During the past two years work has been started on the preparation of a new Swahili dictionary, and while the work is still at an early stage of development it is clear that two particularly difficult problems concern the extent to which lexical material can be systematized, and the manner in which this can best be achieved. We take as one of our aims, that expressed recently by the late Uriel Weinreich, of “…improving lexicographic description by more careful attention to the known syntactic properties of the entries.”1 Two areas in which the existing dictionaries are particularly weak are those of transitivity and verbal extensions. There is, for example, no indication of whether a verb may ‘take’ an object-prefix or not;2 nor, despite the inclusion of many extended forms, is it clear from the absence of a particular form whether this means that it cannot occur, or simply that there is nothing remarkable about its usage and hence it does not need special mention.3 Again, it is not clear on what basis the listed forms have been selected, whether the aim has been exhaustiveness or selectivity. It is, of course, particularly easy to criticize the lexicographer here since the field of verbal extensions is a vast and complex one which has received very little attention from Swahili scholars until the last year or two.4 But the preparation of a major new dictionary seems an appropriate time to try and reach some generalizations about patterns of transitivity, both for minimal and extended

radicals, and this will involve a re-examination of the whole basis for the classification of verbal extensions.