The traditional categories
Early studies of semantic change like those in Bréal (1897) did little more than set up broad categories of change, described with very general, and often vague, labels like ‘weakening’, ‘strengthening’ and so on. In the first instance, we will consider four of these traditional categories of semantic change: specialization, generalization, ameliorization and pejorization. These categories were not part of an effort to explain why meaning change happens, but were meant to contribute to a typology of semantic changes – a precondition of any further explanatory progress.
One common type of change is specialization (narrowing), in which a word narrows its range of reference:
• English liquor used to refer to liquid of any kind: the reference to alcohol was a subsequent specialization.
• English pavement originally referred to any paved surface, but specialized to simply cover the footpath on the edge of a street (called side walk in American English).
• The proto-Romance word for ointment, unctu, specialized in Romanian so as only to refer to a single type of ‘ointment’, butter (as well as undergoing some phonological changes to become unt; Posner 1996: 319).
The opposite tendency is generalization (broadening), in which a word’s meaning changes to encompass a wider class of referents.
• zealot first referred to members of a Jewish resistance movement against the occupying Romans in the first century AD; its contemporary meaning ‘fanatical enthusiast’ is a later generalization.
• French panier ‘basket’ originally meant just a bread-basket; it was subsequently generalized to baskets of any kind.
• The Latin noun passer means ‘sparrow’, but in a number of Romance languages it has generalized to the meaning ‘bird’: this is the case, for example, with Spanish pájaro and Romanian pasa‡re.
• The most common verb for ‘work’ in Romance languages, like French travailler and Spanish trabajar, is a result of a generalization from the Latin *tripaliare ‘torture with a tripalium’, a three-spiked torture instrument (Posner 1996: 322).
• The German adverb sehr ‘very’ originally meant ‘cruelly’ or ‘painfully’ (Kluge 1989; a trace of this meaning survives in the verb versehren ‘injure, hurt’). The shift to ‘very’ is an example of an extreme generalization that has lost almost all connection with the original sense. A similar change is found in many English intensifier terms, like terribly and awfully.
QUESTION Classify each of the following semantic changes as either generalization or specialization. Are there any cases where it is hard to decide?
• Latin curtus ‘short (in space)’ > ‘short (in space and time)’: French court, Spanish corto, Portuguese curto, Italian corto.
• Midde Japanese ake-sita, ‘dawning time, dawn’ > asita ‘tomorrow’ (Traugott and Dasher 2002: 56).
• Dutch drukken ‘to press, to push hard’ > ‘to print (books)’.
• French arriver ‘arrive at the shore’ > ‘arrive (anywhere)’.
Two other traditional categories in the analysis of meaning change are pejorization (Latin pejor ‘worse’) and ameliorization (Latin melior ‘better’). These refer to change in words’ evaluative force. In pejorization, a word takes on a derogatory meaning. This is frequently seen with words for animals, which can be used to refer to people negatively or insultingly, as when someone is called a monkey, parasite, pig, sow, and so on. Another example of pejorization is the adjective silly. This originally meant ‘blessed, happy, fortunate’; its contemporary meaning ‘foolish’ is a later development – and one which, this time, has entirely displaced the original sense. Similarly, boor’s original meaning was ‘farmer’; ‘crude person’ was a later pejorization. Accident originally meant simply ‘chance event’, but took on the meaning ‘unfavourable chance event’.
Ameliorization is the opposite process, in which a word’s meaning changes to become more positively valued. The normalization of previously proscribed taboo words is a good example. Bum, for example, appears to be gaining somewhat in social acceptability, at least in Australian English. It has thus started on the path to what could be full ameliorization: this would be attained if it eventually became fully synonymous with bottom. Another example of ameliorization is provided by English nice. The earliest meaning of this adjective, found in Middle English, is ‘simple, foolish, silly, ignorant’; the basic modern sense, ‘agreeable, pleasant, satisfactory, attractive’ is not attested until the eighteenth century.
These four categories on their own are not at all adequate to describe the complexities and diversity of the types of meaning change encountered in the history of languages. How do we explain, for instance, the shift of Latin ver ‘spring’ to the meaning ‘summer’ in many Romance lan guages (Romanian vara, Spanish verano, Portuguese verão; Bourciez 1967: 207)? This seems to fi t none of the four categories we have mentioned. The same could be said for the development from Latin sensus ‘sensation, consciousness, sense’ to the meaning ‘brains’ in Spanish seso. Similarly, how to account for the shift from ‘count’ to ‘read’ in Tolai luk (Austronesian, Papua New Guinea; Tryon 1995 IV: 509)?
