Problems

Grammaticalization theory has been the subject of critical analysis, which centered mainly around two issues, namely its status as a theory and the significance of the unidirectionality principle. We will deal with each of these issues in turn.

That grammaticalization theory does not qualify as a theory has been claimed in particular by Newmeyer (1998a): He observes that this paradigm involves parameters (or mechanisms) relating to different components of language structure, that is, semantics (desemanticization), syntax (decategorialization), and phonetics (erosion), and that none of these parameters or mechanisms is restricted to grammaticalization; rather, each of them is also relevant to other kinds of language change. Accordingly, he concludes that grammaticalization is not a distinct process (Newmeyer 1998a; see also Campbell 2001).

While this observation is in fact true, the conclusion drawn by Newmeyer is not, for the following reason: Grammaticalization theory accounts for the development and structure of functional categories, and to this end it is concerned with the interaction of the three parameters (as well as of extension), rather than with each parameter individually. Therefore, to the extent that these parameters jointly provide a tool for describing and explaining the rise, development, and structure of functional categories through space and time, and to understand why they are structured the way they are, there is a distinct process that can be accounted for by means of a distinct theory. Note, however, that this is neither a theory of language nor of language change; its goal is restricted to the task just mentioned.

The second issue concerns the validity of the unidirectionality principle.¹ Grammaticalization theory relies on regularities in the evolution of linguistic forms and constructions, especially on this principle and the implications it has for the reconstruction of earlier language states. Still, this process is not without exceptions. A number of examples contradicting the unidirectionality principle have been proposed (see especially Joseph and Janda 1988; Campbell 1991; Ramat 1992; Frajzyngier 1996; and especially Newmeyer 1998a: 260V.), usually referred to as instances of ‘‘degrammaticalization’’. A number of these examples, however, are controversial or have been shown by subsequent research not to contradict the unidirectional principle (see e.g. Haspelmath 2004b; Andersson 2005). While we lack reliable statistics, two authors have come up with more specific estimates on the relative frequency of grammaticalization vis-a`-vis other processes. Newmeyer (1998a: 275–6, 278) argues that only about 90 percent of grammatical changes are in accordance with the unidirectionality principle (i.e. are,, downgradings‘‘in his terminology), and Haspel-math (1999a: 1046) says that about 99 percent of all shifts along the lexical/functional continuum are grammaticalizations. On a conservative estimate then, at most one tenth of all grammatical developments can be suspected to be counterexamples to the unidirectionality principle.

Three different stances have been taken to deal with ‘‘degrammaticalization’’. First, it has been argued that since there are some cases of ‘‘degrammaticalization’’, the unidirectionality hypothesis is false (Camp bell 2001). Second, this hypothesis is largely though not entirely true; it takes care of a robust tendency of grammatical change (Haspelmath 1999a; Hopper and Traugott 1993, 2003). And third, the hypothesis is true and cases of presumed ‘‘degrammaticalization’’ can be accounted for by means of alternative principles. Principles that have been invoked are on the one hand morphosyntactic, like exemplar-based analogical change (Kiparsky 2005), and on the other hand cognitive and communicative forces, like lexicalization, euphemism, exaptation, and adaptation as proposed by Heine (2003).

Considering that no instances of ,,complete reversals of grammaticalization‘‘ have been discovered so far (Newmeyer 1998a: 263), we take the unidirectionality principle to provide a solid basis for linguistic reconstruction. In this respect, grammaticalization theory is comparable to the comparative method, which has also been used extensively for linguistic reconstruction on the basis of crosslinguistic regularities to be observed in sound change: Depending on what kind of evidence one decides to draw on, it seems safe to say that the quantitative magnitude of irregularities in sound change and sound correspondences is not drastically different from that to be found in grammaticalization.

But there is one difference between the two methodologies which is crucial: While both draw on attested cases of language change that occurred roughly within the last five millennia, this exhausts the potential of the comparative method. Grammaticalization theory does not have this limitation: By extrapolating from findings on regular grammatical development to situations of unattested stages of language development, it is able to reconstruct earlier processes of grammatical development—that is, phases in the evolution of human languages that are not accessible to other methods of historical linguistics.

Newmeyer’s reservations But the potential our framework has for reconstructing early language has not received full recognition in the specialized literature. To our knowledge, this framework, first presented in Heine and Kuteva (2002b), has not been challenged so far—with one exception: In his seminal papers on uniformitarianism in linguistic evolution, Newmeyer (2002: 366, 2003: 63) argues that there is a problem with this framework, which he describes thus:

The entire progression from full lexical category to affix can take fewer than 2,000 years to run its course. As they [= Heine and Kuteva 2002b; a.n.] note, if there were no processes creating new lexical categories, we would be in the untenable position of saying that languages remained constant from the birth of Homo sapiens until a couple of millennia ago at which point the unidirectional grammaticalization processes began. (Newmeyer 2002: 366)

