The handbook of historical linguistics
This handbook appeared as the twelfth volume in the valuable 'Blackwell handbooks in linguistics' series, which is still growing and which since the late 1990s has offered us insight into the state of the art of a wide range of subdisciplines within linguistics. It differs from the others in one significant way (apart from having a complete bibliography at the end rather than separate ones per chapter, which I think is more useful and saves space): it has a very long and detailed introduction covering as much as one quarter of the complete text (including thirty pages of [End Page 206] footnotes!). This may seem somewhat excessive. Having read through it, however, I greatly value it for a number of reasons. It deals extensively and often in a delightful (if not provocative) and ruminative way with some knotty but very important issues in historical linguistics, which cannot easily be dealt with in the separate chapters.1 Thus a lot of space is devoted to the subject (or, as it turns out to be, the 'nonsubject') of uniformitarianism; to a discussion of ideas in biology and palaeontology that often are too easily applied to linguistics (such as the notion of 'punctuated equilibrium'), but that are useful areas all the same to pluck ideas from; to the notion of time and the related question of whether there is such a thing as 'linguistic change' (this also includes a discussion on 'unidirectionality'); to the wrong-headed 'organicist' approach to language; to the issue of discontinuity versus stability (the latter often being ignored); to the use or rather the (im)possibility of reconstruction; and, finally, related to this, to the importance of the present to understand the past.
It would take up too much space to do justice to all of the topics covered by the introduction, or to devote time to all of the individual chapters, so I concentrate here on some core issues in the introduction (and how chapters relate to these) and on its main aims and findings.
Most attention is paid by Janda and Joseph to the nature of language, and to the nature of change and of language change. They also address fellow historical linguists directly by reflecting on what their (our) future tasks should be. In addition, in the course of the general discussion, the introduction announces what is and is not to be found in the ensuing chapters, and why, and their decision (a wise one I believe) to include a plurality of voices. As to the latter, we see below that there exist many controversies within the field of historical linguistics, controversies that arise because of different theoretical viewpoints (formal, functional, variationist approaches). Since the theoretical framework chosen has a very strong effect on how data are analyzed and explained, and even on the type of data or the amount of data that are selected, there is a real need to hear all of these voices when one wishes to present an objective view of the field. The editors provide plenty of room for this even though their own preferences become pretty clear from their introduction. But this makes the book spicy, hopefully resulting in fruitful discussions.
Concerning the first issue, the nature of language, the two editors emphasize again and again that languages do not develop as natural objects (i.e. they plead against the 'organicist' approach) and that grammars do not change themselves: it is the 'speakers [that] have something to do with it' (10). Speakers are considered 'the only known organisms which/who come into question as plausible agents of change in languages' (ibid.; see also p. 79ff.).2 I fully agree that this cannot be stressed enough; language change is discontinuous because it involves the replication of forms by human agents, either in a stable fashion or in a changed way (the issue of stability vs. diversity is given a lot of welcome space in Ch. 5, one of the methodology chapters, by JOHANNA NICHOLS). In other words, the forms themselves (the TOKENS used in language) do not CHANGE; forms are REPLACED. This is also emphasized in Chs. 7, 14, and 21 by MARK HALE, DAVID LIGHTFOOT, and BENJAMIN W. FORTSON, IV on phonological, syntactic, and semantic change respectively, and by JEAN AITCHISON in Ch. 25, who uses effective metaphors to describe the two different views: the 'tadpole-to-frog model' vs. the 'multiple birth model' (740).
As to tokens, the introduction provides a useful and extensive discussion of the difficult distinction between 'token' and 'type'. In principle, EACH utterance is a separate token; that is, the same word pronounced again or by a different speaker is ANOTHER token. Together these tokens form a type. This is, however, a very superficial type because the distinctions between these tokens are minimal (in some theoretical models, these types would still be considered [End Page 207] tokens). In such cases, one therefore sees stability rather than discontinuity: the tokens resemble each other 'so closely that no systematic . . . trend of change in form is evident across such a chain of two or more members' (48). Types, then, may be a COLLECTION of more or less the same tokens, but they may also represent more ABSTRACT PATTERNS representing those tokens. Not surprisingly, historical linguists mostly pay attention to instances where there seems to be a 'change' in the more abstract kind of type.
