![]() Uta Reinöhl |
![]() T. Mark Ellison |
A common feature of the grammaticalization of function words is that they develop the requirement for obligatory hosts. For instance, English the does not occur except when followed by a nominal construction. This talk offers an account of the historical development of obligatorification - how hosts develop from optional extras to required accompaniments.
In the first part of the talk, we show that the process leading to obligatorification is semantically and/or pragmatically driven. We further show that it begins before grammaticalization “proper”, rather than being a result of grammaticalization as has been widely, but implicitly, assumed. In fact, specific semantics and/or pragmatics create the need for overt hosts at every synchronic stage of every language. Only rarely does this overt-host requirement develop into a syntactic requirement as a result of grammaticalization. We illustrate this with both diachronic and synchronic examples from diverse parts of speech, from several languages.
This account is supplemented by a computational model of incremental sentence processing. We show that in some circumstances, the desired interpretation of a construction can only be achieved if the host is present. This model shows that obligatorification is not a feature of grammaticalization but rooted in ever-present creative language use.
A common feature of the grammaticalization of function words is that they develop the requirement for obligatory hosts. For instance, English the is ungrammatical unless it is immediately followed by a nominal construction. This talk offers an account of the historical development of obligatorification - how hosts develop from optional extras to required accompaniments. In the literature, this topic is little discussed and is, if anything, directly associated with the transition from content to function word, e.g. from Old English se optionally combining with a nominal construction to the obligatorily combining with one. We argue in this talk that this view is theoretically wanting and empirically inaccurate. Theoretically, this scenario misses intermediate steps that explain the change from optional to obligatory. Empirically, we show that the process leading to obligatorification is i) semantically and pragmatically driven and ii) begins before grammaticalization proper. Reference terms that are semantically and/or pragmatically incongruent with the basic lexical meaning of the future function word must be expressed to enable listener comprehension. An increase in such examples over time paves the way for obligatorification and they also present the bridging cases for the transition from content to function word. This account is underscored by evidence from the grammaticalization of postpositions out of relational nouns in Indo-Aryan (Reinöhl 2016), which we replicate in this talk for a cognate construction in English. In addition, we show that the same process also accounts for adpositions grammaticalizing out of adverbs, as well as for articles grammaticalizing out of demonstratives as in
This account is supplemented by a computational model of incremental sentence processing. Semantic interpretations are represented as acyclic graphs in which links correspond to semantic relations and nodes to entities of specific categories, and where both are annotated for their level of activation (moment-to-moment salience). The interpretations of larger units (e.g. sentences) are constructed incrementally by the unification of the interpretations of smaller units (e.g. words). We show that in some circumstances, the desired interpretation of a phrase can only be achieved if the host is present. An overt host is thus the only means of forcing the correct interpretation for some phrases. This, we argue, is the starting point from which syntactic obligatorification develops.
The data and code in this section will mostly pertain to corpora which we use to investigate the presence (or not) of obligatorification in incongruent constructions.