Startseite / Archive / 2007 / Assibi A. Amidu 2007. Semantic Assignment Rules in Bantu Classes. A Reanalysis Based on Kiswahili.



It seems not too far off the mark to suggest, as Assibi Amidu does, that descriptions of the semantics of Bantu noun class systems have in the past often been crude oversimplifications of the linguistic facts. In this study, which continues a line of investigation initiated in Amidu (1997), Amidu sets himself the laudable goal of accounting for Swahili noun classification and concord on the basis of real usage.


In the work under review as in previous studies, Amidu uses data from Swahili to argue forcefully against simple-minded semantic assignment rules in Bantu noun class systems. As he argues, ‘traditional semantic assigment rules have negative social and ontological implications for grammatical descriptions. Our reanalysis, hopefully goes some way towards eliminating the impression that Bantu speakers have a rather simplistic conceptual view of the world they live in.’ (13)’


Under the rubric of ‘traditional semantic assignment rules’, Amidu groups together so-called pedagogic rules for learners (such as, all animate nouns take their concord in classes 1/2); impressionistic descriptions of what semantic patterns there are to be found in Bantu noun classes (especially those based on simple binary features like animate, arboreal, augmentative, diminutive, elongated, abstract, cf. pp.27ff.); and finally, simplified descriptions of the system by non-Bantuists (e.g. Corbett 1991, Lyons 1968). Throughout the volume, Amidu uses the method of stating a simplistic semantic assignment rule, painstakingly analyzing one or more examples that are problematic for it, and concluding that the rule fails to account for them. In fact, the bulk of the text of the book is taken up by such exercises.


Chapter one, despite being titled ‘Introduction’, assumes full prior knowledge of the workings of the Swahili noun class system and directly plunges into an intricate argument about animacy and the lexical semantics of some animate nouns in class 1/2. Chapter two, elaborating on this, argues that mtu and watu [‘being’ and ‘beings’, md] are the highest generic terms in classes 1/2, and that they have sub-senses that are both [+animate] and [-animate]. The argument here is that this may help account for the occurrence of [-animate] terms in prototypically animate classes and vice versa. Here and in some other places, the argument is slightly marred by the implicit assumption that animacy should be a binary feature. It seems more profitable to think of it as a notion with prototype structure, so that, for example, the occurrence of doll in the prototypically animate class pair 1/2 is not nearly as problematic as Amidu makes it out to be (p.6; see also Amidu 2003:110).


Chapter three, ‘Anomalies of semantic classification’, points out some problems inherent in the view that the noun classes are organized in a neat, logical, and orderly way. It furthermore refines the lexical movement hypothesis (LMH) introduced in Amidu (1997). Briefly, the LMH holds that nouns in any class may be converted into nouns of any other class with or without change of form. This hypothesis is the basis for a dynamic conception of Bantu noun classes in which the conceptual non-homogeneity of class contents is seen as crucial rather than problematic, allowing classes to ‘be elastic and have ability to expand their range of expressions’ (p. 36). This is an interesting thought that merits further elaboration.


Chapter four, ‘What is the domain of semantic assignment rules in Kiswahili Bantu classes?’, argues that simplistic semantic rules not only fail to fully account for class allocation, but also that they cannot form an inductive basis for agreement marking, which is ‘generated grammatically and not semantically’ (p.60). The logic employed is sometimes quite peculiar, for example in the discussion of the NP sanamu ya nyoka ‘drawing of a snake’ being referenced by a class 1 (animate) object marker further on, where Amidu claims that explaining the animate concord as a case of semantic agreement would be an example of ‘reductionist linguistics’, rejected since ‘its only goal is to portray the Bantu as incapable of distinguishing between animate and inanimate objects’ (pp.56-57).


Finally, chapter five once again problematizes the binary feature [±animate] and goes on to sketch a possible alternative, dubbed the polysemic hypothesis (p. 122), which holds, roughly, that class affixes are actually polysemous elements, at least as far as animacy is concerned.


The five chapters were originally intended to be published as separate essays. This accounts for considerable overlap in several places and lack of coherence in others. To give just one example, the polysemous analysis of class affixes (chapter 5) could have been linked to the observation, made in chapter 3, that noun classes are not at all conceptually homogeneous; this would have provided some support for the dynamic view of the noun class system espoused there. In the face of such issues, one wonders whether a bit more editorial care could not have given the work a more unified and focused feel. On the plus side, the book features a very complete and helpful index, in which key terms and principles are put in bold type.


By the end of the book, the reader is thoroughly disabused of the notion that simplistic semantic assignment rules could do the work, but is also left wanting more of a viable alternative. Amidu’s approach, based as it is on actual use cases, carries the seeds of a successful alternative, but does not quite manage to live up to the promise of actually articulating one. For that, the book is too narrow in scope, especially since it fails to situate itself in the context of several alternative (broadly cognitivist) approaches that have emerged over the years. The ‘well known prototype category approach’ (p. x) for example is only mentioned in the preface, where it is dismissed out of hand, even though it offers many advantages over the simple binary feature-based approach debunked over and over again by Amidu. Additionally, the reader will fruitlessly search for references to such illuminating analyses as Spitulnik (1987), Contini-Morava (1994), Palmer and Woodman (2000), and Selvik (2001), the omission of which is all the more surprising because they are very close in spirit to the usage-based, dynamic, radically polysemous view of Bantu noun classes advocated by Amidu.


In sum, Semantic Assignment Rules in Bantu Classes is a work that rightfully and forcefully argues against the danger of crude oversimplification in approaches to Bantu noun class semantics. It will be of interest to Kiswahili specialists and to Bantuists, especially those with a love for unruly linguistic facts.

References cited

Amidu, Assibi 1997

Classes in Kiswahili: a study of their forms and implications. Cologne: Rüdiger Köppe

Amidu, Assibi 2003

'Language teaching and semantic interpretation in Kiswahili classes.' The Linguistic Association of Korea Journal 11,3:103-121

Contini-Morava, Ellen 1994

'Noun Classification in Swahili', (20.12.2007)

Corbett, Greville G. 1991

Gender. New York: Cambridge University Press

Lyons, John 1968

Introduction to Theoretical Linguistics. Cambridge University Press

Palmer, Gary B. and Claudia Woodman 2000

'Ontological classifiers as polycentric categories, as seen in Shona class 3 nouns.' In: Explorations in Linguistic Relativity, ed. by Martin Pütz and Marjoliijn H. Verspoor, pp.225-249. Amsterdam: John Benjamins

Selvik, Kari-Anne 2001

'When a dance resembles a tree.' In: Polysemy in Cognitive Linguistics: Selected Papers from the International Cognitive Linguistics Conference, Amsterdam, 1997, ed. by Hubert Cuykens and Britta Zawada, pp.161-184. Amsterdam: John Benjamins

Spitulnik, Debra A. 1987

Semantic Superstructuring and Infrastructuring: Nominal Class Struggle in ChiBemba. Bloomington, Indiana: Indiana University Linguistics Club




Es liegen noch keine Kommentare vor.

Möchten Sie Stellung zu diesem Artikel nehmen oder haben Sie Ergänzungen?

Kommentar einreichen.

  1. Deutsch
  2. English
  3. Français