To be clear, this term is not meant in any traditional political sense. Rather it is meant to capture the notion that, as a discipline, diachronic studies seem to accept and build on previous theories and empirical findings to a greater extent than do most synchronic subdisciplines. This may be because data are comparatively rare and hard to come by. One result of this scarcity is that, once analyzed, there are fewer opportunities for reanalysis predicated on new data. There are, of course, occasions when more or less radical proposals are brought forward subsequently, which result in debates of the kind which are much more common in synchronic syntax, say, or phonology. The reconstruction of the Indo-European consonant system (Beekes Reference Beekes1995: 132–4 provides a summary), for example, continues to be debated almost two hundred years after it was first proposed.
1.1 Data and explanatory concepts
Most often, however, descriptions of data are well established, as are the conclusions drawn from the data. The conclusions and their theoretical underpinnings are not static, however, but the general trend is to expand or modify earlier analyses rather than to discard them and start as if from the very beginning. Two examples will suffice. First, there have been very few changes in the reconstruction of the Indo-European vowel system since it was established in the nineteenth century. Meillet (Reference Meillet1903), Lehmann (Reference Lehmann1951) and Beekes (Reference Beekes1995) all posit */e/, */e:/, */o/ and */o:/, two reconstructed short vowels and two corresponding long vowels. Discussion continues on the role of these vowels in morphology (via ablaut), but the reconstruction of the system has remained very consistent.
A second example is more theoretical. A standard technique for diachronic studies is to consider the actuation, or first development, of a variant as a scientific question, different and, to a great extent, independent from the question of the spread of that variant. It is the successful spread of a new form which brings about genuine change to any language, but some individual event must, of course, activate the process. In 1968, in a seminal paper, Weinreich et al. (Reference Weinreich, Labov, Herzog, Lehmann and Malkiel1968) discussed at length the spread of change, attributing it in large part to the prestige of the dialect being imitated, where prestige takes many forms, some overt and some covert. The notion of imitation, however, is not a new one. Meillet (Reference Meillet and Rosenberg1908: 22) attributes imitation (which can be associated with prestige, although the link is not explicit) as one of the causes for the spread of the Indo-European languages, a notion taken up again by Bartoli (Reference Bartoli1925: 38), who states it in terms of prestige. Again, practitioners of historical linguistics have expanded and enhanced the notion of imitation, but the concept serves as a well-established aspect of the discipline's foundation.
1.2 Methodologies
Not only do analyses and data endure, although modified in many ways, but so do the methodologies of diachronic linguistics. They fall, traditionally, into two approaches, depending on the nature of data; the dichotomy is simple in that data either are or are not attested. In the case where data exist, the general methodological rubric is philology, most broadly defined as the study of texts (Winters & Nathan Reference Winters, Nathan, Brentari, Larson and MacLeod1992 provide a discussion of the term). Historical linguists have used those data, often through critical editions of texts, as a basis for their knowledge of both older dialects where the modern language continues to exist and also as a source of data for the so-called dead languages. It should be noted that although this approach seems straightforward, careful analysis of the texts is called for, in some cases just to decipher the alphabet, to determine phrasing, even to establish word boundaries in the absence of punctuation or even spacing, and to decide what relationship exists between symbols and oral language.
The second methodology is the equally time-honored one of reconstruction (see Fox Reference Fox1995 for an overview). It is widely applied where textual data do not exist, in order to hypothesize the nature of earlier stages of attested languages or language families; this is the case both for the non-controversial vowel system of Indo-European and its far more disputed inventory of consonantal stops. Again, as with philology, reconstruction has its appeals and its difficulties. Perhaps the most salient is that it is a form of hypothesis building; until new textual data are uncovered (which does occur occasionally), it is impossible to talk about the truth or falsity of a reconstruction, just its degree of plausibility.
To conclude, historical linguistics, more than other subdisciplines, tends not to discard foundational hypotheses, theory, data, or methods completely. Rather, specialists in this field are more likely to expand and enhance what has already been proposed to find new interpretations of past stages (attested or not) of languages.
2 Uniformitarianism as a bridge to innovation
The picture painted above is not at all meant to be negative; diachronic studies have a long and honorable record of exciting discoveries based on previous work. But this is not to say that new ways of thinking and new methodologies are not welcome and should not be fully explored. The collection of articles in the present volume does exactly that, while taking good advantage as well of the legacy of past work in historical linguistics and, specifically, the history of English. The pivot between the well-established analyses and methodologies and the more experimental approaches here resides in the notion of uniformitarianism, to be further discussed in this section.
