linguistlist.org>
Editor for this issue: T. Daniel Seely <seelylinguistlist.org>
eduserv.its.unimelb.edu.au>Here follows the slightly belated summary of comments I received on my query on the development of copulas from pronouns, on Linguist List. I must say I have been heartened by the number and volume of responses I received; I dare say my faith in the Internet as something other than an utter waste of time has been restored! :-) My thanks to (in order of appearance in my mailbox) Larry Trask, David Solnit, Paul Hagstrom, Carsten Peust, Sasha Vovin, Mikael Parkvall, John Verhaar, Cathal Doherty, Gregg Kinkley, Marina Yaguello, Jie Li, Marjory Meechan, John Koontz, Edit Doron, Aya Katz, David Gaatone, Maik Gibson, George Huttar, Eloise Jelineke, Rene Kriegler, Laszlo Cseresnyesi, and Vincent DeCaen. Regrettably, I won't have time to chase up all the tracks pointed out to me by my colleagues right now, but this is an interesting topic, and might make for a good dissertation somewhere :-) . Before I proceed, let me outline my motivation in making this query. The immediate cause was Suzette Haden Elgin's criticism of the inventor of Klingon, Marc Okrand, for describing the pronoun in attributive predicates, like tlhIngan ghaH "Klingon s/he", as a verb, when this is (ostensively) merely a zero-copula construction, something rather commonplace in human language. Haden Elgin goes so far as to accuse of Okrand of "linguistic malpractice" against the wide target audience he could have potentially instilled an appreciation of grammar in. However, the Klingon pronoun can bear the full complement of verbal suffixes in the language, including aspect markers, negators, modal suffixes, and evidentials. The only affix one can be reasonably sure (from Okrand's erratic description) that it cannot bear are verbal agreement prefixes. In that respect, referring to the pronoun as a verb makes a good deal of sense (particularly since the corresponding topic-comment attributive predicate construction, through accident more than design, has ended up reanalysed as a predicate-subject construction): in these constructions, the pronoun behaves morphologically, and arguably syntactically, as a highly suppletive copula. The question which occurred to me then was, how cross-linguistically common was it for a pronoun to be reanalysed into a fully verbal copula. While the zero copula strategy is extremely common-place in language, I was surprised to find no mention of pronoun-to-copula transitions in Comrie's _Language Universals and Linguistic Typology_, or Croft's _Typology and Universals_ and _Syntactic Categories and Grammatical Relations_. There were a couple of reasons I suspected the full transition from pronoun to verbal copula might not be that commonplace. While syntactically one does not have to go far to reanalyse a topic-comment construction like "Comrie, he a linguist" to a subject-predicate construction like "Comrie COPULA a linguist", the pronoun is in some respects a fairly prototypical nominal; and particularly if the pronoun in question remains functional as such in a language, it might be difficult for it to acquire overt verbal morphology --- a major analogical leap is required. Note that, for that reason, it was particularly morphological criteria that I was interested in, rather than syntactic or semantic criteria (like lack of referent agreement), which would fall out rather naturally from the syntactic reanalysis of pronoun to copula. Admittedly, not all languages *can* supply morphological evidence of copula-hood (e.g. Chinese), but what I want to establish is whether the full transition is possible. Furthermore, many languages avoid having to hook verbal morphology onto a zero copula, not by tacking it on to the pronoun, but by retaining copulas in semantically marked contexts, such as the past tense (eg. Russian byl) --- as pointed out by Li & Thompson (1977:436). In that light, I group the responses I received into various groups. The first involves zero-copula languages in which the pronouns are only starting to embark on the road to becoming copulas. The second involves pronouns which have picked up syntactic and semantic hallmarks of copula-hood, but not morphological characteristics. The third involves pronouns which are morphologically verb-like, and can be argued to have become fully copulas in the framework I am talking about. Of course, one cannot afford to be too schematic about these matters; as John Verhaar pointed out to me, one is better off thinking of copulas cross-linguistically as forming a continuum, a cline ranging from 0 through nonverbal copulas (e.g. Indonesian adalah) to verbal intransitive copulas to verbal transitives (e.g. English "this *represents* a problem"). Before going on, I should also point out that Aya Katz' doctoral dissertation, _Cyclical Grammaticalization And the Cognitive Link Between Pronoun and Copula_, provides an interesting connection between this topic and grammaticalisation. According to Katz, there is a grammaticalisation pathway from pronoun to copula, and another pathway from copula to pronoun. Indeed, Hebrew displays the full cycle: Pre-Proto-Semitic to Biblical Hebrew has copula to pronoun (Vincent DeCaen is also currently working on this particular point), and the same pronoun (as discussed below) is becoming a copula again in Modern Hebrew. Katz also