There are a number of questions raised by this passage. First, to our knowledge, neither Heine and Kuteva (2002b) nor any other student of grammaticalization has ever proposed generalizations on the creation of new lexical categories, or argued that languages remained constant in any phase of their history. Second, there is no reason why unidirectional grammaticalization processes should have begun a couple of millennia ago. There is sufficient evidence, for example from Hittite, Ancient Egyptian, Akkadian, or Chinese (Deutscher 2000; Heine and Kuteva 2002a), to show that these processes occurred much earlier and were of the same kind as we observe them today. Third, there is also no reason to assume that the processes creating new lexical categories were any different two or five thousand years ago from what they are today—again, there is documented historical evidence. To conclude, on the basis of findings on grammaticalization we do not see any analytic or theoretical reason for dividing the evolution of human language into two phases: one that covers the last two thousand years, or any larger period for that matter, and another for the rest of the evolution.²

Even if we were to adopt the position that prior to the last two millennia, or at some earlier point in time, there were no processes of creating new lexical categories, this would not mean that there was no grammaticalization or, if there was, that this would have had serious implications for the structure of the lexicon. Our concern here is not primarily with the lexicon; however, a few general remarks on the role of the lexicon are in order. When a verb is grammaticalized to, say, some tense marker then this need not, and frequently does not lead to the loss of that verb; cf. Hopper’s (1991) principle of divergence. Thus, the grammaticalization of keep to an aspectual auxiliary did not have any implications for the lexical structure of English—keep is still available as a lexical item. In a similar fashion, the English possessive verb have (19a) has given rise to a number of grammaticalizations: It developed into a verbal aspect marker roughly a millennium ago (19b), and it also provided the conceptual template for a number of additional grammaticalizations, like that as a functional category of deontic modality (19c). Nevertheless, have is still available as a possessive verb in present-day English, that is, these grammaticalizations did not affect the English lexicon in any significant way.

(19) English

a. They have no children.

b. They have left.

c. They have to pay.

To be sure, it may happen that with the grammaticalization of a given lexical item, that item will disappear from the lexicon—English will being a case in point: The verb of volition, from which it is historically derived, was—except for a few relics (e.g. Do as you will)—lost after it developed into a future tense marker; in fact, it is more likely that a lexical verb that has undergone grammaticalization is lost than one that has not.³ But it does not seem possible to use this observation as a basis for generalization: Lexical loss also happens quite frequently without grammaticalization being involved; English has lost a wealth of verbs on the way from Old English to Modern English without there having been any kind of grammaticalization that could be held responsible for this fact. In other words, there is no significant relationship between the creation of new functional categories and that of new lexical categories.

More recently, Newmeyer (2006) has come up with further problems he has with applying grammaticalization theory to the reconstruction of early language. These problems are:

(a) Grammaticalization (Newmeyer argues) is a cyclical process, and why should one pick one point on the cycle, namely the point where lexical items are in place, as the starting point.

(b) There is no reason to assume that the earliest humans could not express concepts like ‘‘in’’ and ‘‘past time’’, and he wonders why categories such as prepositions and tense and aspect morphemes could not have existed at the outset of human language as independent categories.

(c) If a modern language like Riau Indonesian ‘‘can manifest grammaticalization as poorly as a language spoken 100,000þ years ago putatively did, then it follows that grammaticalization per se cannot tell us very much about the origin and evolution of language.’’

(d) If what is frequently expressed has changed over time, or if the balance of functional and counter-functional factors has not remained constant over time, then the process of grammaticalization might lack sufficient unidirectionality (or at least consistency).

One may wonder whether Newmeyer’s worries are really justified. One worry relates to his assumptions about grammaticalization. With reference to (a), this concerns the notion ‘‘cyclical process’’. Grammaticalization may in fact be cyclic,⁴ but it need not be. To take the example just looked at: As (19) shows, the erstwhile English verb have has been grammaticalized— among others—to a perfect aspect and a modality auxiliary. There is no evidence to assume that this was due to a cyclical process. What we know for sure, however, is that this process started out with a lexical verb of possession, hence there was an unambiguous starting point. And the same applies to the other cases of auxiliaries that we discussed in “The present approach”: Irrespective of whether or not there was, or is, a cyclical process, the ultimate source of functional categories can be traced back in some way or other to lexical categories.

Problem (b) concerns the method of reconstruction. If Newmeyer believes that categories such as prepositions or tense and aspect markers could have existed at the outset of human language then one would like to know what the methodological basis of this belief is. We are not aware of any possible basis other than that provided by grammaticalization theory (see “Methodology”). And according to this theory, neither prepositions nor tense and aspect markers are diachronic ‘‘primitives’’ (see “The fourth layer: demonstratives, adpositions, aspects, and negation”): Both are historically derived from lexical categories. For example, in “Desemanticization” we mentioned the English prepositions in front of and in back of, which are both of nominal origin, and the tense and aspect auxiliaries keep, used to, and be going to, all of which are historically derived from verbs, and much the same observations can be made in many other languages. Accordingly, Newmeyer’s belief must of necessity remain speculative as long as it is not supported by any method of reconstruction. To be sure, there are a number of functional items for which no reliable etymology exists; but we do not see any reason why they should behave diachronically any differently from those for which there is sufficient diachronic information.

