Matthias Urban
The topic of this chapter is lexical semantic change, the change of the meaning of languages’ lexical items through time. To give a first idea of this process by example, the basic meaning of Old High German bein was ‘bone’ (the original or source meaning), while in Modern German, it is ‘leg’ (the innovative or target meaning). In this example, there is a complete change in meaning, with the original meaning disappearing almost entirely. However, arguably, semantic change is also present if the original meaning persists and a new one is merely added, since already then there is a difference from the original state.
Inherent in early studies of semantics from the nineteenth century is a strong focus on diachrony; hence, diachronic semantics at large predates the rise of synchronic semantics. Research from this period was mostly carried out by scholars from France and Germany. They developed a number of catalogues of types of semantic change, the terminology of which has become common in historical semantics.
A major dichotomy that dates back to this research is that between semantic broadening (or generalisation) and semantic narrowing (or specialisation): broadening involves an expansion of the range of application of a lexical item. For instance, German gelenk ‘joint’ is a Middle High German derivative of lanke ~ lanche ‘hip’ and originally denoted the ‘hip joint’ specifically, but has later been generalised to ‘joint’ in general (Kluge 2002). The mirror image of broadening is narrowing: this type of change involves a restriction of the denotational range of a lexical item from an erstwhile more comprehensive one. A classic example is English meat, which goes back to Old English mete ‘food’.
In a metaphorical change, the source and target meaning stand in a relationship of similarity to one another. For instance, German bein also denotes the ‘leg of a table’ (often occurring in the compound tischbein, with tisch meaning ‘table’). The two denotata are similar along several dimensions: they are similar in appearance in that they are both longish in shape, they are attached at the bottom of a larger structure (the trunk of a person or animal and the top of a table respectively), and they fulfil a similar function in supporting that larger structure. Conversely, there is also metonymic change, which is not based on similarity, but on the often somewhat ill-defined notion of contiguity. Broadly speaking, contiguity is a relationship between two meanings based on their co-occurrence in space and/or time, or on a relationship of causality (cf. Geeraerts 1997: 96). The change from ‘bone’ to ‘leg’ in German is metonymic, based on spatial contiguity. There are several subtypes of contiguously related meanings. A particularly noteworthy type is synecdoche, when the original and target meanings stand in a part-whole relationship to one another.
Together, broadening, narrowing, metaphor and metonymy form the ‘classic’ and most commonly recognised types of semantic change pertaining to denotation and coded meaning.
A further aspect of the meaning of a linguistic sign is its connotation, the emotional and discourse-related nuances associated with it. There are two possible directions of change with regard to connotations: one, amelioration, causes an erstwhile derogatory term or a term more generally associated with negative attitudes to lose these connotations and become ‘neutral’ or even to be positively beset. Conversely, pejoration goes in the opposite direction such that a term acquires negative connotations. A classic example that illustrates both amelioration and pejoration is the semantic development of West Germanic *knehta-. The probable original meaning of this term is simply ‘boy’, which is still attested in Old High German and Old English. In both languages, the term was extended to ‘boy hired as a servant, labourer’ and then also to ‘male servant of any age’. The latter sense prevails in Modern German knecht, and hence the word has acquired a negative connotation. In English, however, there was a further specialisation to ‘military servant’ and finally, the term began to be used with positive connotations to refer to the holder of a military rank in medieval feudalism, namely a knight. Another example of pejoration would be the development of the term semantics itself, which in English, some time after its coinage by Bréal (1897/1921), came to be used to refer to manipulative rhetoric in politics and advertising.
Related to the changes in connotation just discussed, words may come to be perceived as unpleasant, either because they refer to ‘delicate’ activities or places, or because they are linked to some other culturally sensitive material or behaviour. This may trigger tabooing and subsequent loss of the original term and/or euphemistic extension of another item. For instance, big has acquired an additional sense ‘fat, obese’ in (some varieties of) English.
Another dichotomy is that of hyperbole, in which semantic change comes about by exaggeration and overstating of a case, as when in present-day English to starve is used in the sense ‘to be (really) hungry’, and litotes, in which it conversely comes about by understatement. For the latter, Blank (1999: 67) provides the example of Vulgar Latin male habitus ‘in a bad state’ which gives rise to French malade and Italian malato ‘ill’. Hyperbole and litotes have a particularly clear pragmatic grounding: they come about by speakers’ choices for and against a particular wording to express a certain event or state of affairs (more on pragmatics in semantic change in section 2).
Changes in semantics may also be based on analogy, namely if words with similar or identical meanings undergo similar semantic changes. Stern (1931: 185–191) shows that adverbs meaning ‘rapidly’ in English as spoken before 1300 have also developed the meaning ‘immediately’, and other evidence from English is presented in Lehrer (1985).
Geeraerts (1997: 101) also discusses semantic borrowing under the category of analogy. For instance, in Hui (Sinitic), words for ‘thousand’ and ‘ten thousand’ are homophonous with those for ‘day’ and ‘bowl’ respectively. Now, Baonan (Mongolic), which lacked single morphemes for high numerals traditionally, copied this homophony through contact with Hui, using its indigenous words for ‘day’ and ‘bowl’ for ‘thousand’ and ‘ten thousand’ (Li 1985: 327–329). But lexical borrowing is a two-sided process, since words have two sides, the formal and the semantic. In the simplest scenario, then, a word is borrowed in form (with possible adaptations to match phonological constraints in the target language) and in meaning. However, lexical borrowing can arguably be typologised more finely, depending on whether phonological form and/or semantic structure is or is not borrowed.
Additionally, one can treat certain changes that would traditionally be classified as generalisations as contact-induced, namely when native lexical material is generalised to also serve as the name for a newly encountered item (e.g. an artefact; see Table 16.1). For instance, in Toaripi (Elema family, New Guinea) ma ove originally denoted a ‘reflection in water’ but has been extended to ‘mirror’ because “[b]efore glass or metal mirrors were available, reflections in water served as a mirror” (Brown 1968: 217). This shows that assignment of particular cases is not always unambiguous (this is a statement that typically accompanies classificatory schemes from early on, e.g. in Stern 1931: 164). Also, there is sometimes an entanglement of descriptive categories with the explanations of the changes. This becomes particularly noticeable in the last types of change to be discussed. Some authors attribute certain semantic changes to the avoidance of homonymy, a situation which obtains when two semantically unrelated words have the same form. For instance, Menner (1936: 242) argues that English light in the sense of ‘bright’ experienced considerable semantic narrowing after it became homonymous in Old English with light ‘not heavy’, because some of its former senses (e.g. ‘clear, lucid’) may have led to confusion with the latter in actual discourse. Similarly, there are proposals to the effect that synonymy, when two different words have the same meaning, is a cause of change in meaning. An early exponent is Bréal (1897/1921). Further, there are also claims that polysemy, roughly speaking when a word has several distinguishable but related meanings, is avoided diachronically. Anttila (1989: 181–182) mentions as an example of polysemy avoidance Finnish kutsua, generally meaning both ‘to call’ and ‘to invite’. This polysemy has been resolved in the Värmland dialect by borrowing kalloa ‘to call’ from Swedish, while kutsua retains the meaning ‘to invite’ only. Avoidance of homonymy, synonymy, and polysemy can be conceptualised as instances of a more general efficiency principle, namely that linguistic systems strive to associate one meaning with one form. Homonymy, synonymy and polysemy are violations of this principle because of their ambiguity. Recently, the ‘one form-one meaning’ principle and its proposed instantiations have been called into question because of their underestimation of the disambiguating role of context (Traugott and Dasher 2002: 53–54).
