Lee Morgan was selected to participate in the Ninth Annual Harvard Undergraduate Linguistics Colloquium.
Morgan presented work from his Div III thesis, entitled “The Nature and Process of Metaphorical Interpretation.” In it, he synthesizes research in philosophy, linguistics, and literary theory in order to present an original analysis of metaphorical expressions.
While it might seem that accepting the invitation to participate in such a prestigious undergraduate colloquium was an obvious choice, Morgan did have to make a decision. He submitted an abstract of his Div III thesis both to Harvard’s colloquium and to the Southern California Undergraduate Linguistics Conference at UCLA, and was accepted into both. As they were scheduled on the same weekend, he chose the Harvard event.
The colloquium is hosted each spring by the Harvard College Linguistics Group, a student-run organization, and brings together undergraduates from around the globe.
Each Division III student at Hampshire works closely with a faculty committee. Members of Morgan’s committee are Daniel Altshuler (chair), who is an Andrew W. Mellon Postdoctoral Fellow in Language, Mind and Culture at Hampshire this year, and linguistics professor Steven Weisler.
In his presentation and in his Div III, Morgan argues that the inferences a hearer draws from a metaphor are not meant by the speaker and therefore cannot be analyzed by the usual linguistic tools of semantics and pragmatics. Rather, interpretation is based in everyday processes of logical reasoning, relying on an understanding of conventional facts within the hearer’s community, even if these “facts” are not true or believed.
A metaphor's literal meaning can differ greatly from the inferences that hearers draw from it. For example, from the metaphor in (1), which literally means that John is an elephant, a hearer might draw the inference in (2).
(1) John is an elephant.
(2) John is large.
Searle (1979) and Grice (1989) treat the disparity between the literal meaning of a metaphor and the inferences drawn from that metaphor as a distinction between sentence meaning and speaker meaning.
Davidson (1978), however, argues that in metaphors, inferences like (2) are neither literally nor speaker meant, a view that has been recently advanced by Lepore and Stone (2010). According to this view, the metaphor in (1) merely causes a hearer to reach (2). As Davidson writes, metaphor can, "like a picture or a bump on the head, make us appreciate some fact—but not by standing for, or expressing, the fact." He argues that metaphorical inferences are not produced through linguistic mechanisms, which would entail that the speaker means (2). He instead proposes that metaphorical interpretation is based on imagination; a hearer infers (2) from (1), for example, by having an experience of imagining John as an elephant.
In the first part of the talk, I present a new argument for the position that metaphorical inferences are not speaker-meant, based on Grice's (1957) definition of speaker meaning and a comparison of metaphors to hints. In particular, I argue that hints intended to produce a belief are not speaker-meant, while hints intended to produce an action are speaker-meant. Subsequently, I provide novel data illustrating that metaphors are akin to belief-oriented hints and not action-oriented hints, and thus are not speaker-meant.
In the second part of the talk, I argue against Davidson's position that metaphorical interpretation is based on imagination and adopt Hobbs’s (1990) idea that axioms of world knowledge play the crucial role. These axioms are granted by the hearer for the sake of interpretation, but need not be true or believed. I argue that there are three classes of axioms that are linguistically relevant for metaphorical interpretation: necessarily true axioms, contingently true axioms, and contingently false axioms. In (1), the axiom necessary to produce the inference in (2) is that elephants are (generally) large, which is contingently true. In (3) and (4), the relevant axiom, i.e., that giants are large, is necessarily true.
Finally, in (5), the axiom necessary to produce the inference in (6) is contingently false; lemmings are not, in fact, particularly suicidal.
(3) Sean is a giant.
(4) Sean is large.
(5) Alan is a lemming.
(6) Alan is suicidal.
I examine what these three classes of axioms have in common, and conclude that in each case the axiom is said to be true by a particular interpretive community.
In sum, I argue that metaphorical inferences are not speaker-meant, and then analyze the process of metaphorical interpretation, focusing on the nature of inference selection. The proposed theory has broad implications for a theory of metaphor, hinting, speaker meaning and speech act theory.