2.1 Uniformitarianism
Although uniformitarianism is not always explicitly mentioned, every article here clearly reflects its force. A concept which was first developed in the realm of nineteenth-century physical sciences (Janda & Joseph Reference Janda, Joseph, Joseph and Janda2003: 23–37), uniformitarianism proposes, to state it as comprehensively as possible, that what we can discover in the present and about the present is relevant to what we discover in and about the past. In a sense it is a statement that, absent direct evidence, historical scientists of all sorts must extrapolate backwards, into the undocumented past. Traditional linguistic reconstruction depends on uniformitarianism, for example, since the plausibility of a hypothetical earlier form is based on the extent to which that form is a likely ancestor of the documented one it is said to precede. The degree of likelihood is based in turn on what changes may have taken place in the reconstructed form to result in the documented form. These hypothesized changes are, in at least most cases,Footnote 2 those which have been seen to take place in the documented past.
2.2 Linguistic applications of the Uniformitarianism Principle
This principle seems to be straightforward and is, indeed, open to very little dispute at the highest level of abstraction: the present or the documented past should be used to interpret the prehistoric (undocumented) past. The challenge, however, is equally straightforward: what is it in the present or other documented period of language use which might serve as a guide to the past? The responses are, to date, rather less straightforward. A range of historical linguists state the basis for this extrapolation in a range of ways. Bynon (Reference Bynon1977: 45) and Fox (Reference Fox1995: 195) both cite principles as the recurring feature of language, while Fox adds the notion ‘that the regularities that can be observed in current and attested languages are equally valid for reconstructed languages’. Hock (Reference Hock1991: 630) talks of observable phenomena as possible sources for reconstructing the past, while Christy (Reference Christy1983, quoted in Janda & Joseph Reference Janda, Joseph, Joseph and Janda2003: 30) refers to processes and Trask (Reference Trask1996: 232) to natural laws. Finally, Croft (Reference Croft2003: 49) points to the assumption that the rules that govern language structure today are the same as those that governed it previously and will govern it in the future. It is beyond the scope of these remarks, however, to do more than point to these different characterizations of the Uniformitarian Principle without providing an analysis of the use of ‘principle’, ‘phenomenon’, ‘process’ or ‘rule’.
2.3 Questions raised by the principle
Of greater relevance here are two other observations stemming from the principle. First, several discussions of the concept go further than simple (but often non-straightforward) definitions. Christy (Reference Christy1983, quoted in Janda & Joseph Reference Janda, Joseph, Joseph and Janda2003: 30) remarks that
[w]hat is observed in the present can be proposed for the past but what is not observed in the present cannot simply be banished from the realm of possible.
In saying so, he is opening up the possibility that there are, indeed, conditions (again to use a neutral term) which we have not (yet) come across in the present, by which he means as well the documented, historical past.
Secondly, questions are raised as to another aspect of the scope of the principle. Labov (Reference Labov1994: 23) accepts that uniformitarianism may act on the physiological basis of language, but finds it more problematic in talking of its social setting, citing the danger of explaining pre-urban societies through the present. Trask, on the other hand (Reference Trask1996: 285) states explicitly that it applies to such social factors as we can extrapolate backwards, at least at the level of the abstract notions of prestige (or lack thereof) and of stigmatized features. Again, the consequences of these diverging views deserve greater exploration than is appropriate here.
Labov (Reference Labov2001: 21, quoted in De Smet & Van de Velde this volume) proposes the Historical Paradox. Underlying historical linguistics at its most basic level is the idea that languages change over time; there is certainly well-documented evidence of all sorts that this is the case. However, framing explorations of this idea, through the principle of uniformitarianism, is tantamount to saying that – in some way or another – languages (or at least Language) do not change after all; the same principles, rules, phenomena, regularities must be said to have existed in the past. The point, therefore, is to find where these seemingly contradictory conditions can harmonize. The response may be in the level of abstraction at which uniformitarianism is understood, or in the components of language and human linguistic behavior to which it may be applied. Most of the articles in the current volume are seeking new responses to this series of puzzles, using the theoretical framework and functional apparatus of Cognitive Grammar.
3 New approaches to diachronic Cognitive Grammar
To date, diachronic studies within Cognitive Grammar have almost entirely addressed semantic questions, be they the semantics of the lexicon or of more abstract morphological or syntactic entities. Early approaches involved the configuration of radial sets and the identity of prototypes. Later diachronic studies, including some in this volume (Harbus, Hoffmann, Traugott), addressed constructions and the changes over time which they may undergo. In every case uniformitarianism is called upon either implicitly or explicitly, with greater or less discussion. In this respect, the volume is a collection of solid articles on various phenomena in the history of English ranging, among others, from the meaningful literary role of alliteration (Harbus), through the lexicon understood broadly (Cole, De Smet & Van de Velde), to syntax (Hoffmann, Traugott) and discourse-level matters (event structure in the case of Pentrel and the effectiveness of extravagant language in the case of Petré).Footnote 3
The volume, however, reflects various new approaches as well. First, the availability of data for diachronic analysis has changed dramatically with the proliferation of searchable online, computer-based corpora. This is a far cry from earlier eras of research when, to present primary data (and not just to reuse data which were already quoted in articles or historical manuals by others) meant to read as many paper or microfilm texts as one could find, searching for examples of whatever grammatical or lexical unit was being studied. One never was sure of finding all the primary data, especially in languages with a large extant literature like Old English or Late Latin. It is not necessary to review here the details of flagging forms and running comprehensive searches, but the results are clearly more comprehensive from several points of view.