Finally, a lexical item may become enriched semantically by ellipsis, the shortening of a more complex construction. In this way, English gum has acquired the additional meaning ‘chewing gum’ by omission of chewing (Hock and Joseph 2009: 286).
These are merely the most commonly recognised categories of semantic change, and research in the early twentieth century has spawned a variety of frameworks for its classification. An overview is presented in Geeraerts (2010: 25–41), and a more complete description of the development of (diachronic) semantics in Europe in Nerlich (1992). As noted by Geeraerts (2010: 43), by and large these classifications do not use textual data, but present examples of semantic change in isolation without context (Sperber 1923 is a notable exception). Likewise, while early authors frequently discuss the causes that initiate changes in meaning, they neglect somewhat the question of how semantic change actually comes about (though this is an oversimplification and e.g. Paul 1920/1960: 74–75, Sperber 1923: 22–25, and Bloomfield 1933/1965: 430 were interested in semantic change as a process). Still, with the advent of corpus linguistics and the systematic study of pragmatics, scholars were in a better position to describe the ongoing processes in detail.
| Borrowing of phonological form, but not full semantic structure | German star is borrowed from English with the meaning ‘celebrity’. Since for ‘star’ in the sense of ‘celestial body’ ‘celestial body’ indigenous Stern is retained, there is a mismatch between the semantic structure of the word in the recipient and donor languages. This is common in lexical borrowing. |
| Borrowing of phonological form and semantic structure | Gawwada (Cushitic) borrowed nafse from Amharic(Semitic) näfs in both of its meanings ‘soul’ and ‘life’(Tosco 2009). |
| Borrowing of phonological form with extensive or complete shift in semantics | Dolgan (Turkic) muora ‘tundra,north’ <Russian more ‘sea’ (Eugénie Stapert p.c.). |
| Borrowing of semantic structure, but not phonological form | German lesen, originally ‘to gather, collect’, acquired the nowadays dominant reading ‘to read’ due to influence from Latin legere, which had both meanings |
| Borrowing of phonological form, superimposing indigenous semantic structure (=relexification) | In Bislama, a creole language of Vanuatu, harem, refl ecting English hear, means ‘to hear, taste, smell,feel either by touch or by emotions’, mirroring the semantic range of rogo in the Austronesian language Tangoan (Camden 1979:55), which Camden uses as a representative for other languages of the area. |
| Contact-induced broadening | In Interior Salish languages (as well as some other languages of the region), indigenous terms for ‘grass’ acquired an additional sense ‘hay’ as their speakers became familiar with livestock keeping (Turner and Brown 2004). |
The notion of polysemy plays a key role in understanding semantic change as a process, because a common way for semantic change to take place is by an intermediate stage of polysemy (Evans and Wilkins 2000, as well as Enfield 2003, also speak of ‘bridging contexts’). That is, a word does not change its meaning abruptly over night across an entire speech community (though from the perspective of individual innovating speakers and speakers adopting innovations, the change is instantaneous, Traugott and Dasher 2002: 34). Rather, it acquires a novel conventionalised sense, with the original one at least initially remaining intact (e.g. Sweetser 1990: 9; Wilkins 1996: 269; Campbell 1998: 268). Eventually the original meaning may, but need not, be lost. Diagrammatically, the process may be depicted as follows (Traugott and Dasher 2002: 12):
Thus, even though the semantic history of each single word is different, there is nevertheless a typical line of development on a more general level of description. But how do new polysemies arise? Pragmatics, the study of language in use, which is concerned not with the coded meanings lexical items and utterances have independently of context, but, among other things, with the meanings that arise only in context of communicative events, is vital here. As summarised by Traugott and Dasher (2002: 35), there is a common diachronic pathway from coded meanings via initially context-bound pragmatic inferences to new coded meanings (for a diagrammatic representation involving the stage in which the innovative meaning is only pragmatic see Enfield 2003: 29). That is, ultimately new senses of words and hence semantic change come about through the “semanticization of pragmatics” (Traugott 2010: 32). To understand the process in more detail, some more background on pragmatics is necessary. The relevant notion (also in semantic change in early stages of grammaticalisation) is the socalled conversational implicature (Traugott and König 1991; see Horn 2004 for more details of this notion).
Drawing on earlier sources, a general framework for the process of semantic change is developed in Traugott and Dasher (2002). In the first step, there are so-called invited inferences (similar but not identical to ad hoc conversational implicatures in neo-Gricean pragmatics). These arise ‘on the fly’ (Traugott and Dasher 2002: 17) in the process of communication. They may be based on general world knowledge or the particular circumstances of the conversational setting, and they may be either consciously exploited or emerge unconsciously. For instance, Traugott and Dasher (2002: 36–37) offer a corpus-based case study of Old and Middle English swa lange swa, the precursor of present-day English as long as. In Old and Middle English, there is coexistence of swa lange swa in a spatial sense ‘as long as’ and a temporal sense ‘for the same length of time as’, with the spatial sense presumably being the older one. Now, in particular contexts, it was possible to infer a conditional reading (‘provided that’). Then, in early modern English, there are the first examples of the conjunction (already spelled as long as) in which the conditional reading dominates the temporal one, although a temporal interpretation is still possible. Terminologically, at this stage one would say that the invited inference has become a generalised invited inference (similar but again not identical to generalised conversational implicature in neo-Gricean work). This means that the conditional interpretation has become regularly associated with the lexical item in the language (though not necessarily in other languages, generalised invited inferences are language-specific). At this stage, the inference no longer requires a particular context. However, it is still cancellable: this means that a purely temporal interpretation is still possible, which shows that the new conditional reading is not (yet) part of the coded meaning. Still later, around 1850, examples begin to appear in which the conditional reading seems the only one possible. That is, the generalised invited inference of conditionally has been conventionalised as the coded meaning of as long as.
While there thus is a general scheme available for the acquisition of new meanings, the loss of meanings, Traugott and Dasher (2002: 39) say, is irregular and hence unpredictable.
But does all semantic change come about in this way? The answer Traugott and Dasher (2002: 4) provide themselves is in the negative. For example, some instances of semantic change are due to contingent change in the extra-linguistic world, such as socio-cultural circumstances and technological innovations (see section 4 on causes of semantic change).
