another (thego/changesentences) and (b) something making something else have a particular characteristic over a
period of time (thekeepsentences). Because similarly structured complex concepts appear in many (and possibly all)
semanticfields, it is convenient for the language to use the same words as it switches from onefieldto another.^184
This underlying abstract syste mis most evident in se manticfieldsthat vary along a one-dimensional range of values:
they invariably are expressed in terms of the linear oppositesupanddownorhighandlow. Numbers (and hence prices,
weights, and temperatures) go up and down, military ranks go up and down, pitches on the musical scale go up and
down, and so does one's mood. (Time concepts, however, are a partial exception to this generalization. In just about
every language, they are expressed by terms that also apply to space; but instead ofupanddown, they use a front-to-
back continuum, for examplebeforeandafterin English.)
Similar grammatical and lexical patterns appear in language after language. Cognitive linguists such as Talmy (1978;
1985) and Langacker (1987) essentiallytake the mfor granted and build theories of conceptual structure around the m.
Lakoff and Johnson (1980; Lakoff 1987; 1990) argue that they are part of a vast syste mof metaphor, inextricably
embedded in the conceptual structure expressed by language. They further argue that the pervasiveness of metaphor
makesitimpossibletobasea semantictheoryona simplisticnotionof“literaltruth,”and impossibletotreatmetaphor
as non-truth-conditional and therefore marginal.
While acknowledging the ubiquity of metaphor in thought, I would argue for afiner-grained account for these
particular phenomena (Jackendoff 1983; 1992a: ch. 3; Jackendoff and Aaron 1991). Lakoff and Johnson use the term
“metaphor”foranyextension of terms from one semanticfield to another. By contrast, the traditional notion of
metaphor is reserved for creative, novel expressions, often with a patent semantic clash, used to make speech more
colorful. The parallelsillustrated in (13)–(16) exhibit no semantic clash, and theyare the only ways availablein English
of expressing theconcepts in question. Thus theyare notmetaphorical in thetraditional sense. I wouldcontend rather
that they reflect a set of precise abstract underlyingconceptual patterns that can be applied to many different semantic
fields. These patterns are the basic machinery thatpermits complex thought to be formulated and basic entailmentsto
be derived in any domain. Among these domains, the spatial domain exhibits a certain degree of primacy due to its
evolutionary priority
358 SEMANTIC AND CONCEPTUAL FOUNDATIONS
(^184) “Convenientfor the language”is an abbreviation for a variety of factors. Among them: speakers who lack a word for something they want to say will press a conceptually
related word into service; learnersfigure out less concrete meanings more easily if more concrete (especially spatial) parallels are made evident by phonological identity.