THE COMPLEXITIES OF ARGUING ABOUT COMPLEXITY

Wouter Kusters

Leiden Centre for Linguistics (ULCL)
P.O. Box 9515
2300 RA Leiden
The Netherlands

Pieter Muysken

Dept. of General Linguistics and Dialectology

P.O. Box 9103
6500 HD Nijmegen

The Netherlands

John McWhorter is to be commended for keeping the issue of complexity of creoles on the research agenda, and putting it squarely in the arena of linguistic typology and language change, where it belongs. Particularly insightful is the explicit comparison between a creole and an analytical language like Lahu. However, our task here is to raise critical points rather than simply concur, and there are a number of issues that come to the fore.

First, the issue of time. Languages with histories of thousands of years of uninterrupted transmission and ongoing elaboration are contrasted with recent languages (creoles). However, for the vast majority of the world's languages we simply do not know how long ago they emerged and took on their present form. Current work by one of us (Muysken) suggests that e.g. the past two millennia  were certainly ones of upheaval and frequent restructuring for many languages of the Andean/Amazonian fringe, highly complex with McWhorter's measures. Furthermore, there is no independent evidence on how long it takes for a language to become complex: is there a crucial difference between 300 and 600 years here? In the absence of such evidence, it is vacuous to make claims about development. Furthermore, it is questionable to contrast historical objects such as creoles with a-historically viewed other languages.

A second issue is the source of complexity in language, which McWhorter, following earlier work by Bickerton, attributes to "baroque accretion". McWhorter's perspective is a functionalist one: language is there to convey meaning and complex features of individual languages which do not contribute to this are historical accidents. This perspective leaves much to be desired. Carstairs-MacCarthy (1999) draws attention to the fact that many languages have complex systems of conjugation classes, which show a remarkable persistence over time (cf. also Kusters, in prep.). According to this author, such classes have a basis in the evolutionary basis of our basic language capacities. One need not agree with Carstairs-MacCarthy's explanations, but the very stability and vitality of these systems runs counter to the idea that they are simply the debris of history.

A third issue was raised by Lightfoot (1979) in a review of Li (1977), in terms of Ebeling's Law. David Lightfoot attributed to the Dutch slavicist Carl Ebeling the observation that the less we know about a language, the more regular its phoneme system appears. The sad fact is that we know very little about most creole languages. One of the few good descriptions of a creole is Kouwenberg's (1991) study of the moribund Berbice Dutch Creole, of which only the rudiments were still available at the time of description. Certainly there is no description of Saramaccan available with anything like the level of detail of Matisoff's (1973) study of Lahu. Any time we do work on Saramaccan or on its sister language Sranan, new facts are discovered. Recent work on the supposedly well-described Hawaiian Creole English TMA system by Vellupilai (in prep.) is unearthing completely new facts about the language. Altogether, creole languages were often assumed to be instantly known by observers in the colonial era and theoretical linguists in the post-colonial era, due to their European lexicon and simple root shapes. This has stood in the way of serious description. Methodologically, Ebeling's Law would imply we need to compare two systems which are equally well-described.

Where, given the eloquently phrased contrastive analysis by McWhorter of Saramaccan, Tsez and Lahu, could the complexity of creoles lie? Clearly not in the phonological shapes and morphological endings (although there is some of that, too), but in the complex phrasal compounding, and in the combinatorics of roots. As in English, it is precisely the absence of morphology on roots that makes the system complex.

This last point, lexical complexity, leads us to a final issue: the vague and hybrid nature of McWhorter's complexity measures: half theoretical, half empirical, half system-oriented, half item-oriented. This ambivalence can be illustrated with a number of examples.

1) Germanic V2 is hard to acquire for L2 learners, but perfectly straightforward for L1 learners (Clahsen & Muysken, 1986). While the facts are complex when viewed from the outside, theoretical grammarians have formulated perfectly simple and elegant rules to describe it. Complexity lies to some extent in the nature of the description. A description intended to grasp the acquisition process of an L1 learner leads to a different notion of complexity than a description for the L2 acquisition process, or the description of a language structure itself. These distinctions are blurred by McWhorter's intuitive notion.

2) Similarly, McWhorter claims that Lahu is more complex than Saramaccan in having an overt marker linking causatives to their non-causative counterparts. However, the reverse could also be argued, and has been argued by Mühlhäusler (1974): lexical rules simplify a lexicon.

3) McWhorter’s notion of complexity is not sufficient when comparing, for instance, the following phonological inventories: p, t, k, p^h, t^h, k^h, b, d, g, b^h, d^h, g^h, which can be analysed with the help of four distinctive features, and p, t, k, s, d, k^h, which needs five distinctive features. Are the number of distinctive features decisive here, or the number of phonemes? McWhorter’s reluctance to base his notion of complexity on language processing leaves such questions insoluble.

Thus McWhorter's complexity notion remains half-baked. We think the reason for this is that it is not sufficiently modular. Many linguists would argue that a language is the product (among other things) of the interaction of a lexicon and various rule systems. This perspective would lead us to see where in the trade-off between these subsystems the complexity lies. In McWhorter's undifferentiated perspective, this remains unclear.

It would be nice if there were independent tests for the complexity of a language, e.g. if it were easy for a L2 learner to pass as a native speaker (a measure rejected by McWhorter). On this score, creole languages do not seem encouraging for McWhorter. We do not know whether anybody has claimed to come near to sounding like a native speaker of Saramaccan, but for Sranan even many Surinamese (let alone non-Surinamese) are criticized as non-fluent. For Papiamentu, the same thing: only a handful of non-native speakers are grudgingly acknowledged as speaking decent (mind you, not fluent) Papiamentu.

In conclusion, McWhorter has successfully put the notion of complexity in language back on the research agenda. However, his notion of complexity remains too vague and is intended to cover too much of a grammar to be of much use. It does not tell whether a language will be difficult for an L1 learner or an L2 learner, neither does it evaluate language on anything more than an intuitive level.

References

Carstairs-MacCarthy, Andrew (1999). The origins of complex language: an inquiry into the evolutionary beginnings of sentences, syllables, and truth. Oxford: Oxford University Press.

Clahsen, Harald and Pieter Muysken. (1986). The accessibility of universal grammar to adult and child learners. A study of the acquisition of German word order. Second Language Research 2, 93-119.

Kouwenberg, Silvia (1991). Berbice Dutch Creole : grammar, texts, and vocabulary. Doctoral dissertation, University of Amsterdam.

Kusters, Wouter (in prep.) Complexity in language; its socio-cultural causes and its morphological appearances. Doctoral dissertation, University of Leiden.

Lightfoot, David L. (1979) Review of Charles N. Li (ed.) (1977). Mechanisms of syntactic change. In: Language, vol.55, 381-395.

Matisoff, James (1973). The Grammar of Lahu. Berkeley: University of California Press.

Mühlhäusler, Peter (1974). Pidginization and simplification of language. Canberra: Linguistic Circle.

Vellupilai, Viveka (in prep.) The Hawaiian Creole English TMA System. Doctoral dissertation, Max Planck Institute for Evolutionary Anthropology, Leipzig/University of Nijmegen.