Since Traugott and Dasher’s data are drawn from the well-attested history of English and Japanese, they can, advantageously, draw on large corpora of texts from different periods. This is a prerequisite for following semantic developments through time in detail. However, there are not many languages with a long enough history of literacy for corpus approaches to be effective (see Fleisch 2008 on the situation in Bantu studies). Even if data are available, the majority of them are embedded in context, so one first has to extract the coded meaning of a given form from context. There are no speakers to consult, so how can one assess what is coded and what is only implicated from written records, and how can one judge at what point semanticisation has occurred? The only available evidence for the semanticisation of pragmatics in written records is indirect (Traugott and Dasher 2002: 44–48). Finally, a complication lies within the very notion of polysemy, which is hard to pin down theoretically. Lexical semanticists have developed tests for polysemy, but they obviously cannot be readily applied in a diachronic context; moreover, the tests are problematic in that they sometimes yield mutually incompatible results (Geeraerts 1993).
These are serious issues, but the model is still highly important for the understanding of semantic change it offers and the generalisations it has spawned (cf. section 5). For instance, bearing in mind that there may be gaps in the available data to the effect that not all possible meanings are recorded and that there may be differences between dialects, the available evidence for the semantic change from ‘bone’ to ‘leg’ in German is compatible with the polysemy-based model of semantic change: in the eighth century AD, bein is only recorded with the meaning ‘bone’ (Seebold 2001: 83), but in texts from the ninth century from various dialects, the word is also used to translate Latin tibia ‘shin bone’ specifically, but also already crus ‘leg’ (Seebold 2008: 153). Thus, bein acquired an additional (metonymically motivated) sense which would eventually oust the original one after a long period in which it still prevailed. Traugott’s work on semantic change has focused on the semantic development of more grammatical items, and there are few detailed empirical studies of semantic change via pragmatically-induced polysemy in the lexical realm. While examples such as the above suggest that pragmatically-induced polysemy is crucial here as well, details may differ: Hansen and Waltereit (2006: 244–245), for instance, suggest that mouse must have acquired the coded meaning ‘input device for computers’ directly from a particularised conversational implicature without it becoming generalised first, questioning Traugott and Dasher’s (2002) elaboration of the role of implicatures in semantic change on this and other grounds.
From the 1980s onward, the rise of cognitive linguistics has also had an impact on research in semantic change. A concept that has proven particularly fruitful is that of the prototype structure of meaning (though according to Geeraerts [1997: 27], the beginnings of prototype-based views of semantic change already appear in earlier work): departing from the conviction that there are necessary and sufficient features available to define a category, prototype theory postulates a degree of typicality among members of the category denoted by a lexical item, with blurry edges where membership is uncertain and may vary across judgements by different individuals. Moreover, around the typical or most salient meanings, less central meanings often revolve in clusters, so-called radial sets. Geeraerts (1997: 32–47) contains a corpus-based case study showing how prototype theory can be applied to describe semantic change. It deals with the semantic development of Dutch legging and its synonyms leggings and calegon (the latter is common in the variety of Dutch spoken in Belgium). When the respective garment and hence the English and French loanwords denoting it appeared around 1987, they most commonly referred to a long, tight-fitting, creaseless garment worn by women as a piece of outerwear and made from elastic material. Geeraerts hence identifies this cluster of features as characteristic of the prototypical ‘legging’. As time passes, the referential range of legging and its synonyms becomes wider, but not completely randomly. Rather, in each subsequent year the application becomes more flexible with regard to the features. Each of the innovative combinations of features is found on the periphery and is least common, while the abovementioned dominant prototypical core remains constant. Geeraerts concludes that prototypical semantic areas are diachronically more stable than peripheral ones. Another case study in Geeraerts (1997) shows the diachronic development of radial sets.
There is a long tradition of trying to locate the causes of language change in the different needs of the speaker and hearer in the process of communication (e.g. Martinet 1952; Haspelmath 1999): on the one hand, the speaker is said to attempt to communicate with the least possible effort on his behalf, while on the other hand making sure that communication is successful, that is, that he is understood by the hearer. In a variety of guises, this tension is adduced as a cause of semantic change as well. Geeraerts (1999) sees the emergence of prototypicality as functionally motivated in a speaker-oriented fashion: it optimises speech production by providing the necessary flexibility to adapt to changes in the extra-linguistic world elegantly, but also the stability at the core necessary to maintain the linguistic system in its general characteristics. Conversely, he views certain types of semantic change, such as that due to avoidance of homophony, as hearer-oriented because, by removing ambiguities, they help to process the message. Another elaboration concerning the influence of the general tension between the conversational needs of the speaker and hearer in the rise of implicatures and semantic change, and inspired by Zipf’s principle of least effort, is invoked in Horn (1984).
Causes for semantic change are also discussed by Blank (1999), with the general overarching motivation for semantic change also seen as the drive to increase efficiency and expressivity in communication. But Blank also recognises six more specific motivations: first, trivially, there may simply be the need for a new name for some newly encountered referent which can efficiently be handled by semantic extension and/or change. Second, there may be the need to make accessible abstract concepts that are hard to verbalise otherwise (and semantic extension and/or change is seen as an efficient means of doing so). Third, there may be changes in social organisation, structure, or (material) culture (as emphasised early by Meillet 1948), as when Toaripi ma ove, mentioned above, has kept up with technological innovation by expanding its denotation. Further, change may occur by referential indeterminacy in actual discourse, but yet with a “close conceptual or factual relation” between the relevant meanings; some traditional categories of semantic change discussed in section 1 fall under this rubric. Blank says that semantic change may also be instantiated in order to reduce complexity and irregularity in the lexicon on behalf of the speaker in order to communicate successfully with minimal effort (folk etymology can be conceptualised as one of the instances of this type). Finally, semantic change may be due to tabooing or emotional load (emphasised earlier by Sperber 1923).
While it is frequently acknowledged that semantic change has a sociolinguistic dimension (e.g. Blank 1999: 62), actual research from a sociolinguistic perspective is scarce. Robinson’s (2012) study of skinny in British English shows the applicability of sociolinguistic and variationist approaches to questions of semantic change that could be combined fruitfully with corpus-based approaches such as Traugott’s. Likewise, there is little empirical research on which group of speakers is particularly involved in bringing about semantic change. Traugott and Dasher (2002: 41) assume that because the process presupposes the mastery of discourse-pragmatic rules, it is not children who are responsible for innovations in semantics, but rather adults, particularly younger adults; this would converge with Robinson’s (2012) empirical results. Likewise, frequency may play a role. Ogura and Wang (1995), building on Williams (1976), suggest that frequently used items are more likely to undergo semantic change than non-frequent ones.
