Vortrag: 18-20 Uhr, Raum 051 RG, Schellingstr. 3
This paper traces the roots of the phenomena associated with constituent structure to the two domain-general processes of chunking and categorization. Chunking is the process by which repeated sequences come to be stored and accessed as a single unit. Categorization in this context refers to the similarity mapping that occurs when a linguistic unit in comprehension activates a stored representation to which it is identical.
Both of these processes create a gradient: units in sequence may be more or less chunked together depending primarily on the extent to which they have been repeated together; activation of a stored unit can be greater or stronger. This leads to the hypothesis that constituent structure is gradient and that it can differ according to the particular items serving as potential constituents.
The gradience of constituent structure allows us to solve the problem of gradual change that results in reanalysis – change in category or constituent structure. For instance, a controversial case concerns complex prepositions such as in front of, in back of, because of and in spite of. These phrases resemble in some ways simple prepositions while maintaining in some cases properties of their sources in two prepositional phrases. A combination of increased frequency of use and gradual semantic change leads to gradual change in constituent structure evidenced by the loss of nominal properties of the central noun and the reassignment of of to the items preceding it rather than to the item following it.