Morphological Productivity Structural Constraints in English Derivation Conclusion

We started out with the statement that the mechanisms responsible for the productivity of morphological processes are still ill-understood. After having clarified the notion of productivity and how it can possibly be measured, a broad range of suffixes were investigated in order find out more about the machinery necessary to deal with their combinatorial properties. The gist of the argument was that most of the responsible factors lie with the individual properties of a given process and that more general mechanisms can be severely restricted.

We then narrowed our focus down to one of the most intriguing problems in derivational morphology, the nature of rival processes, with basically the same conclusions emerging. The close inspection of the structural properties of the individual verb-deriving processes revealed that the productivity and distribution of these processes fall out once these properties are correctly described.

On the theoretical level, the foregoing study has argued for a sign-based, output-oriented model of derivational morphology, and against theories that want to separate meaning and form in morphology. Coming back to the general, i.e. not rule-specific, restrictions proposed in the literature, it seems that only the last three hold up against the empirical evidence: the unitary output hypothesis, blocking, and stratal constraints. The unitary output hypothesis is inherent in an output-oriented approach and need not be commented on in more detail. Token-blocking (but not type-blocking) is certainly a relevant factor restricting productivity, but it was argued that it is much less a structural than a processing constraint. Stratal constraints are also relevant in some areas of derivation but contra earlier claims by Gussmann (1987), for example, they seem to be of minor influence when it comes to verbal derivation. Nevertheless, further investigation of the stratal constraints is needed to clarify their nature and relevance.

This topic has also made a contribution to methodological issues pertaining to the study of word structure. Thus, the most important insights into the nature of the verbal derivation were gained on the basis of the analysis of large numbers of neologisms. By restricting the data in this way it was possible to arrive at significant generalizations concerning the properties of possible words that go beyond what can be found in earlier treatments. The OED on CD turned out to be a powerful tool for this kind of approach and allowed the coverage of a broad range of data. Furthermore, it was demonstrated that the OED can be used even for measuring the productivity of a given process, provided that one is aware of its shortcomings. The application of Baayen's text-corpus-based measures to derived verbs as manifested in the Cobuild corpus indicated that these statistical measures are useful for the determination of the productivity of overt affixation, if employed carefully, and if complemented by a qualitative investigation.

Returning to the central problem of productivity, this study, like many others before, has argued that the productivity of a given morphological process can largely be predicted on the basis of the process's peculiar structural properties and restrictions. However, I have also demonstrated that even in a seemingly well-described language like English surprisingly little is known when it comes to the exact formulation of these properties. The foregoing study can therefore be read as a case study laying out a possible research agenda for future in-depth investigations of other derivational processes in English and other languages.