With (c), Newmeyer raises a problem that is beyond the scope of grammaticalization theory. As we hope to show here, grammaticalization theory offers an account of grammatical evolution from early language to the modern languages; but it does not tell what happened thereafter in the history of modern languages. Riau Indonesian in fact manifests grammaticalization poorly: It is said to lack grammatical distinctions normally found in other languages (Gil 2001). But why this is so is open to question. That a language loses many of its grammatical distinctions in the course of its history is nothing unusual, French and English being cases in point: Both lost a large part of their inflectional and derivational morphology in the course of the last two millennia. And what led to the situation characterizing Riau Indonesian is also a question that concerns the particular history of this language rather than grammaticalization theory, which is concerned with regularities of grammatical change across languages. Assuming that a language such as Riau Indonesian, where it is not even possible to distinguish—linguistically—between nouns and verbs, let alone grammaticalized structures, represented an earlier stage of language evolution, this would in no way present a challenge for grammaticalization theory. On the contrary, such might be a language which can be used as living proof that the kind of system we arrive at, going back in time when we use grammaticalization theory as a reconstructing tool, is something not only plausible but also very possible, since it exists as part of our present linguistic diverse reality.

The nature of the last concern (d) would need further analysis; but irrespective of what the nature of the counter-functional, or counter adaptive factors may be,⁵ grammaticalization is overwhelmingly unidirectional and regular, as Newmeyer (1998a: 275–6, 278) himself admits (see above); hence, it provides an appropriate basis for linguistic reconstruction. To conclude, we do not see how Newmeyer’s reservations could affect the application of grammaticalization theory to the reconstruction of early language.

¹ For an explanation of unidirectionality in terms of ‘‘rhetorical devaluation’’, see Haspelmath (1999a: 1059–61).

² To be sure, future research might reveal that there is justification to divide the evolution of grammar into two salient phases, say, an earlier phase characterized by some specific state of grammaticalization and a later one showing a different state of grammaticalization; for the time being, however, we see no reason to assume such a division.

³ With reference to the development from verbs for ‘say’ to complementizers in Chadic languages, Frajzyngier maintains: ‘‘The development of verbs of saying into complementizers resulted in the bleaching of the lexical content of these verbs. As a result, new verbs of saying were introduced in many Chadic languages. This explains the large number of lexical innovations in the verbs of saying as compared with the large number of retentions for those verbs that have not served systematically as sources of grammatical morphemes’’ (Frajzyngier 1996: 470).

⁴ This term refers to the fact that grammaticalization can be cyclical, leading from lexical to grammatical category and finally to the loss of the latter, and there may be renewal, in that the lost category is replaced by another lexical category (Givo ´n 1979c; Heine and Reh 1984).

⁵ Newmeyer cites Haspelmath (1999c) in support of this assertion.

اخر الاخبار

اخبار العتبة العباسية المقدسة

تستمر طيلة شهر رمضان.. إقامة ختمة قرآنية تعليمية في مقام الإمام المهدي (عجل الله فرجه)

في مستشفى الكفيل.. جراحة دقيقة بين الأذن والدماغ تنقذ مريضة من التهاب حاد

المجمع العلمي يُطلق مسابقة (فرسان التنزيل الفرقية الأولى) ويقيم ختمات قرآنية في محافظة المثنى

ضمن أنشطته الإنسانية.. قسم الشؤون الفكرية يفتتح البئر الـ 72 في نيجيريا

المزيد

الآخبار الصحية

دراسة تحذر من الإفراط في ممارسة الرياضة.. كيف يضر دماءك؟

الطهي بزيت الزيتون.. مضر أم مفيد؟

دماغ الرضع في خطر.. دراسة تحذر من "العدو الخفي"

التمر بالحليب عند الإفطار.. "كنز صحي" في رمضان

المزيد

الآخبار التكنلوجية

أثر نفسي غير متوقع للرياضة.. مرتبط بالغضب

القمر يواصل الانكماش.. رصد تشققات حديثة على سطحه

إطلاق أول تطبيق مراسلة "مشفّر".. ثورة في عالم الدردشة

خلل تقني يؤجّل أول رحلة مأهولة إلى القمر منذ عقود

المزيد

مواضيع ذات صلة

Marked coding: the passive construction

Case

Grammatical functions and transitivity

Valence at the clause level

Clauses

An outline of grammatical evolution Introduction

The present volume

Grammatical morphemes and agreement

Constructional schemas

Composite structure

Prototypical stems and affixes

Semantic autonomy and dependence