As Haspelmath (2004: 26–27) puts it, “diachronic semanticists have not even begun collecting the systematic cross-linguistic data that would allow us to arrive at empirically well-founded … universals of lexical semantic change.” With this statement, Haspelmath is in a century-old line of lamenting the lack of a systematic empirically-based investigation of semantic change (cf. Brugmann 1895; Tappolet 1905). A catalogue of general tendencies in semantic change that such an investigation might allow us to establish would be of great practical value as a guide for etymologists in the search for cognates, and in reconstructing meaning (cf. e.g. Sperber 1923: 60; Wilkins 1996: 266). The assemblage of such data is now on its way: Zalizniak et al. (2013) offer an online database of semantic shifts (a theoretical notion which does not only include semantic change proper) on a cross-linguistic basis; a description of the associated project is in Zalizniak et al. (2012) (moreover, Buck [1949] has proven an invaluable resource for semantic change in Indo-European languages). In addition, there are some suggestions about likely directions in semantic change or even claims as to unidirectional developments for individual meanings or semantic domains. Table 16.2 bundles some of the most important available proposals. They differ in whether their proponents conceived of them as violable tendencies or unidirectional laws of semantic change and in how the generalisations were arrived at; columns 3 and 4 provide pertinent information, where authors are clear on these points. Such an overview necessarily glosses over many details, and consultation of the original publications is recommended to obtain a fuller picture of the context and background of each study. Not listed are works arguing for certain universal pathways in which languages acquire terms for certain meanings, which strictly speaking fall in the domain of lexical, not semantic change. Notable work in this area includes Berlin and Kay (1969) on colour terms and Berlin (1972) on ethnozoology.
Generalisations from grammaticalisation research are in principle also relevant, because typically, in grammaticalisation the input of the change bears lexical meaning and undergoes semantic bleaching or other semantic alteration. Heine and Kuteva (2002) is a compendium of grammaticalisation processes in the languages of the world. For empirical data on meaning change in grammaticalisation, the reader is referred to their ample data and to Traugott and König (1991) and Eckardt (2006) for theoretical perspectives.
There are also a number of proposals on a more general level of description. Lehrer (1985: 286) argues on the basis of data from the history of English that “semantically related words are more likely to undergo parallel semantic changes than semantically unrelated ones precisely because of their semantic relationships. Semantic relationships tend to remain constant, so that if one word changes meaning, it will drag along other words in the domain.” For instance, metaphorical use of English gorilla with the sense ‘brute, brutish person’ caused ape and baboon to acquire this extension as well, even though ape at an earlier time already had the metaphorical reading ‘fool’ (Lehrer 1985: 288).
On the basis of mainly Indo-European data, Sweetser (1990: 45) argues that “[t]he internal self is pervasively understood in terms of the bodily external self, and is hence described by means of vocabulary drawn (either synchronically or diachronically) from the physical domain” (e.g. ‘to see’ > ‘to know’, ‘to hear’ > ‘to understand, obey’, ‘to taste’ > ‘to like’). This generalisation can be seen as a special case of a more general principle formulated e.g. by Paul (1920/1960: 96), according to which meanings in the spatial domain give rise to Table 16.2 Overview of proposed generalisations in semantic change meanings in the temporal and cognitive domain. Thus, English to grasp has a secondary sense ‘to understand’, and Latin explicare ‘to untangle, unfold’ also means ‘to explain’. This in turn is a special case of an even broader principle, identified by Bloomfield (1933/1965: 429) among others: there is a frequent pathway in semantics from the concrete to the abstract.
Table 16.2 Overview of proposed generalisations in semantic change
Furthermore, Traugott’s work discussed above has led to generalisations about semantic change. These are discussed in Traugott and Dasher (2002: 94–95) and other publications of Traugott’s. The most important generalisation is subjectification (see Traugott 2010 for more details): “Meanings tend to become increasingly based in the speaker’s subjective belief state/attitude toward the proposition.” An example would be the expansion of English while from temporal ‘during’, grounded in the real world situation described, to concessive (‘although’), grounded in the speaker’s perspective on that situation. Subjectification is regarded as the dominant tendency of semantic change that is fed by two other related ones. One of these is that “[m]eanings based in the external described situation” give rise to “meanings based in the internal (evaluative/perceptual/cognitive) described situation.” The traditional categories of amelioration and pejoration can be conceptualised as instances of this tendency, as can Sweetser’s (1990) generalisation mentioned above. The other is that “[m]eanings based in the external or internal described situation” yield “meanings based in the textual and metalinguistic situation.” An example of this type of change is the earlier shift from Old English pa while pe ‘at the time that’ to Modern English while ‘during’.
In spite of all the proposals mentioned so far, many researchers are sceptical that general principles of semantic change can be identified (regularity in semantic change may depend on part of speech, with change in closed-class grammatical items and in the verbal domain more regular than in nominals, cf. Traugott and Dasher 2002: 3–4). However, even if one chooses to not accept the existence of laws or even the weaker notion of directional tendencies in semantic change, it is arguably never the case that semantic change is literally random: even if we are unable to predict semantic developments that may happen in the future, any semantic change that is going to occur is necessarily based on and constrained by the general cognitive mechanisms by which meanings may be related to one another diachronically that have been mentioned in section 2, or by relations that arise from cultural facts (see e.g. Sperber 1923: 34 for similar statements). Since synchronic polysemy is commonly an intermediate step in semantic change, a smallest common denominator concerning methodology is formulated by Evans and Wilkins (2000: 550): “Insistence on synchronic attestation of polysemy places strong constraints on postulated semantic changes, providing an important antidote to the unbridled imagination in postulating etymologies.”
Yet one cannot help but note that more large-scale empirical work in the area is needed. Such research faces similar challenges as Traugott’s corpus-based approach, such as the lack of attested diachronic data from a wide variety of languages. Hence, to base claims as to what is a natural semantic change on a cross-linguistically balanced survey of diachronic data is possible only to some extent. Therefore quite a large number of the proposals in Table 16.2 are either derived mostly from the history of a single language (English), or not actually based on diachronically attested data, instead bypassing the issues by referring to other (synchronic) data. Moreover, if one, in the face of the dearth of cross-linguistic historical data, uses the pathway from reconstructed proto-meanings to the actually attested meanings as the basis of generalisations, one simply gets the information out of this enterprise that one (or etymologists) has previously invested in them. There is an issue with circularity in this. And at present no general method of semantic reconstruction is available, making reconstructed meanings doubly suspect as the input for sound generalisations (see further section 6).
The importance of inclusion of historical data from lesser-known languages is underscored by Evans and Wilkins (2000). Simplifying, they show that verbs for cognitive processes or states do not necessarily need to be derived historically from verbs for seeing, as argued by Sweetser (1990) on the basis of mainly Indo-European data, but can also be extensions of verbs for hearing in Australian languages. Generally, Evans (1990: 137) maintains “that the whole continent of Australia is characterisable as a linguistic area from the point of view of certain types of polysemy and semantic change that are common right across the continent, but rare or unreported elsewhere”; these are likely culturally based and diffused linguistically. Further notable work on semantic change from an areal perspective, in this case dealing with Southeast Asia, is in Matisoff (1978) and Enfield (2003). Theoretically, there is room for an areal aspect in semantic change, because “bridging contexts” are embedded in the context of everyday communication which is in turn embedded in everyday social conduct (cf. Evans and Wilkins 2000: 550). Then, using data only from a particular language family or language in formulating cross-linguistic generalisations is dangerous, because the patterns found may turn out not to be universally recurrent, but restricted to the bounds of a certain language family or linguistic area.