However, there is far more intriguing innovation to be found: many of the articles are based on the premise that not only can one draw more certain conclusions about the behavior of a given language or language group from far more robust sources of data, but one can also hypothesize more widely about human cognitive linguistic activity in earlier times. This is a bold step in that it takes uniformitarianism in a new direction. As was said above, the principle has been applied variously to rules, phenomena and processes, but not directly to human thought. In several of the articles here, however, psycholinguistic methodology is being extrapolated into the past to draw conclusions about the cognitive basis for the change being studied.
It should be said that this approach is not entirely novel since syntactic (and other) studies have proposed changes based in cognitive function in the past. Lakoff's (Reference Lakoff1987: 440–4) notion of Image Schema Transformations is just that; a construction (using the term broadly) can be manipulated in the mind to conceptualize and ultimately express something new. Lakoff developed the term for synchronic use; it can, however, be applied diachronically as well. In a article on the Early Modern English construction of a dative-experiencer subject (methinks has survived in very specific contexts and will stand here for a range of such constructions), I proposed that its change over time to a nominal subject (I think) shows a cognitive shift in the way a thought is construed. Instead of conceptualizing the subject matter of the thought as coming toward the speaker from the outside (as symbolized by the dative-experiencer subject), thinking is conceived of as an activity engaged in by the speaker. The grammatical result (agentive rather than experiencer subjects) would follow from a new way of construing the interior, epistemic world (Winters Reference Winters2010).
What is proposed in this volume goes far beyond this kind of approach. While Winters (Reference Winters2010), and others like it, simply hypothesize how thinking (tacit or explicit) may have changed over time, the present articles include psycholinguistic experiments (better characterized as real data filtered through thought experiments) using the ample data in online historical corpora of English in lieu of the results of direct testing of human subjects. A few short examples should suffice. Petré, using Keller's (Reference Keller1990) work on the actuation of change, proposes that the birth (and rebirth) of usage for particular effect (extravagant or vivid language) has as its basis a change which symbolizes an increase in the direct involvement of speakers in a given event. Bringing the self into the discourse, he hypothesizes, results in greater cognitive salience on the part of the listener. The change, therefore, would take place, presumably, in the cognitive architecture of both speaker and listener.
Three further articles are even more explicit in their psycholinguistic approach to the language of earlier periods. Pentrel explores sentence production, specifically the placement of adverbial clauses in Samuel Pepys's Diaries of the seventeenth century. The explanatory importance of iconicity, that is, the use of linguistic order to reflect experience, is strongly confirmed, with less clear results when it comes to the role of adverbial length or other factors in accounting for this placement. Pentrel is able, cautiously, to confirm clause order as responding to at least one factor that has been identified through synchronic means.
Hoffmann turns his attention to ease of processing, a constraint on grammars which may be uniform across time. His data come from three competing correlative structures in Old English, looking not only at the structures themselves, but the role of each of them in the entire constructional network. The surviving version of the construction is supported not only by ease of processing, but also by the fact that it filled a gap in the network. Hoffmann therefore takes uniformitarianism one step further in that he brings in the concept of network, not often discussed in Cognitive Grammar beyond descriptions of, usually, a single unit. He proposes that gaps in a network (in his case a network of constructions) serve as a stimulus for language change based on the cognitive phenomena of identifying such gaps and remediating them within what is conceptualized as a structural unit.
Finally, De Smet & Van de Velde look at the morphological compositionality of adverbs and priming, exploring some relatively modern, though diachronic, data to determine whether priming may or may not become less effective as compositionality (the degree to which, in this case, adverbial bases are recognizable by speakers) decreases. Priming is a tool of psychological and psycholinguistic research normally carried out with subjects in real time. To measure diachronic effects, De Smet & Van de Velde use segments of texts, looking at earlier instances of adverbs (within 500 words) as priming textually later instances.
