Extending beyond space

Extending Beyond Space Brooke O. Breaux (brookebreaux@louisiana.edu) Institute of Cognitive Science, University of Louisiana at Lafayette, Lafayette, LA 70504 USA Michele I. Feist (feist@louisiana.edu) Institute of Cognitive Science, University of Louisiana at Lafayette, Lafayette, LA 70504 USA Abstract Investigations into the semantics of the spatial and non-spatial uses of in and on have tended to assume that a type-level similarity exists between these two prepositions. However, their syntactic distributions, while overlapping, are not equal in scope (Navarro, 1998). In this paper, we ask whether these distributional differences might be related to semantic differences between the two terms. The preliminary evidence collected here suggests that in and on have slightly different levels of interpretability, even in their prepositional uses. Thus, both semantically and syntactically, the assumption of type-level similarity may need to be qualified. Keywords: Semantics; prepositions; metaphor; language Introduction Investigations into the semantics of prepositions such as English in and on have tended to treat these lexical items as though they are different tokens of the same semantic and syntactic type. Such treatment seems to follow from the generative grammar tradition in which lexical category – rather than meaning – determines syntactic behavior. For example, Cook and Newson (2007) suggest ―that arguments are interpreted in a particular way due to the structural positions they occupy‖ (p. 263). This assumption is also reflected in introductory linguistics and psycholinguistics text books, which state that words belonging to the same lexical category, or word class, are typically interchangeable syntactically (cf., Carroll, 2004; O’Grady, Archibald, Aronoff, & Rees-Miller, 2005). Together these suggest that different lexical items drawn from the same word class may interact with the rest of language in very similar ways. Even in more cognitive views of language, we find evidence that prepositions are treated as a lexical category without indication that the individual differences between the prepositions will have important repercussions for the functions of the individual lexical items within the linguistic system. As a case in point, type-level equivalence has been assumed in examinations of the semantics of prepositions (e.g., Coventry & Garrod, 2004; Feist, 2000, 2008, in press; Feist & Gentner, 2003; Tyler & Evans, 2003; Vandeloise, in press). Much of this work focuses on the criteria that distinguish the meaning of one preposition from that of another, without discussion of the possibility that prepositions may differ in additional ways beyond their meanings. For example, while Tyler and Evans (2003) do acknowledge the importance of context in establishing the meaning of a lexical item and the fact that different prepositions will occur in different contexts, such contextual factors do not lead to different proposals regarding the nature of the meanings of individual prepositions. However, evidence from corpus-based studies of prepositions challenges this assumption of distributional equivalence. For example, in his investigation into the semantic structure of English topological prepositions, Navarro (1998) found a differentiation between in and on based not only on their meanings but also on their syntactic distributions. While on tends to occur primarily in prepositional constructions, in is also quite prevalent within ―a wide range of morphosyntactic usages that make it controversial to categorise it on behalf of a single syntactic construction‖ (Navarro, 1998, p. 273), including use as a full adverb, as an adverbial particle of a phrasal verb, and as a prefix for nouns, adjectives, and adverbs. This difference in syntactic distribution suggests that, despite their similarity as topological prepositions, in and on may behave quite differently within the language system as a whole. Following up on these observations, we searched for uses of in and on in the more than 400 million word Corpus of Contemporary American English (COCA; www.americancorpus.org). Our first observation was that the frequencies of occurrence of in and on are highly unequal overall, with in (7,333,378 instances) appearing more than 2½ times more frequently than on (2,723,768 instances). Secondly, and more importantly, we examined the combinatorial possibilities for both in and on across a set of naturally occurring uses within a limited syntactic context (i.e., prepositional phrases containing the preposition immediately followed by a noun). Within the hundred most frequent collocations for each preposition, we observed an inequality in the distribution of uses, χ 2 (1, N = 200) = 21.34, p = .0003 (see Table 1). Table 1: Noun types collocating with in and on In On Proper Nouns Noun Phrases Idioms Concrete Nouns Abstract Nouns