Abstract
The principle of end-weight has frequently been applied to the word order patterns of Old English (e.g. Colman, 1991; Kohonen, 1978; Reszkiewicz, 1966, 1971; Strang, 1970). Although there are strong indications that heavier constituents are preferred later in the Old English phrase or clause than lighter constituents, the motivations for this pattern have not been sufficiently explained. This thesis examines patterns of constituent order and adjacency in Old English noun and verb phrases within the framework of Performance Theory (Hawkins, 1994, 2004). The distribution of syntactic weight is explained as a consequence of processing considerations. Heavy constituents are preferred in postposition in a language such as Old English, in which phrasal nodes are typically constructed on their left periphery. This ordering pattern is predicted to improve processing efficiency through domain minimization.
To test the predictions of Performance Theory (and the principle Minimize Domains in particular), I examine the corpus frequencies of variant ordering patterns for adjectives, genitive nominals and relative clauses in the noun phrase, and for adverbs, subject predicates and adposition phrases in the verb phrase. I then compare the predictions of Performance Theory to the patterns found in the selected Old English prose corpus and to the predictions made in previous studies of Old English word order.
In general, the corpus data support a processing-based analysis of constituent order and adjacency in Old English prose. However, Performance Theory alone does not satisfactorily explain all observed patterns. On the basis of my findings, I propose that an information-processing approach also be used, in conjunction with Hawkins' syntactic-processing approach, to explain differences in the functions of constituent position. The way in which constituents are presented in the larger context of the discourse appears to be a strong motivation for variation in constituent position.
Performance Theory makes predictions for both synchronic and diachronic patterns of order and adjacency. I suggest in response that some of the ordering patterns observed in the Old English corpus are best explained with reference to the processes of language change. Moreover, Performance Theory makes correct predictions for English more frequently than do previous accounts of word order change (e.g. Vennemann, 1974). I also outline the advantages and disadvantages associated with a diachronic approach to syntactic processing and the Old English data.