Given the early interest in historical semantics mentioned above, it is at first glance surprising that semantic reconstruction is a topic which has not received much attention and is in need of methodological founding. A related problem is that the methodology of semantic reconstruction, and to some degree also the search for cognates, is contingent on knowledge about naturalness in semantic change (Sweetser 1990: 25).
At the beginning of the twentieth century two leading etymologists, Antoine Thomas and Hugo Schuchardt, debated the question of whether phonetic plausibility alone is sufficient in assessing the accuracy of etymologies, or whether the semantic side should be weighted equally (cf. Tappolet 1905). Some 100 years later, it seems fair to say that the former often wins out, also when it comes to reconstructing proto-languages (cf. Koch 2004: 90–91 and Nichols’ 1996: 57 understanding of the Comparative Method). Thus, although in diachrony one encounters variation along two dimensions, the formal (phonetic-phonological) and the semantic, historical linguists have paid more attention to the more accessible formal shape of a linguistic item at the expense of its semantics.
In semantic reconstruction, one can aim at a generally applicable method or be interested in a particular semantic field in a particular language or language family only. Research of the latter kind, which is often interested not only in semantics per se, but also in inferences about the culture, society and environment of speakers of the proto-language, is summarised in Friedrich (1992); see also Epps (this volume). Generally, Lehrer (1978, 1985) highlights the importance of contrasts within semantic fields for understanding developments in semantics. Blust (1987) provides semantic reconstruction of words in the semantic field ‘house’ in Austronesian languages with recourse to the Saussurean distinction of signification and value, which reflects the structuralist view of meaning as defined by contrast to the meaning of other words within its semantic field and the overall lexicological system. Pawley (2011), working from an essentially extensionalist perspective, attempts to determine the range of meaning of higher-level terms for aquatic creatures in Proto-Oceanic. For instance, reflexes of *ikan in daughter languages are commonly glossed as ‘fish’ in lexical sources, but, depending on the language, may also denote other aquatic creatures. The obvious question is, of course, where the cut-off point of Proto-Oceanic *ikan itself was. From the distribution of the semantic range of reflexes, Pawley concludes that *ikan had a narrow sense ‘typical fish’, but more broadly could at least refer to sharks, rays, whales, dolphins, eels, turtles, and crocodiles (the situation seems to be describable well in a prototype theory framework, though Pawley himself does not do so). Finally, Dyen and Aberle (1974) (using an approach criticised e.g. in Blust [1987]), Ehret (2008), and McConvell (2013) reconstruct aspects of kinship systems in Athapaskan, African, and Pama-Nyungan languages respectively.
There are also a number of proposals on how to approach semantic reconstruction more generally beyond individual semantic fields. But before these are discussed, it is instructive to look at what etymologists do in actual practice rather than theory when reconstructing semantics. Koch (2004: 91–96) surveyed some etymological dictionaries, and arrived at the following typology: first, there is the additive type, in which one “simply assigns . as proto-meaning an addition of the cognates’ contemporary meanings” (2004: 92). In other cases, etymologists select one of the present-day meanings of cognates as godfathers of the proto-form; this is Koch’s selective type. Next, there is the so-called taxonomic-abstracting type (alongside the similar engynomic type), which is well known from traditional Indo-European reconstructions and yields reconstructed meanings such as ‘swelling, rotundity’ generally for cognates denoting body parts with roughly roundish shape (this is the approach criticised most prominently by Sweetser [1990: 24]). The most famous and sophisticated example of such a reconstruction is Benveniste’s (1954) analysis of cognates meaning ‘path’ and/or ‘road’ in some Indo-European languages, but ‘bridge’ in Latin, ‘ford’ in Armenian, and ‘sea’ in Ancient Greek. The structuralist background of Benveniste’s approach is brought to light by Fox (1995: 115–116), who schematises Benveniste’s account in terms of semantic features: since the only feature that is consistently present in the semantics of all daughter language terms is that they are passages in the broadest sense, this is the meaning assigned to the proto-term (Sweetser [1990: 24] notes the similarity to feature analysis in phonology). However, this schematisation does not do full justice to Benveniste’s discussion. In fact, Benveniste (1954: 255) emphasises that one has to search for contexts that may have triggered a differentiation in meaning. In this, his reasoning is even akin to modern approaches emphasising the role of context and synchronic polysemy. Anttila (1989: 370) agrees that in semantic reconstruction one ought to compare intersections between different meanings of the cognates in daughter languages (just like in phonetic reconstruction), but cautions that “the context need not be elevated into a protomeaning’ (emphasis original). Meillet (1937/1966: 382) explicitly states that a semantic reconstruction along these lines is only an abstraction and should not be taken as ‘real’ in the sense that the proto-word actually ever bore the reconstructed meaning.
Cognitive linguistics has inspired a number of approaches to semantic reconstruction: Jurafsky (1996) proposes a radial category for the semantics of the diminutive, in which there is a particularly strong semantic-pragmatic link with the meaning ‘child’. His approach shows its virtues in particular when combined with the previously formulated unidirectional line of development from concrete to abstract meanings. He applies this generalisation to a particular problem in semantic reconstruction, namely that of the Indo European suffix *-ko-, which can appear with a wide range of meanings in modern languages, including ‘diminutive’. Jurafsky’s radial category allows for a diachronic interpretation supporting an original core sense ‘child’ with more abstract semantic developments being secondary. Niepokuj (1994) is concerned with Indo-European verbs traditionally reconstructed as meaning ‘curve’ or ‘bend’. There are very many of these, which suggests that they bore semantic nuances not hitherto worked out, and Niepokuj employs cognitive linguistics theory to do so. Györi and Hegedüs (2012), aided by the generalisations from Wilkins (1996), attempt to show a connection between Indo-European roots for ‘chin, jaw’ and ‘knee’ by appealing to image schemata, another Cognitive Linguistics notion.
Urban (2011) suggests a way to refine semantic reconstruction by a different route: drawing on data from a synchronic-typological sample, he proposes that directionality in word-formation in some cases provides a means to find out about likely and unlikely pathways of semantic change. In a second step, diachronic semantic maps of likely developments can be drawn along which one can work backwards in semantic reconstruction and ultimately arrive at probable, though not incontrovertible, proto-meanings.
However, it seems unlikely that it will ever be possible to operationalise semantic reconstruction by mechanically applying ‘laws’; more likely, it will continue to require philological competence on behalf of the historical linguist carrying out cognate searches and/or semantic reconstruction. What follows are some further observations that may be useful when reconstructing semantics in a particular language or language family. Not all of them are specific to semantic reconstruction itself, but can also be applied to other areas of reconstruction.
