This book presents a new theory of grammatical categories - the Universal Spine Hypothesis - and reinforces generative notions of Universal Grammar while accommodating insights from linguistic typology.
Date: Mon, 31 May 2004 19:50:29 -0500 From: Mike Maxwell <email@example.com> Subject: Twice as Meaningful: Reduplication in Pidgins, Creoles
EDITOR: Kouwenberg, Silvia TITLE: Twice as Meaningful SUBTITLE: Reduplication in Pidgins, Creoles and Other Contact Languages SERIES: Westminster Creolistics Series - 8 PUBLISHER: Battlebridge Publications YEAR: 2003
Mike Maxwell, Linguistic Data Consortium / University of Pennsylvania.
For those who speak languages like English, reduplication seems at the same time exotic, rare, and (for linguists, at least) fascinating. Fascinating it may be, but at least in Creole languages it is not rare. This volume brings together a wealth of studies on reduplication in Creoles.
The relevance of this book for creolists and for morphologists interested in reduplication should be obvious. Lest other readers turn to other matters, at the end of this review I will suggest some reasons this book may be important for language documentation and description.
Most of the papers here were originally presented at the 4th Westminster Creolistics Workshop on reduplication in contact languages, in 1999. But there appears to have been a great deal of revision between workshop and publication, and a few of the chapters have other sources.
The editor, Silvia Kouwenberg, does an excellent job of setting the stage for the collected papers. First, she notes that defining reduplication is not as simple as it might seem. She proposes a working definition, but notes that this very definition is in need of study. (And I would add that it may in part be a theory-internal question.)
With this caveat, Kouwenberg outlines the issues discussed in the remainder of the book:
o whether the substrate languages are the source of the reduplicative processes in the daughter languages, or reduplication arose independently in creoles;
o reduplication in creoles vs. pidgins;
o the form of reduplication in creoles (complete vs. partial reduplication);
o the meaning of the reduplicated forms.
To attempt to summarize and comment on all 34 papers would result in a truly boring review. It should suffice to say that while there is the occasional misunderstanding of some theoretical point, or evidence which is not quite strong enough to support the conclusion that an author attempts to draw, I did not find a single paper which did not make some interesting points. And while this is not really a comment on the value of the papers, I found it refreshing to read a linguistic analysis of other languages in which I could more or less understand the example sentences without laboriously working through the word-by-word translations. For instance, example (38a) from Hubert Devonish's article on reduplication in Guyanese Creole:
Mi sii wait daag, wait daag, wait daag, aal oova di plees. ''I saw large numbers of white dogs everywhere.''
(Not all the languages are English-based creoles, of course!)
So instead of reviewing each paper, I will try to summarize some of the main issues which the papers focus on, including the issues mentioned by Kouwenberg (above).
I begin with the genesis question: is reduplication in creoles inherited from the languages out of which the creoles arose, or did it arise independently?
One form of the independent origin theory would say that reduplication arose from the innate language learning program, operating not just to support learning of a language, but in the creation of a new language. Demonstrating this, however, requires demonstrating the implausibility of other explanations for the origin of reduplication in the creoles.
Naively, this might seem an easy question to answer: if reduplication happens in the same way, performing the same function, in the substrate language and in the creole, then surely the simplest explanation is that reduplication in the creole was borrowed from the substrate (or substrates). And indeed one can point to similar kinds of reduplication in many of the predecessor languages (e.g. the West African languages spoken by the presumed ancestors of the later speakers of Caribbean creoles).
Nevertheless, it is not at all clear that the reduplication processes found in the creoles were inherited. The functions of reduplication (and in many, if not most, of languages, both creole and substrate, reduplication plays many roles) are not an exact match, and the form of reduplication is often quite distinct. (More on the latter below.) Moreover, the function of reduplication is so often iconic (bearing meanings like ''more of'', ''repeated action'', etc.) that even where functions match between substrate and creole, one is left wondering whether this is simply coincidence.
Given that reduplication is so often iconic, the best evidence for transfer might come from non-iconic reduplication, provided this can be related to similar non-iconic reduplication in one of the source languages. An example might be the use of reduplication to indicate attenuation (such as 'redi-redi' for ''a little red'', an example in Parkvall's paper ''Reduplication in the Atlantic Creoles''). But in the end even this is questionable; relatively few productive uses of reduplication are truly non-iconic, and for those that are non-iconic, the correspondences between reduplication in substrate and creole often seem shaky.
Concerning this last point, Bakker writes (pg. 79):
Reduplication exists in (almost?) all West African language families, although its semantic functions vary widely. It should be easy to find productive process of reduplication in West African languages that are similar to [Saramaccan] reduplications.
I suspect that it may be all too easy, so that any similarity between the function of reduplication in Saramaccan and a West African language might be purely coincidental.
But suppose for a moment that reduplication in creoles is in fact carried over from pidgins. One might then hope that the mechanism of this transfer would be discoverable. In particular, if one assumes that creoles arose from the substrate languages by way of an intermediate pidgin, then one might suppose that reduplication entered the creole by way of that intermediate. And until recently, it was believed that reduplication was commonplace in pidgins, allowing for just such an uninterrupted trail. Un- fortunately for this theory, Bakker shows in his paper (and several other authors further substantiate the point) that contrary to the received notion, there are very few attested instances of reduplication in (true) pidgin languages. Assuming Bakker to be correct--and his evidence seems quite convincing, once languages which have been called pidgins but which for one reason or another do not fall squarely into this category (I will resist the temptation to say 'pigeonhole') are removed from the sample--the presumed simple chain of transfer is broken.