Ch. 10 by RAIMO ANTTILA on 'Analogy: The warp and woof of cognition' provides a very interesting discussion as to how analogy (involving both form and meaning) plays a primary role in the recognition of similarities between tokens and between types, and also of mismatchings or approximations between them (after all, language is analog rather than digital, which makes precise correspondences or clear oppositions the exception rather than the rule). The latter may lead to 'change'. The editors have given a lot of space in this volume to the notion of analogy, which, after having been ignored for quite a while—especially within formal models that considered it too loose a notion—seems to have come back with a vengeance. Ch. 11 by HANS HENRICH HOCK on 'Analogical change' shows how pervasive analogy is in matters of change; PAUL KIPARSKY (Ch. 6, 'The phonological basis of sound change') points to the role played by analogy in lexical diffusion; while WOLFGANG U. DRESSLER's paper (Ch. 12) on 'Naturalness and morphological change' discusses analogy under the heading of iconicity. Other recent studies showing the renewed interest in analogy are Hofstadter 1995, Holyoak & Thagard 1995, Gentner et al. 2001, Itkonen 2005, Wanner 2006, and Fischer 2007.
Because language tokens do not change but are replaced, the editors further argue that there is no such thing as 'linguistic change' (80ff.), and, consequently, that there is also no such thing as unidirectionality. This is a controversial issue among historical linguists. Grammaticalization linguists, for instance, strongly adhere to a principle of unidirectionality (see Ch. 18 by BERND HEINE and Ch. 20 by ELIZABETH CLOSS TRAUGOTT on grammaticalization; it is also implicit in Ch. 19 by JOAN BYBEE, which concentrates on the role played by frequency), while linguists of a generative persuasion (here represented in Ch. 14 by Lightfoot and Ch. 15 by SUSAN PINTZUK)3 reject any such notion.4 GREGORY R. GUY, in Ch. 8 on phonological change (394–95), writes that certain changes such as mergers are unidirectional (but cf. Trudgill 1974), and that other phenomena such as deletion and assimilation are more common than insertion and dissimilation;5 he emphasizes, however, that these are tendencies, no more.
A parallel situation with respect to unidirectionality can be found in existing ideas about biological evolution. Some scholars (e.g. creationists or believers in 'intelligent design') consider the appearance of humans the pinnacle of some inevitable trend in nature toward greater complexity, unidirectionally steered as it were by some invisible agent. Others, Darwinists such as Stephen Jay Gould, maintain that humans are simply a contingent product of an accumulation of chance events. If the evolution were to occur all over again, the chances are one in a million (or much less) that humans would appear on the horizon again. This is because it is most unlikely that exactly the same path would be followed, due to the fact that this would require that the circumstances under which evolution takes place do not themselves change.
Thus each option in the evolution, taken out of a range of possible options, in fact closes off the development of the options that have not been selected. The same is true in language (cf. [End Page 208] Aitchison's (1987) metaphor of change as a spaghetti junction). It is for this reason that some grammaticalizationists (e.g. Haspelmath 2004) reject the possibility of de- or antigrammaticalization (of grammaticalization going backwards)—which, by the way, simultaneously strengthens their idea that the process must be unidirectional—because they consider as counterevidence to unidirectionality only a process that constitutes an exact reversal of the 'tokens' (note that in fact these tokens in their model are already more abstract 'types', cf. above) that formed a chain in the grammaticalization process. Unidirectionality is also essential to grammaticalizationists because it implies that the grammaticalization process cannot be an epiphenomenon,6 that is, cannot be reduced to more common changes occurring elsewhere. Other grammaticalizationists are more lenient in that they also recognize cases as degrammaticalization when they involve more abstract or schematic 'types', such as those involved in a 'cline' (on this issue see e.g. Rosenbach 2004:85ff.).7
This controversy shows again how important it is to make a clear distinction between tokens and the various kinds of types. As indicated above, as far as 'change' is concerned, it may be more fruitful to look at what happens to 'types', to look at the more abstract patterns. When one considers grammaticalization on such a higher type level (as evidenced in grammaticalization theory in the notion of clines, see n. 7), it is clear that degrammaticalization must also be said to occur even when it is not a true mirror image of an earlier grammaticalization process (cf. Joseph's Ch. 13:477).