The introductory example showed that German bein underwent semantic change from the original basic meaning ‘bone’ to ‘leg’. In the older stages of other West and North Germanic languages, ‘bone’ is uniformly found as the general meaning of bein’s cognates (Orel 2003: 32).2 If this is true, positing an original meaning ‘leg’ requires the assumption of a change at many nodes of the family tree, while positing an original meaning ‘bone’ necessitates only change at a low node of the tree. By the principle of parsimony, then, reconstructing ‘bone’ would be preferred on methodological grounds. Therefore, if we did not already know from the available internal evidence regarding the development of German that ‘bone’ is the likely original meaning, the comparative evidence would direct us towards the same conclusion.
The original meaning of a lexical item may sometimes be preserved in conventionalised figures of speech or fixed phrases (cf. Bloomfield 1933: 432–433). Likewise, it can be useful to look for the lexical item one is interested in as parts of compounds, derivatives, and other kinds of more complex constructions. For instance, the German name for the ‘human coccyx’ is steißbein, consisting of bein and steiß; ‘rump’ as the other constituent. Further, there is the collective derivative gebeine ‘bones, mortal remains’, and other complex terms in which bein is (historically) a constituent. In all these terms, assuming the meaning ‘bone’ for bein at the time of their coinage leads to intuitively natural semantic associations, whereas positing ‘leg’ does not make too much sense. Again, if we did not already know the original meaning to be ‘bone’ by the available textual evidence, the internal evidence would provide a hint towards the correct analysis (if, as here, two pieces of evidence converge, the argument is strengthened considerably). Internal semantic reconstruction is also carried out by Traugott (1986) on the basis of generalisations such as those mentioned in section 5.
Most languages are dialectally diverse, and this diversity can manifest itself on all levels of linguistic structure, including semantics. Several surveys of dialect geography have shown that often the dialects spoken on the geographic or sociolinguistic periphery are more likely to preserve archaic features than those spoken at the centre (Hock 1986: 440–441, who, however, also justifiably cautions against “preconceived notions as to what should be central or outlying/remote areas”). This generalisation can be potentially useful for reconstructing semantic change as well. For instance, in Dyirbal (North Queensland, Australia), dialects on the northern and southern periphery have the term juja for ‘back, dorsum’, while more central dialects have mambu. Since in some northern dialects mambu occurs with the meaning ‘loins’, Dixon (1982: 66–67) infers that ‘loins’ is the likely original meaning of mambu, having shifted in the central dialects to ‘back, dorsum’ (due to tabooing of juja).
A frequently employed technique for consolidating a posited original meaning is to point to similar developments in other unrelated languages (cf. Benveniste 1954: 251; Hock 1986: 308). For instance, one might attempt to consolidate the original meaning ‘bone’ for German bein by pointing out that in Nzebi (Bantu), the word for ‘bone’ has the same specialised sense ‘shinbone’ that was an intermediate stage in the development from ‘bone’ to ‘leg’ in German (Bastin 2001: 31). Such an argumentation is of course closely akin to investigations of regularity in semantic change on a broad scale as described in section 4, except that it lacks the systematicity by which such investigations are characterised.
Evans and Wilkins’ (2000: 550) abovementioned suggestion can be helpful in a similar vein: since polysemy is a frequent (though not the exclusive) source of semantic change, synchronic polysemies can be helpful to validate a postulated etymology. This is a particularly important point because here the relevant data is synchronic, and hence much more readily available.
Benveniste (1954: 252) notes that in French, voler means both ‘to fly’ and ‘to steal’. In the former sense, the verb is surrounded by a large word family derived from the root, including voleter, s’envoler, survoler, volée, volatile, volaille, and volière, but in the latter sense, there is only voleur ‘thief’ derived from it. This suggests to Benveniste that one should look at contexts which may have conditioned a semantic split from ‘to fly’ to ‘to steal’, which implies that ‘to fly’ is the original sense (in this case, Benveniste identifies the relevant context in the language of falconry).
Anttila (1989: 366–368) supplies methodological help that is akin to Benveniste’s: if there are two sets of cognates for what appear to be identical or highly similar meanings, it is worthwhile to study carefully the differences of occurrence of the two roots in attested daughter languages and derivatives to flesh out rather subtle semantic nuances (compare Niepokuj’s [1994] work as well as Pawley [2011] for a similar situation, in which the aforementioned Proto-Oceanic *ikan goes along with *pinagoda, which also denoted sea creatures but apparently stood in contrast with *ikan). In this, morphological and syntactic evidence can be of use in establishing semantics.
The aim of this chapter was to provide an overview of the current knowledge about semantic change and its theoretical underpinnings. The categories of semantic change described in section 2 are well-known and have become part and parcel of historical semantics because they offer useful terminological distinctions to describe individual cases of semantic change. Clearly progress beyond those has been achieved, not least by the recognition of the importance of pragmatics and synchronic polysemy. Also, a number of generalisations on and tendencies in semantic change have been proposed. Still, the general rift between believers and sceptics (e.g. Hock and Joseph 2009: 240) as to the possibility about generalisations in semantic change has remained remarkably constant over the last 100 years or so, and many historical linguists consider semantics the most underdeveloped and least well understood area of language change. Also, there is still a gap between etymological practice (Mailhammer, this volume) and theoretical concerns highlighting the perceived non-scientific and arbitrary nature of judgements about semantic relatedness in cognacy assessment (cf. Brugmann 1985: 17; Sweetser 1990: 24–26). The associated desideratum is to convert the semantic side of etymology and semantic reconstruction to an enterprise based on empiricism and a set of general procedures. In both regards, the field is open to investigation.
1 Thanks are due to Claire Bowern, Michael Cysouw, Bethwyn Evans, and Elizabeth Closs Traugott for commenting on earlier drafts of this chapter. Mistakes and shortcomings are my responsibility, not theirs.
2 However, it must be borne in mind that the association between ‘leg’ and ‘bone’ is present in cognates in modern North Germanic languages as well, perhaps indicating that it is of rather early Germanic vintage.
Allan, Kathryn and Justyna A. Robinson (eds). 2012.Current methods in historical semantics. Berlin/Boston: Walter de Gruyter.
Nerlich, Brigitte. 1992. Semantic theories in Europe 1830–1930. From Etymology to Contextuality. Amsterdam/Philadelphia: John Benjamins.
Traugott, Elizabeth Closs and Richard B. Dasher. 2002.Regularity in semantic change. Cambridge: Cambridge University Press.
Allan, Kathryn and Justyna A. Robinson (eds). 2012.Current methods in historical semantics. Berlin/Boston: Walter de Gruyter.
Andersen, Elaine S. 1978. Lexical universals of body-part terminology. In Joseph H. Greenberg (ed.)Universals ofhuman language, Vol. 3:Word Structure. Stanford: Stanford University Press, 335–368.