However, it seems quite possible that creoles and other languages having reduplication co-existed for a period in an adstrate relation, and that bilingual creole speakers borrowed reduplication from the latter. Something very much like this situation existed (and still exists) in some of the Creole Englishes of Liberia, one of which Singler discusses in his article. Unfortunately for our purposes, this is not a good test case for the borrowing theory, since true reduplication turns out not to be productive in this language. The situation is clearer for some of the Indo-Portuguese Creoles, where borrowing of reduplication from the adstrate languages seems indisputable. This does not of course establish that this form of inheritance accounts for all cases, but it is at least suggestive.
The form of reduplication is given some attention in many of the papers, although for some of the languages with partial reduplication, I would have preferred more details. In particular, there are a few cases where the rules or generalizations do not seem to match the examples.
Whole word reduplication is common in the creoles, almost to the exclusion of partial reduplication, whereas partial reduplication seems the norm in ''older'' languages, including many of the substrate (and adstrate) languages where these are known. The difference in the form of reduplication again makes the case for a substrate source of reduplication again seem weak (and it might argue against an adstrate source, although this is perhaps less clear).
But this difference between the form of reduplication in the substrate or adstrate language and in the creoles seems to me to be one of the more interesting stories these studies have to tell; for it seems at least plausible that we have found reduplication in its basic form, and that we can track its historical development from the beginning stages of iteration (repetition, presumably at the syntactic level), through its early morphological stages (full word reduplication), and on to various sorts of partial reduplication--if not in a single language, at least in related languages. Indeed, several of the creoles show what look like early stages of partial reduplication, attributable to the differential affect of phonological rules on base and reduplicant, or length-based effects (such as reduction of unstressed syllables).
One of the generalizations that one is tempted to make, is that repetition consists of an essentially unbounded number of copies of a word (think of ''big, big, big!'') or even a phrase; whereas reduplication is usually grammaticalized to allow a single, not necessarily complete, copy of the base. Intriguingly, one paper (that of John Ladhams, Tjerk Hagemeijer, Philippe Maurer, and Marike Post) alludes to what may be a counterexample: multiple iterations with only partial copying. Unfortunately, no data are given.
The meaning of reduplication also receives a great deal of attention. I mentioned earlier that the this meaning is often iconic. But there is a substantial range of these iconic meanings, as well as many non-iconic (or even seemingly anti-iconic) meanings, and most of the authors go to great pains, within the limits of article length, to elucidate and illustrate this. In sum, there is a wealth of information on the meaning of reduplication here.
At the beginning of this review, I suggested that this book may hold interest for language documentation and description, that is, for efforts to preserve for future generations the properties of languages which may some day go extinct. And extinction is a possibility for nearly all the languages of this book--indeed, for Limonese Creole (of Costa Rica), two of the authors report that the language is already experiencing both loss in number of speakers, and loss in terms of the usage of reduplication among the remaining speakers. In my opinion, reduplication is the morphological process that can tell us the most about the nature of both morphology and phonology. It is bad enough when a language becomes extinct; speaking as a linguist, it is even worse when that language holds clues that would help us in a unique way to unlock one of the mysteries of human language. The papers in this book, then, serve an especially important role in documenting a particular aspect of languages, many of which may become extinct, and the editors are to be commended for encouraging this.
However, I believe there is more that this book can teach us about language documentation than just the familiar precept ''document before extinction.'' The terms ''language documentation'' and ''language description'' cover a wide range of practices, from simple (if extensive) transcription of texts at one end, to more or less detailed grammatical and lexical analysis at the other. The argument in favor of the former is that there are so many endangered languages that there is time for only text collection. The argument in favor of the latter is that detailed analysis may uncover rare but interesting phenomena, and that unless further data collection is guided by analysis, there may be insufficient examples of these rare phenomena in the corpus for future linguists to be able to tell what is going on.
The data collection methodologies in this book span this range of practices: Nordlander and Shrimpton explore reduplication in a comparatively small (and translated) corpus, namely the New Testament; and at several points, they are forced to refrain from making stronger generalizations because of the sparseness of the data. Gooden's article, on the other hand, benefits from elicitation of acceptability judgements from native speakers, including the author himself, allowing for a correspondingly more detailed analysis. (Dictionaries are a third commonly cited source of information, although I question whether lexicographers will document productive reduplication.) My point is not to that Gooden's methodology is better than that of Nordlander and Shrimpton (both have their place), but rather that the range of studies in this book of a single phenomenon, reduplication, provides a unique window into the strengths and weaknesses of the various methodologies.
(For the record, the studies do show that even substantial corpora leave unanswered questions. Particularly unclear in the corpora-based studies is the productivity of reduplication. This can be a crucial question: since morphology, unlike syntax, can in principle be simply memorized, one should hesitate to base ones theory on a process which may turn out to be unproductive. My conclusion is that if you are documenting a language which has reduplication, you should do more than collect corpora.)
Finally, I wish to add what may seem like a trivial comment, about the cover. Finding appropriate illustrations for covers of linguistics books must tax the abilities of whoever does this. In the case of this book, however, the illustration is wonderfully appropriate. (And no, I won't give it away!)
ABOUT THE REVIEWER:
ABOUT THE REVIEWER Mike Maxwell works on morphological processing and language resource collection at the Linguistic Data Consortium. He has also studied indigenous languages of Ecuador and Colombia, under SIL.