ABSTRACT This research develops an approach to the structure of complex words in natural languages which treats relations among words as defined by a network of interacting rules instead of by their shared content of minimal signs (or 'morphemes'). It is thus 'a-morphous' in dispensing with morphemes rather than in the sense of lacking form. Three general problem areas will be addressed. First, if rules rather than morphemes are the foundation of word structure, words should have much less internal non- phonological form than commonly assumed. The investigator will examine the idea that most such structure, such as that usually associated with the presence of boundaries or junctures, and with a phrase-marker-like arrangement of morphemes into higher level constituents, can be eliminated. Secondly, the research will pursue the notion that a single class of phonological operations on word forms characterizes phonology, word level morphology, and also the introduction of phonologically dependent clitics. Treating clitics as "phrase level morphology" has significant implications for several problems in syntax as well as for phonology and morphology. Finally, the investigator will explore the implementation of grammars of word structure of this type in the form of computer programs. Such implementation involves several difficult issues, such as the proper representation of the full autosegmental and metrical structure of a word; the balance between general and language particular principles in the interaction of sound structural rules; and the degree and nature of parallelism appropriate to the parsing of morphological structure. %%% The linguistic theory of grammatical structure at the level of the word -- "morphology" -- has long presented special problems. In some ways structure at the level of the word seems intuitively more obvious than other levels of grammatical struc- ture, such as the sentence, and yet it has proven much more resistant to a comprehensive, coherent theoretical account. One reason for this is the great diversity from one language to another in the nature of what the speakers perceive as individual words. The investigator in this project has been developing a quite novel theory of morphology, which accounts for a number of phenomena which have been unexplained in other theoretical models. The next phase of the research will test the theory against certain problematic cases from a variety of different languages, and will also implement the theory as a computer program, to enable further testing of its consistency and applicability to a wide variety of cases.