Anttila, Raimo. 1989.Historical and comparative linguistics. 2nd Edition. Amsterdam/Philadelphia: John Benjamins.
Bastin, Yvonne. 2001. Les themes pour ‘os’ en bantou: perspectives diachroniques. General Linguistics 38(1): 1–40.
Benveniste, Émile. 1954. Problèmes sémantiques de la reconstruction. Word 10 (2/3): 251–264.
Berlin, Brent. 1972. Speculations on the growth of ethnobotanical nomenclature.Language in Society 1(1): 51–86.
Berlin, Brent and Paul Kay. 1969.Basic color terms: their universality and evolution. Berkeley/Los Angeles: University of California Press.
Blank, Andreas. 1999. Why do new meanings occur? A cognitive typology of the motivations for lexical semantic change. In Blank and Koch, 61–89.
Blank, Andreas and Peter Koch (eds) 1999.Historical semantics and cognition. Berlin/New York: Mouton De Gruyter.
Bloomfield, Leonard. 1965 [1933].Language. London: George Allen & Unwin.
Blust, Robert. 1987. Lexical reconstruction and semantic reconstruction: the case of Austronesian ‘house’ words. Diachronica 4(1/2): 79–106.
Bréal, Michel. 1921 [1897]. Essai de Sémantique. Science des significations. 5th Edition. Librairie Hachette.
Brown, Cecil H. and Stanley R. Witkowski. 1983. Polysemy, lexical change and cultural importance. Man 18(1): 72–89.
Brown, H. A. 1968. A dictionary of Toaripi with English-Toaripi index. Sydney: University of Sydney.
Brugmann, Karl. 1895. Announcement of Die Ausdrücke für den Begriff der Totalität in den indogermanischen Sprachen. Eine semasiologisch-etymologische Untersuchung. Anzeiger für indogermanische Sprach-und Altertumskunde 5: 17–19.
Buck, Carl D. 1949. A dictionary of selected synonyms in the principal Indo-European languages, Chicago: University of Chicago Press.
Camden, William G. 1979. Parallels in structure of lexicon and syntax between New Hebrides Bislama and the South Santo language as spoken at Tangoa. Papers in Pidgin and Creole Linguistics 2: 51–117.
Campbell, Lyle. 1998. Historical linguistics. An introduction. Edinburgh: Edinburgh University Press.
Dixon, R. M. W. 1982. Problems in Dyirbal dialectology. In John A. Anderson (ed.) Language form and linguistic variation. Papers dedicated to Angus Mcintosh, Amsterdam/Philadelphia: John Benjamins, 43–73.
Durie, Mark and Malcolm Ross (eds). 1996. The comparative method reviewed: regularity and irregularity in language change. Oxford: Oxford University Press.
Dyen, Isidore and David F. Aberle. 1974.Lexical reconstruction. The case of the Proto-Athapaskan kinship system. London: Cambridge University Press.
Eckardt, Regine. 2006. Meaning change in grammaticalization. An enquiry into semantic reanalysis. New York: Oxford University Press.
Ehret, Christopher. 2008. Reconstructing ancient kinship in Africa. In Nicholas J. Allen, Hilary Callan, Robin Dunbar and Wendy James (eds) Early human kinship. From sex to social reproduction. Malden, Massachusetts/Oxford: Blackwell, 200–231.
Enfield, N. J. 2003. Linguistic epidemiology. Semantics and grammar of language contact in Mainland Southeast Asia, London/New York: Routledge Curzon.
Evans, Nicholas. 1990. Without this child – some regularities of semantic change in the Australian linguistic area. In Peter Austin, R. M. W. Dixon, Tom Dutton and Isobel White (eds) Language and history: essays in honour of Luise A. Hercus. Canberra: Pacific Linguistics, 137–155.
Evans, Nicholas and David Wilkins. 2000. In the mind’s ear: the semantic extensions of perception verbs in Australian languages. Language 76(3): 546–592.
Fisiak, Jacek (ed.). 1985. Historical semantics, historical word-formation. Berlin: Mouton Publishers.
Fleisch, Axel. 2008. The reconstruction of lexical semantics in Bantu. In Dymitr Ibriszimov (ed.) Problems of linguistic-historical reconstruction in Africa. (Special volume of Sprache und Geschichte in Afrika). Cologne: Rüdiger Köppe, 67–106.
Fox, Anthony. 1995. Linguistic reconstruction. An introduction to theory and method. Oxford: Oxford University Press.
Friedrich, Paul. 1992. Semantic reconstruction. In William Bright (ed.) International encyclopedia of linguistics, Vol. 3. New York/Oxford: Oxford University Press, 389–394.
Geeraerts, Dirk. 1993. Vagueness’s puzzles, polysemy’s vagaries. Cognitive Linguistics 4(3): 223–272.
——1997. Diachronic Prototype Semantics. A contribution to historical lexicology. Oxford: Oxford University Press.
——1999. Diachronic Prototype Semantics. A digest. In Blank and Koch, 91–107.
——2010.Theories of Lexical Semantics. Oxford: Oxford University Press.
Györi, Gabor and Irén Hegedüs. 2012. A cognitive approach to the methodology of semantic reconstruction: the case of English chin and knee. In Allan and Robinson, 313–333.
Hansen, Maj-Britt M. and Richard Waltereit. 2006. GCI theory and language change.Acta Linguistica Hafniensia 38(1): 235–268.
Haspelmath, Martin. 1999. Optimality and diachronic adaptation.Zeitschrift für Sprachwissenschaft 18(2): 180–205.
——2004. On directionality in language change with particular reference to grammaticalization. In Olga Fischer, Muriel Norde and Harry Perridon (eds) Up and down the cline – The nature of grammaticalization. Amsterdam/Philadelphia: John Benjamins, 17–44.
Heine, Bernd. 1997.Cognitive foundations of grammar. New York/Oxford: Oxford University Press.
Heine, Bernd and Tania Kuteva. 2002.World lexicon of grammaticalization. Cambridge: Cambridge University Press.
Hock, Hans Henrich. 1986.Principles of Historical Linguistics, Berlin: Mouton de Gruyter.
Hock, Hans Henrich and Brian D. Joseph. 2009. Language history, language change, and language relationship. An introduction to Historical and Comparative Linguistics, 2nd Edition. Berlin: Mouton de Gruyter.
Horn, Laurence R. 1984. Toward a new taxonomy for pragmatic inference: Q-based and R-based implicature. In Deborah Schiffrin (ed.) Meaning, form, and use in context: linguistic applications. Washington, DC: Georgetown University Press, 11–42.
——2004. Implicature. In Laurence R. Horn and Gregory Ward (eds), The handbook of pragmatics. Malden, Massachusetts: Blackwell, 3–28.
Jurafsky, Daniel. 1996. Universal tendencies in the semantics of the diminutive. Language 72(3): 533–578.
Kluge, Friedrich. 2002. Etymologisches Wörterbuch der deutschen Sprache. 24th Edition. (Edited by Elmar Seebold.) Berlin: Walter de Gruyter.