discusses some of the other cases discussed below: Turkic, Chinese, and Finnish. Katz establishes fairly convincingly, it would seem, the pervasiveness of the pronoun-to-copula grammaticalisation pathway. I. INCIPIENT COPULAS Sasha Vovin pointed out the me the use of promonimal affixes as copulas in Turkic, eg. Osmanli Turkish T"urk-"um "[I] am Turk". I'm not sure what to make of this; this looks a lot like a grammaticalisation arising out of a zero-copula strategy, fusing the subject pronoun into the predicate noun, but the pronoun seems to have been decategorialised, rather than reanalysed as a full copula in itself. Katz considers this an instance of copula to pronoun (if I understand him correctly); this would probably fit the data more smoothly. In another instance of 'right morpheme, wrong syntax' :-) , John Koontz sees in the Siouan stative verb formant *-ka (e.g. *htaN-ka 'be big') a reflex of *ka 'yon' --- although this is not yet universally accepted. In a non-polysynthetic language, this would have developed to a copula; like Turkish, however, it does not represent evidence for the nominal-verbal transition I had in mind. John Verhaar finds an incipient copula in Bahasa Indonesia determiner itu, which also serves to mark topics (indeed, John has found instances of itu used twice after a constituent in his corpus --- first as a determiner, and then as a topicaliser.) Since as a topicaliser itu introduces the predicate in topic-comment constructions, he considers it is not far off from becoming a copula proper itself --- all that is needed is to drop the pause between the topicaliser and the predicate. Irish also seems to have an incipient copula in the use of a pronominal in equative sentences in addition to a non-verbal copula; this is discussed in Doherty (1996); See also Dechaine & Vinet (1992). Rene Kriegler points to a similar situation in Swahili (Ashton 1944:92-94), in which subject pronouns as copulas coexist with zero-copula and uninflected-copula strategies. II. NON-MORPHOLOGICAL COPULAS Charles Li & Sandra Thompson (1977) deal with a number of these in their paper _A mechanism for the Development of Copula Morphemes_. My thanks to Larry Trask for bringing the paper to my attention; I had in fact read through the volume it was published in at the start of my long-suffering dissertation, but the article in question seems to have slipped my memory. Chinese shi4 (reconstructed as d'iek, according to Gregg Kinkley) is a textbook case of the topic-comment to subject-predicate reanalysis I alluded to earlier. Originally a demonstrative, as in (1), shi4 was reanalysed to a copula around 100 AD, as in (2) (examples from Li & Thompson): (1) qiong2 yu4 jian4, shi4 ren2 zhi3 suo3 wu4 ye3 Poverty and debasement, *that* [is what] people dislike (Analects, v BC) (2) zhu1 ke4 yue4, "ci3 shi4 An1-shi2 sui4 jin1" All the guests said, these *are* An-shi's bits of gold (Shi4 Shuo1 Xin1 Xu3, v AD) shi4 as a demonstrative disappeared in Chinese by vi AD. There are no morphological criteria for shi4 being a copula in Chinese; as David Solnit astutely put it, "Chinese being what it is, there is no verb morphology for shi4 to pick up". It does however pass the synactic test of negation by bu4. In Modern Chinese, it can take predicative nominal phrases, but not adjectival phrases. That its grammaticalisation is incomplete is also demonstrated, as Laszlo Cseresnyesi points out, by the fact that shi4 does not take aspect markers in Modern Chinese. While Modern Hebrew is not as far advanced as Chinese in this process, it also belongs under this rubric. (Whether the peculiar circumstances of the renewal of Hebrew are causally implicated in this is a question for the specialists to resolve; I am interested to see how Aya Katz addressed it.) Hebrew does not use the hyy root as a copula in the present tense, but a zero-copula topic-comment strategy with a demonstrative or personal pronoun. It has been argued by Berman & Grosu (1976) and Faltz (1973) that the pronouns are behaving as copulas: Hebrew topics cannot be indefinite, but the subjects of these copulas can be; the copulas can be questioned; and there is no intonation break between the putative topic and the pronoun. Finally, a non-3SG 'topic' can occur with a 3SG-copula in some contexts. (The illustration Li & Thompson gave was a cleft sentence, although David Gaatone gave me the simple sentence ani hu hamelekh 'I am the king' as an illustration; it seems individual judgements vary in Modern Hebrew, and the status of this pronoun/copula is still in flux.) More information on the developments in Modern Hebrew can be found in Doron (1983), Rapoport (1987), and Katz's dissertation; according to Li & Thompson, the facts for Palestinian Arabic are very similar. Carsten Peust informs me Neo-Assyrian Akkadian demonstrative shu came to serve as a copula as well --- I'd have to consult the grammars for confirmation and details, though. Creoles (which, as Mikael Parkvall points out, Chinese resembles in many ways) display similar develops. Mikael pointed to John McWhorter's recent work on Atlantic creoles (da < 'that?'; see the 1996 Journal of Pidgin & Creole Studies, American Speech 4.4 (1995), and his recent book _Towards a new model of creole genesis_), and Marguerite Fauquenoy-St-Jacques' work on French Guiana creole (optional copula sa < Frence cela). George Huttar adds that Jacques Arends has convincingly established the da < 'that' connection for Sranan in the Journal of Pidgin & Creole Studies, and Huttar & Huttar (1994: 132ff) discuss passim the same development in Ndyuka. Finally, indeclinable former demonstratives ce?