A solution to this sort of problem comes from recognizing specialization and generalization as just two types of metonymic change. Metonymy (see 7.2.4) is the process of sense development in which a word shifts to a contiguous meaning. ‘Contiguous’ has a number of meanings (Peirsman and Geeraerts 2006), but the essential idea is that two senses are contiguous if their referents are actually next to each other (either spatially or temporally), or if the senses underlying the words are closely related conceptually. The Spanish ‘consciousness’ > ‘brains’ shift exemplifies conceptual contiguity (brains and consciousness have a close conceptual association), while the ‘spring’ > ‘summer’ shift exemplifies both temporal and conceptual contiguity: summer is next to spring in time, and also a closely related notion conceptually. Some instances of pejorization and ameliorization can also be considered metonymic: the shift of boor from ‘farmer’ to ‘crude person’ could be considered to be based on the close association of these two notions.
Metonymy is a powerful category: it can describe many types of change which can’t otherwise be accommodated. The English noun bead originally meant ‘prayer’ in Old English. Its present meaning can be explained through the widespread use of the ‘Rosary’, a chain of beads which Christians used to keep track of prayer sequences. This association established a conceptual link between the notions ‘prayer’ and ‘bead’, explaining the metonymic transfer of bead to the latter meaning in the Middle English period. This is neither generalization/specialization, nor ameliorization/pejorization, so a new category is clearly needed to describe it.
Metonymic changes are common. A particularly colourful one under lies the word pupil, which in English refers both to a student and to the opening in the eye through which light passes. This puzzling polysemy goes back to Latin, where pupilla means both ‘small girl, doll’ and ‘pupil’. This can be explained by metonymy. Our eyes have ‘pupils’ because of the small doll-like image that can be observed there: spatial contiguity, in other words, underlies the shift. Greek khōrē has exactly the same metonymically related meanings. Another example of a metonymic meaning shift is the Romanian word ba‡rbat ‘husband’, which derives from the Latin barbatus ‘bearded’. If husbands often have beards, the ideas will be conceptually associated.
Metonymy was a notion adopted into linguistics from rhetoric, the traditional study of figurative, literary and persuasive language. Another originally rhetorical concept with linguistic application is metaphor, discussed from the synchronic point of view in 7.2.4. Metaphors are based not on contiguity, but similarity or analogy. English germ is a good exam ple of a metaphor-based meaning change. The earlier meaning of this word was ‘seed’, clearly visible in a sentence like (2), from 1802:
(2) The germ grows up in the spring, upon a fruit stalk, accompanied with leaves (OED germ 1a).
The word’s application to the microscopic ‘seeds’ of disease is a metaphorical transfer: ailments are likened to plants, giving them ‘seeds’ from which they develop. The Old French word for ‘head’, test, is another example of metaphorical development. Originally, test meant ‘pot’ or ‘piece of broken pot’: the semantic extension to ‘head’ is said to be the result of a metaphor current among soldiers, in which battle was colourfully described as ‘smashing pots’ (Hock 1991: 229). Exactly the same metaphor explains the sense development of German Kopf ‘head’, which used to mean ‘cup’. The use of monkey, pig, sow, etc. in pejorative reference to people can also be seen as the result of a metaphor based on perceived similarity with the animals concerned. Another, very common, metaphor relates space and time. It is seen in the use of verbs with spatial meanings in temporal ones, as when English conveys the ‘immediate’ future tense using go (I’m going to stop now), or French uses venir ‘come’ to express events in the recent past (je viens de terminer ‘I have just finished’, literally ‘I come from finishing’). In these expressions temporal events are expressed in language on the analogy of spatial ones. Many investigators have commented on the deep-seated nature of this transfer, which is widely attested cross-linguistically (Bybee, Perkins and Pagliucca 1994).
The centrality of metaphor and metonymy in semantic change is due to the fact that they jointly exhaust the possibilities of innovative word use and thus subsume all the other descriptive categories. If you want to express yourself innovatively and be understood, then ‘there are only two ways of going about that: using words for the near neighbours of the things you mean (metonymy) or using words for the look-alikes (resemblars) of what you mean (metaphor)’ Nerlich and Clarke (1992: 137).
