Functional Interpretations: Borderline Idiosyncrasy in Prepositional Phrases and Other Expressions

There is a type of prepositional phrase which not only specifies a spatial relationship between the located object (LO) and reference object (RO), but also forces entailments drawing on knowledge about such things as the canonical function of an entity or the stereotypical interactions which it has with other entities. Often, this restricts the range of possible meanings of the PP and hence the contexts in which it may appear. For instance, (1) means that Gertrude was seated in frontal proximity to the piano, but forces the additional entailment that she is looking at it, playing it or otherwise interacting with it. Replacing before with in front of renders this interaction optional. before is impossible in contexts like (2) which exclude such an interaction. (3) and (5) give examples of 'functional' uses of prepositions. on in (3) entails that Cuthbert is in a position to use the train in its canonical function as a means of transport. Since the train in (4) cannot thus function, on must be replaced by in. In (5), the preposition with a bare noun complement forces the inference that Nelly is on the school premises in order to participate in educational activities. If we provide school with an article, this implication does not necessarily hold, and the bracketed phrase becomes acceptable. (1) Gertrude was sitting before the piano. (2) Everyone in the room was fixated on the television. Fran and Stan were on the sofa, Gordon was sitting near the fire and Gertrude was sitting {in front of/*before} the piano, her back resting on the keyboard. (3) Cuthbert is on the train. (4) Cuthbert is {in/*on} the old train on display in the museum. (5) Nelly is at/in school (*to pick up her little sister). Apart from PPs, we also study a number of other constructions displaying interpretations arguably related to the ones just seen. One such construction is known as the 'verb-particle combination', 'phrasal verb' or 'particle verb', where information about the functional or associative relationships between objects is often enlisted in deducing the type of entity which is the implicit RO of the prepositional relation expressed by the particle. For instance, in the VP put a record on, the implicit RO of on must be the type of object on which the record can fulfill its function and on which records are canonically found, namely a record player. Likewise, many denominal verbs are specialised to situations involving typical uses of a noun. For instance, our knowledge of the things denoted by cage and gaol tells us that they are designed for accommodating animals and prisoners. When these nouns are converted to verbs, their object selection properties are sensitive to this knowledge. One can cage lions and gaol criminals, but one cannot *cage the lions' food or *gaol a new prison warder. Also, the recoverability of an ellipsed verbal argument is often restricted to cases where its an entity is in a functional or stereotypical relationship to a surviving argument. Thus, pour a glass (unlike pour something into a glass) is usable only if the implicit substance is something for which a glass is designed, viz. a drink (cf. logically possible, but unacceptable cases like *I got dishwashing liquid and pouted the glass. Various linguists in personal communication suggest that functional interpretations 'come from the context'. I later note that functional interpretations of some expressions indeed result from context-based inference, there is ample proof that this is not always the case and thus that the existence and distribution of functional readings is not a trivial problem. It is circular to claim that the functional construal of the RO in I put a record on comes from the context, since put on only allows contexts where this construal is possible, i.e. one where the LO is put in a position where it can be used in its normal function (e.g. put clothing on, put makeup on, put the kettle on). Likewise, the fact that I put the plate before Ann suggests that Ann is meant to interact with the plate might be thought to emerge as a trivial implication of the context being described, at least if I conceed that putting something in someone's frontal proximal region is usually meant to instigate an interaction. However, the relevant use of before (studied in 6.1) is specialised to contexts where the LO and RO interact in some fashion, while its near synonym in front of is not, cf. I put Dave's plate {in front of/*before} Ann while I wiped the table in front of Dave. The interactional entailment does not come from the context in which before appears; rather the preposition insists that it be embedded in a context where interaction is possible. One perhaps genuine instance of a contextually inferred functional reading is the intuition that the PP in Ann is at her desk (without further context) implies that she is using it as a work surface. This can be seen as a rather obvious type of contextual guessing based on our knowledge of the kinds of things which one normally does in the at-region of a desk. However, this reasoning appears to be equally valid in a context where at is replaced by near, but such a context excludes the functional reading. We must find out why. Furthermore, the functional interpretation of at the desk is a 'weak' functional reading, being easily cancelled in a suitable context: During the office party, Dave stood as far away as possible from Ann, who was sitting at her desk, drinking vodka and talking to Fritz. This raises the question of why the functional reading of at the desk can be nullified by the context when that of before a person cannot. This essay tries to come to grips with all these issues.