(e?) in Wappo, and u in Zway, also mentioned by Li & Thompson, also count in this category as copulas not yet displaying verbal morphology (in languages otherwise rich in verbal morphology). III. MORPHOLOGICAL COPULAS The one good example I have found of this is in several dialects of Arabic (including Egyptian, Tunisian, and some Palestinian variants; thanks to Carsten Peust, Eloise Jelineke, and Maik Gibson.) Eid (1983) discusses Cairene Arabic in some detail. Like Hebrew and Palestinian Arabic, Cairene Arabic uses a personal pronoun (huwwaa) in present-tense attributive predicates. The difference here is that Cairene and the other Arabic dialects have a negative circumfix, ma-sh, which can only apply to verbs (and adjectives in Moroccan), and which is applied to huwwaa in such constructions. So Cairene huwwaa is here explicitly behaving like a verb morphologically. Note that Cairene was not 'obligated' to develop verbal morphology for the pronoun, since it also has a negative particle, mish, which can apply to non-verbs and verbs alike. As is frequently the case in language change, one cannot really speak of the development being motivated to fill a gap in the system: 'drag-chain' accounts do not seem to sit well with grammaticalisation. In the same neck of the woods, Old Egyptian demonstrative pw became a copula by the second millenium BC. According to Carsten Peust, "one cannot say that this pw ever accepts verbal morphology", but from 1300 BC on (Late Egyptian) it is inflected in number and gender, although not in the same manner as fully fledged verbs. Number and gender marking may not be as prototypical verbal qualities as negation and tense/aspect/mood, but provisionally I count pw as also a morphologically verbal copula. Wolof has an interesting variant of the pronominal copula (Robert 1992, Yaguello & Diouf 1991). Regrettably neither reference is available in Melbourne, but the basic story is, Wolof (and neighbouring West-Atlantic languages) lack a copula as such, but have a suppletive pronoun paradigm, distinguished by tense, aspect, mood, stativity, and so on, and which behave as copulas when followed by NPs. The three values pronouns can bear when behaving as copulas are identity/belonging, focus, and location: suma warit la: he is my friend (my friend + he) moo suma warit : HE is my friend (he emphatic + my friend) mu ngi fi: he is here (he +locative + here near the speaker) (Thanks to Marina Yaguello and Marjory Meechan for this information.) The acquisition of TMA properties, prototypically verbal, by pronouns, also seems to me indicative of a pronoun-to-copula shift. CONCLUSION Even Blind Freddy (to use an Australianism) knows that there is a well-established pronoun-to-copula pathway, which seems to follow the Chinese route of topic/comment -> subject/predicate fairly consistently. Not all languages have the syntactic/morphological machinery to allow the resulting forms to be 'fully-verbal' in the narrow, morphological sense I adopted --- Siouan and Chinese being cases in point. (The morphological criterion was, of course, partially prompted by the musings of two linguists on categoriality in a language cooked up for Star Trek, and thus are not necessarily compelling for the entire discipline :-) .) Nonetheless, this too occurs, certainly in North African Arabic, apparently also in Late Old Egyptian, and in a sense also in Wolof. And the non-morphological instances are abundant, notwithstanding the frequent TMA restrictions on the pronominal copula. And as is frequently the case with grammaticalisations, in some languages the binary distinction between pronouns and copulas (or if you like, nominals and sorta-verbals) is significantly weakened --- Modern Hebrew seems to provide the best illustration of this here. REFERENCES Ashton, E.O. 1944. _Swahili Grammar_ London: Longman. Berman, R. & Grosu, A. 1976. Aspects of the Copula in Modern Hebrew. In Cole, P. (ed.) _Studies in Modern Hebrew Syntax and Semantics_. Amsterdam: North Holland. Dechaine, R.M. & Vinet, M.T. 1992. Une Structure Predicative sans Copule. _Revue quebecoise de linguistique_ 22.1:11-44. Doherty, C. 1996. Clausal Structure and the Modern Irish Copula. NLLT 14.1:1-46. Doron, E. 1983. _Verbless Predicates in Hebrew_. Doctoral Dissertation, University of Texas. Eid, M. 1983. The Copula Function of Pronouns. _Lingua_ 59: 197-207. Faltz, A. 1973. Surrogate Copulas in Hebrew. MS. (Cited in Li & Thompson) Huttar, G. L. & Huttar, M. L. 1994. _Ndyuka_. London: Routledge. Li, C. & Thompson, S. 1977. A mechanism for the Development of Copula Morphemes. In Li, C. (ed.) _Mechanisms of Syntactic Change_. Austin: University of Texas Press. 419-444. Rapoport, T. 1987. Doctoral dissertation, MIT. Robert, S. 1992. _Approche enonciative du systeme verbal du wolof_ Paris: CNRS Yaguello, M. & Diouf, L.. 1991. _Damay jang wolof_. Paris: Karthala. - O Roeschen Roth! Der Mensch liegt in tiefster Noth! Der Mensch liegt in tiefster Pein! Je lieber moecht' ich im Himmel sein! --- _Urlicht_ n.nicholasMail to author|Respond to list|Read more issues|LINGUIST home page|Top of issue