Fabb (1988)

Fabb investigates the combinability of 43 English suffixes, restricting himself to suffixes that attach to otherwise free forms (i.e. excluding bound roots). In view of the attested and non-attested combinations he comes to the conclusion that level ordering restrictions and syntactic category re strictions limit the number of a possible 1849 combinations to 459. Fabb, however, finds only 50 combinations attested, which is a strong indication that there must be restrictions operative other than those proposed by level orderists. Although Fabb's arguments against level ordering are substantial, the alternative proposed by Fabb is much less attractive. He posits a model of English derivation that contains four kinds of suffixes, grouped according to the restrictions they show.

Although in principle individual selectional restrictions may constitute a serious alternative to a level-ordered morphology, it seems that Fabb may have thrown out the baby with the bathwater because some interesting generalizations observed by level-orderists are lost under Fabb's approach.

Fabb's model also has consequences for another much debated issue in generative morphology, namely the above-mentioned visibility of word-internal structure to affixation processes. Fabb's model assumes that suffixes are sensitive to whether the base is already suffixed, which contradicts one of the central claims of Anderson's A-morphous Morphology (1992), or Kiparsky's Bracket Erasure Convention (1982a).

The remainder will deal mainly with two questions. The first one is whether Fabb's observations are empirically correct, and the second one is to what extent the stacking restrictions of affixes may be explained by independent mechanisms of English morphology instead of idiosyncratic affix-driven selectional restrictions. Especially the second question is extremely difficult to answer and it should be stressed that I will not attempt to provide an all-embracing account. What I rather aim at is to outline which kinds of mechanisms we need to acknowledge and how they may possibly interact. For many of the individual problems to be discussed, only tentative solutions will be offered, which will have to be further refined or backed up by future investigations. For instance, the role of semantic compatibility of suffixes certainly deserves further attention since it seems that in this domain a number of interesting restrictions can be located.

It should also be noted that the present treatment is not exclusively meant as a reaction to Fabb's article. My aim is not only to refute some of Fabb's claims, but to come a few steps closer to the solution of some important problems of English morphology, which many linguists have dealt with before. Fabb's study only serves as a reference point because it is justly praised for showing the inadequacy of preceding approaches. Furthermore, Fabb's predictions can be easily tested, which is a virtue that more traditional treatments of English morphology tend to lack (e.g. Marchand 1969, or Jespersen 1942).