Can developments in language be compared to biological evolution? Yes and no, as Janda and Joseph make clear, for instance, in connection with their very illuminating discussion of 'punctuated equilibrium'. Referring to punctuated equilibrium is a highly fashionable pastime among some linguists, who believe that linguistic 'change' (or, more properly, 'replacement') can be both catastrophic and gradual. Janda and Joseph point out that indeed we may find in language major structural changes as well as periods of stasis, but to use punctuated equilibrium as 'proof' that major grammatical changes may take place across one generation (as is done in generative models) is a misuse of the biological notion, where 'punctuations' still take 'thousands or ten thousands of years' (73). They write in this respect that external disciplines can be 'most helpful if scouted out heuristically—as available sources for borrowing (or generating) novel hypotheses and other ideas—[but we should be careful not to take them] as models for emulation' (68).
It is quite clear that language and biology do not evolve in the same way. First of all, language as a cultural product can evolve much faster than is possible in biological change, as is clear from a comparison of, for example, Old English and Modern English, which differ only by 1,000 years but which are definitely not recognized by beginning students of English as being at all alike (in contrast, humans have not changed that much in 1,000 years). Second, changes in language can be passed on directly to the next generation, whereas biological evolution is indirect, because favorable traits can only evolve by genetic change. Finally, language can change across linguistic branches (for instance, through language contact) whereas in biology branches that have diverged cannot subsequently again be joined (65).
Returning now to the vexed problem of unidirectionality, it is true that with hindsight a trend may be discovered, but each stage in the development of a language is in principle independent of a previous stage. What it does depend on is the speakers, and the contours of the grammatical system that they have extracted from the primary linguistic data (PLD) around them. There will be continuity, of course, because communication must be maintained between generations, but what will be replaced and what will remain stable depends not on the way in which a language had developed so far but on synchronic variation and on the frequencies of the variants (on the importance of frequency see Bybee's Ch. 19; the variationist approach is discussed by Guy in [End Page 209] Ch. 8).8 The emergence of variation, as Nichols shows in Ch. 5 on 'Diversity and stability in language', is related to how stable or unstable elements within a linguistic system are, and also on the general borrowability of elements, which in turn may depend on the intensity of contact. (For the latter see also Ch. 23 by SARAH GREY THOMASON and Ch. 24 by WALT WOLFRAM and NATALIE SCHILLING-ESTES.)
Does this also mean that there are no linguistic universals? That depends. Universals based on general cognitive principles shared by all human beings most probably do exist, such as the so-called ease-of-effort principle (the principle of economy),9 or the use of analogy that enables language users to find patterns in the surrounding data. It must be clear that unidirectionality is not a cognitive principle; speakers do not order data on the basis of previous historical developments. 10 Because speakers cannot change language at random but most obey conventions of the system in order to keep communication possible, however, it is also evident that the system of language itself, as it has developed over the years, is an important force in the way forms are replicated by speakers. Due to the continual variation existing among linguistic forms in the PLD offered to speakers (variation due to dialect mixing, generation mixing, style mixing, changes in frequency, etc.) and shifting frequencies in these variants, speakers (and this is true for children acquiring the language and building up their grammar, as well as for adults) may discover new patterns by the force of analogy and the need for economy, and these patterns by analogical extension may then slowly affect the language output as well as the underlying system that speakers derive from it.
Another aspect of historical linguistics that is considered very extensively in the introduction is the difference between 'innovation' and 'change'. Innovation is the beginning of a change and involves a replacement made by an individual. Only when more individuals replicate this innovation can it be called a change. The editors emphasize (see also Thomason's Ch. 23:687) that a change should be seen as the result of a spread of an innovation through the community by social forces (speakers imitating others for reasons of prestige, group awareness, etc.). I think that too much emphasis may have been put here on change as SOCIAL spread, on an innovation spreading like a blot of ink among speakers. Some innovations may well be the result of instabilities in the system, and they may therefore start with numerous individuals at the same time. It is possible that phonological change is more likely to be spread in such an external, social fashion, whereas analogical pattern shifts in syntax or morphology are perhaps more likely to be linguistically internal. Similarly, it is possible that the former are more likely to be spread by adults, whereas the latter may well arise more easily, or be speeded up considerably, during the period of language acquisition when children are still forming the patterns of their grammar. Clearly the whole question of innovation vs. change is intimately tied up with the locus of change: adults or children.11 Here again theoretical models differ considerably, as the plurality of voices [End Page 210] in this volume show. What has not been mentioned in connection with innovation and change is the distinction made by Keller (1994) between 'final' or intentional innovations brought about by individuals, which may accumulate into a change that was not intended and that could be called a 'causal' phenomenon. If Keller is correct, then not all change can be said to be the direct result of speakers as agents, as mentioned by the editors. In this sense the system too may play a larger role than is admitted here.
