Serial verb constructions. Research notes
Discussion notes on the ethno-cognitive view of SVC
Some linguists are inclined to see behind this way of expressing processes and relationships different ways of experiencing the world, events in the world, and relationships between entities involved in these events. "… the Ewe people describe every detail of an action or happening from beginning to end, and each detail has to be expressed by a special verb: they dissect every happening and present it in several parts …" Westermann, Diedrich (1965). A Study of the Ewe Language, p. 265. Westermann's ethno-cognitive construction of what happens in Ewe SVC is strikingly accurate from a European standpoint and applies just as well to Akan SVC.
Nevertheless, a word of caution is in order here. (For a radical critique of the ethno-semanticist view, see Delplanque 1998.) Languages may be said to use various strategies for modelling the world as it is being experienced. Languages differ in many ways on account of this modelisation of experience. From this, however, it does not follow that the psychological, perceptual and cognitive mecanisms involved in processing this experience are necessarily different. Similar or identical experiences are open to widely divergent modelisation by verbal means. As can be seen from the possibility of paraphrasing a sentence in various ways without altering its meaning, this is even the case in one and the same language (although one has to be careful not to gloss over, in the name of truth-conditional semantic invariance, subtle semantic differences expressed e.g. by word order, passivisation, etc).
It remains that some aspects of the Akan language, perhaps the most striking among them being verb serialisation, represent a different type of modelisation of human experience. By saying so we do not exclude nor do we precipitately conclude that different strategies used in organising and processing the experience could not or could in fact correlate with some deeper and potentially pervasive difference in the way events are cognitively experienced in Akan. We do not deny either that differences of modelisation through lexico-grammatical structures tend to be highly systematic and should therefore be of greatest relevance to typological inquiry.
- the critique by Delplanque
Discussion notes on the grammaticalisation view of SVC
It has been claimed that verbs which are typically used for expressing some particular role of a participant in relation to some action, such as the role of beneficiary or of instrument, are no longer to be counted as 'real' verbs but are in a more of less advanced state of recategorisation as prepositions or case markers, due to a process of reanalysis. The theoretical underpinning of this view is the framework of grammaticalisation.
Grammaticalisation theory helps to understand - and to learn - a number of characteristics typical of serial verb constructions. Yet at the same time it raises a number of questions and objections which cannot be brushed under the carpet. It will be sufficient to briefly mention some of these questions, using the same example of the verb maŠ for illustrative purposes:
For a balanced approach to serial verbs in Akan which retains fundamentally a grammaticalisation perspective but at the same time rectifies its exaggerated claims, see Osam (1994a). See also the excellent discussion in Osam (1994b, Ch. 6).
For a more radically critical view,
Verb serialisation denotes an ingenious lexico-grammatical strategy of conflating two aspects of the function of language in one surface operation, namely those of identifying the referential characteristics of processes and states on the one hand and the role of their arguments on the other hand. To the extent that the same category - verbs - is used for indicating role relationships, verbs do indeed show a functional affinity with case markers. Where grammaticalisation theories go wrong is when they derive from this affinity a claim of a shift towards a case-marking-system which, moreover, is suspiciously similar to more familiar Latin type case systems. This view disregards the typologically different inherent properties of the SVC language typology.]