Koch, Peter. 2004. Diachronic cognitive onomasiology and semantic reconstruction. In Wiltrud Mihatsch and Reinhild Steinberg (eds) Lexical data and universals of semantic change. Tübingen: Stauffenburg, 76–106.
Lehrer, Adrienne. 1978. Structures of the lexicon and transfer of meaning. Lingua 45(2): 95–123.
——1985. The influence of semantic fields on semantic change. In Fisiak, 283–296.
Li, Charles N. 1985. Contact-induced semantic change and innovation. In Fisiak, 325–338.
Martinet, André. 1952. Function, structure, and sound change. Word 8(1): 1–32.
Matisoff, James A. 1978. Variational semantics in Tibeto-Burman: the ‘organic’ approach to linguistic comparison. Philadelphia: Institute for the Study of Human Issues.
McConvell, Patrick. 2013. Granny got cross: semantic change of kami ‘mother’s mother’ to ‘father’s mother’ in Pama-Nyungan. In Robert Mailhammer (ed.) Lexical and structural etymology. Berlin: Walter de Gruyter, 147–184.
Meillet, Antoine. 1948. Comment les mots changent de sens. In Linguistique historique et linguistique generale, Paris: Librairie Ancienne Honoré Champion, 230–271.
——1966 [1937]. Introduction â l’étude comparative des langues indo-européennes. University of Alabama Press.
Menner, Robert J. 1936. The conflict of homonyms in English. Language 12(4): 229–244.
Nerlich, Brigitte. 1992. Semantic theories in Europe 1830–1930. From etymology to contextuality. Amsterdam/Philadelphia: John Benjamins.
Nichols, Johanna. 1996. The comparative method as heuristic. In Durie and Ross, 39–71.
Niepokuj, Mary. 1994. Reconstructing Semantics. Or, a bad case of the bends. Berkeley Linguistics Society 20: 374–386.
Ogura, Mieko and William S-Y. Wang. 1995. Lexical diffusion in semantic change: with special reference to universal changes. Folia Linguistica Historica 16(1/2): 29–73.
Orel, Vladimir E. 2003.A handbook of Germanic etymolog. Leiden/Boston: Brill.
Paul, Hermann. 1960 [1920]. Prinzipien der Sprachgeschichte. 6th Edition. Darmstadt: Wissenschaftliche Buchgesellschaft.
Pawley, Andrew. 2011. Were turtles fish in Proto Oceanic? Semantic reconstruction and change in some terms for animal categories in Oceanic languages. In Malcolm Ross, Andrew Pawley and Meredith Osmond (eds) The lexicon ofProto Oceanic, Vol. 4:Animals. Canberra: Pacific Linguistics, 421–452.
Robinson, Justyna A. 2012. A sociolinguistic approach to semantic change. In Allan and Robinson, 199–232.
Seebold, Elmar. 2001. Chronologisches Wörterbuch des Deutschen Wortschatzes: der Wortschatz des 8. Jahrhunderts (und früherer Quellen). Unter Mitarbeit von Brigitte Bulitta, Elke Krotz, Judith Stieglbauer-Schwarz und Christiane Wanzeck, Berlin: Walter de Gruyter.
——2008.Chronologisches Wörterbuch des Deutschen Wortschatzes: der Wortschatz des 9. Jahrhunderts. Unter Mitarbeit von Brigitte Bulitta, Elke Krotz und Elisabeth Leiss, Berlin: Walter de Gruyter.
Sperber, Hans. 1923. Einführung in die Bedeutungslehre, Bonn/Leipzig: Kurt Schroeder.
Stern, Gustaf. 1931. Meaning and change of meaning. With special reference to the English language. Bloomington: Indiana University Press.
Sweetser, Eve E. 1990. From etymology to pragmatics. Metaphorical and cultural aspects of semantic structure, Cambridge: Cambridge University Press.
Tappolet, Ernst. 1905. Phonetik und Semantik in der etymologischen Forschung. Archiv für das Studium der neueren Sprachen und Literaturen 115: 101–123.
Tosco, Mauro. 2009. Gawwada vocabulary. In Martin Haspelmath and Uri Tadmor (eds) World Loanword Database. Munich: Max Planck Digital Library. [Available at: http://wold.livingsources.org/vocabulary/3; accessed 4 September 2013.]
Traugott, Elizabeth Closs. 1986. From polysemy to internal semantic reconstruction. Berkeley Linguistics Society 12: 539–550.
——2010. (Inter)subjectivity and (inter)subjectification: a reassessment. In Kristin Davidse, Lieven Vandelanotte and Hubert Cuyckens (eds) Subjectification, intersubjectification and grammaticalization. Berlin: Walter de Gruyter, 29–70.
Traugott, Elizabeth Closs and Richard B. Dasher. 2002.Regularity in semantic change. Cambridge: Cambridge University Press.
Traugott, Elizabeth Closs and Ekkehard König. 1991. The semantics-pragmatics of grammaticalization revisited. In Elizabeth Closs Traugott and Bernd Heine (eds)Approaches to Grammaticalization, Vol. 1:Focus on theoretical and methodological issues, Amsterdam/Philadelphia: John Benjamins, 189–218.
Turner, Nancy J. and Cecil H. Brown. 2004. Grass, hay, low growth and weeds: utility and semantics of Interior Salish botanical terms. In Donna B. Gerdts and Lisa Matthewson (eds) Studies in Salish linguistics in honor ofM. Dale Kinkade. (Occasional Papers in Linguistics 17). Missoula: University of Montana, 410–428.
Urban, Matthias. 2011. Asymmetries in overt marking and directionality in semantic change. Journal of Historical Linguistics 1(1): 3–47.
Viberg, Âke. 1984. The verbs of perception: a typological study. In Brian Butterworth, Bernard Comrie and Östen Dahl (eds)Explanations for language universals. Berlin: Mouton Publishers, 123–162.
Wilkins, David P. 1996. Natural tendencies of semantic change and the search for cognates. In Durie and Ross, 264–304.
Williams, Joseph M. 1976. Synaesthetic adjectives: a possible law of semantic change. Language 52(2): 461–478.
Witkowski, Stanley R. and Cecil H. Brown. 1985. Climate, clothing, and body-part nomenclature. Ethnology 24(3): 197–214.
Witkowski, Stanley R., Cecil H. Brown and Paul K. Chase. 1981. Where do tree terms come from? Man 16(1): 1–14.
Zalizniak, Anna A., Maria Bulakh, Dmitrij Ganenkov, Ilya Gruntov, Timur Maisak and Maxim Russo. 2012. The catalogue of semantic shifts as a database for lexical semantic typology. Linguistics 50(3): 633–669. (Special issue: Maria Koptjevskaja-Tamm and Martine Vanhove (eds) New directions in lexical typology).
——2013. Database of semantic shifts in the languages of the world. [Available online at: http://semshifts.iling-ran.ru; accessed 10 February 2013]