Quite clearly, enough remains to be discovered. This handbook provides a very good starting point in this respect because it is critical and at the same time broadminded. It points the way, but also makes sure that different theoretical standpoints are highlighted. It can be recommended to any aspiring historical linguist fresh to the field, but also to every old hand among us, whose model-determined way of thinking may need to be shaken up a little.
Universiteit van Amsterdam
1012 VT Amsterdam
1. The chapters are organized into seven parts that deal either with change on the different levels of language (phonology, morphology, etc.—Parts 3–6), or with methodologies (Part 1) and explanation (Part 7). Together these chapters reconsider and sum up the state of the art in historical linguistics, which should be the main aim of any handbook.
2. 'Speakers' here includes also 'hearers'. The role of the latter is often neglected, quite undeservedly, as JOHN J. OHALA shows in Ch. 22 in the mechanisms of phonetics. That listening is very important for language interpretation and change must also be clear from the fact that children learn language in the first place by LISTENING.
3. BRIAN D. JOSEPH, in Ch. 13 on 'Morphologization from syntax', also follows formal linguistics in this respect.
4. Unidirectionality is an important notion in historical linguistics and as such it is surprising that the term has not been given more prominence in the index; it can only be found under 'grammaticalization'. Generally, I find the index not too easy to use since many other terms important in historical linguistics, such as the principle of 'economy' or of 'distance', notions such as 'ease of effort', 'markedness', 'universals', and so on are not given as main headings even though they are all discussed in various places in the volume. They may be represented as subheadings, as is the case with 'unidirectionality', but this is not easy to find out.
5. Ohala provides a very interesting discussion of dissimilation, which makes it much less strange as a sound change than hitherto acknowledged. He shows that it is the result of 'hypercorrection' when listeners 'inappropriately apply... corrective processes' (Ch. 22:678).
6. For the view that grammaticalization is in fact an epiphenomenon, see the articles in the 2001 special issue on grammaticalization in Language Sciences 23.2–3 and the studies referred to there.
7. Examples of such clines are a noun > adverb > preposition > affix cline, as discussed in this volume by Heine (Ch. 18:589) and by Traugott (Ch. 20:627–29).
8. Pintzuk (Ch. 15) discusses syntactic variation in connection with parameter settings but here the variant patterns are more abstract than in Guy's more token-directed approach. This is not surprising in that Guy deals with phonology and Pintzuk with syntax. Still, for Pintzuk competition basically involves only two broad types, that is, head-initial vs. head-final structures within the IP.
9. In their interesting discussion of 'uniformitarianism' (23–37), Janda and Joseph show that this principle should be brought down to a much more general 'theorem' of 'informational maximalism' (35)—itself an off-shoot of the principle of economy—because only the uniformity of law and the uniformity of process are valid aspects of this 'principle'; it cannot be applied to phenomena. They also argue, since 'uniformitarian-ism', or rather, 'informational maximalism', involves 'a heavy concentration on the immediate present' (37) that it might be better for historical linguists to rely much more on (variation in) the present in order to acquire insight into how languages develop. This is illustrated convincingly by Ohala in his explanatory paper on 'Phonetics and historical phonology' (Ch. 22), where he shows the advantages of studying sound change in the laboratory.
10. This statement must be toned down a little bit since Guy shows that 'speakers do have an awareness of what is old and new, what is archaic and moribund in their language versus what is fresh and expanding' (Ch. 8:398); this is due to the social distribution of an innovation.
11. Aitchison (Ch. 25:737–40) provides an extensive discussion of the controversy about the locus of change, that is, child vs. adult.