4 Cognitive functioning
The application to cognitive functioning of uniformitarianism is a crucial theme in this volume and the one to which these remarks have paid the greatest attention. But other aspects of thinking which human beings bring to language are also treated diachronically here. Iconicity, best known synchronically in the work of Haiman and others (Haiman Reference Haiman1985) features diachronically in the articles by Pentrel and by Hoffmann, serving as a way of viewing the order of the symbolic world such that linguistic change occurs to reflect that order. It is not far apart from analogical thinking which must, at least in some cases, also underlie the modification of the order of units toward one modeling the non-linguistic exterior or epistemic experience being expressed.
The intriguing question of mental units is also considered in several articles. Those by Traugott, Hoffmann and Harbus all use Construction Grammar as their framework. Traugott looks at what elsewhere have been called small clauses, units smaller than the independent clause (see Progovac et al. Reference Progovac, Paesani, Casielles and Barton2006). Traugott considers them as constructions and demonstrates their durability and even expansion in use over time. Her article prompts Bergs, as respondent, to question the very status of the sentence as the basic linguistic unit, a question which has been occasionally raised in the past as well and certainly deserves much more attention. Cole, on the other hand, looks at much smaller units, tying syntax (in the form of poetic half lines) and phonology (source of the alliterative segments) into linked units which make Old and Middle English verse meaningful to the reader.
All this is to say that several of the major topics of this volume transcend the history of English, although it remains, like the history of the Romance languages, a testing ground for theoretical explorations. This has been the case since long before twentieth- and twenty-first-century technology gave researchers access to large amounts of data; the ability to test new ideas has been thereby greatly expanded. Another, perhaps less obvious point, is that the same topics (uniformitarianism and the nature of linguistic units) also transcend theoretical frameworks in a certain sense. While the articles in this volume are informed by Cognitive Grammar, these wider issues can be, I believe, fruitfully explored as well in other functional and formal frameworks, either strictly within the theories themselves or at points of intersection. In this volume, Hoffmann's remarks on the integration of Structuralist theory into Construction Grammar are a step toward this kind of innovation.
5 Conclusion: prospects and cautions
There is, obviously, a great deal of further exploration needed for several aspects of these proposals. First, uniformitarianism, both in definition and in application, has not yet been rigorously defined, as exemplified by the array of theoretical constructs (rules, principles and/or processes) about which constancy or uniformity is claimed. Although they may all be appropriate, such a conclusion cannot be reached until there is an overarching sense of what uniformitarianism is meant to do. Related to this is a lack of clarity as to whether it can apply only to linguistic phenomena, to social conditions, or – as in this volume – to mental states. Lastly, the question arises as to its limitations: if something linguistic (as abstract a term as possible) is not attested in the present, does it mean it cannot exist in the past? A response to this question, which must necessarily take the form of a hypothesis (as opposed to a flat statement or maxim), will shape how uniformitarianism is put to work in diachronic linguistics.
As a result of the application of uniformitarianism in several of these articles to cognitive activity (iconicity and event structuring, reactions to priming) more questions of limitations arise. There is a great deal still to be learned about both cognitive processing and mental storage, with questions as basic as the very nature of the sentence as a grammatical entity. Is it a basic unit? An epiphenomenon? All research to date and, indeed, instinctive reactions to this question seem to lead to the conclusion that it is not a primitive, but constructed as we speak. What does this say about smaller units, Traugott's MIS (‘Monoclause Introduced by Subordinators’) or small clause, and its role in processing and storage? Traugott herself looks in the direction of broader discourse for explanations, but the complex network of questions as to what has remained constant in the cognitive make-up of human beings in the storage of linguistic units still remains open to a great deal of further work.
It is, in a very particular way, hard to judge the articles in this volume. This is not at all to say that any of them is poorly constructed or badly presented – quite the contrary. They all instantiate what was brought out in the introduction of these remarks, that diachronic linguistics builds on the theory, data and methods of previous studies. What stands out, however, in these contributions to the history of English is the pervasive underlying hypothesis that ascertaining the cognitive activity of past speakers through documentation may be feasible, as inspired and supported by the Uniformitarian Principle. Any evaluation, therefore, of the quality of the conclusions, as opposed to the discussion of data in these articles must include the phrase ‘it depends’.
How far, then, can we as historical linguists extend these innovative notions? I suspect that any response to that question must also, in the end, come down to ‘it depends’. The ideas are certainly intriguing, but need a great deal of testing through careful application to other kinds of data. And that work will not be carried out in a vacuum. There is still a great deal we do not know about the nature of the human mind that is being explored directly, through the work of neuroscience as well as psycholinguistics. This is not to say, however, that historical linguists should wait for all to become clear synchronically; historical, document-based experimentation, as exemplified largely for the first time in this volume, should be able to inform synchronic, even non-linguistic, work on the nature of human cognition. We have a long way to go, but cannot overestimate the need, addressed here, for ongoing interconnected experimentation.
