1.0 Application of the Comparative Method (Part 2) Li2 - Language Variation Genetic Relationships II Dr. Deborah Anderson http://www.ling.cam.ac.uk/li2 (The following exercise is taken from Campbell 2004:125:147) Step One: Assemble cognates Italian Spanish French (Latin) Eng. gloss capra cabra chèvre capra goat caro caro cher caru dear capo cabo chef caput head/top carne carne chair carō/carn- meat/flesh cane *can chien canis dog *archaic form Step Two: Establish sound correspondences But some sound correspondences are purely coincidental: Examples: Italian /k-/, Spanish /k-/, French /ʃ/ Italian /p/, Spanish /b/, French /v/ Hawaiian aeto eagle mele sing kau summer Ancient Greek aetos eagle melos melody kauma heat Source: Trask(1996: 220-1) (Note that the hyphen indicates that the sound is in initial position.) To eliminate such false correspondences, one must look for evidence that the same sound correspondence recurs in other sets of cognates. Step Three: Reconstruct the protosound. Some guidelines: 1. Follow established generalizations regarding the directionality and naturalness of certain sound changes. For example, /k/ /ʃ/ is a typical change but not /ʃ/ /k/. Even more typical would be /k/ /ts/ /ʃ/, a change actually attested in French. 2. Majority wins. Unless there is compelling evidence to the contrary, one should assume that the sound that is attested in the majority of daughter languages should be chosen as the reconstructed proto-form. Example: With reference to the cognate set above, notice that Italian and Spanish both have /k/, where French has /ʃ/. In fact, the more widespread occurrence of /k/ is well recognized in Romance daughter languages. University of Cambridge 1
It s important to exercise caution in following this second guideline, however, since very common sound changes might reasonably occur on an independently basis across a number of languages. Also, it is possible that changes have occurred in such a way that no language in a set of daughter languages actually retains the original sound. 3. When multiple sound changes can be postulated, the one that requires the fewest independent changes is the one most likely to have occurred. Example: With reference to the cognate set above, if we assume that /ʃ/ /k/, then we must assume 2 independent sound changes took place in Italian and Spanish. If we assume /k/ /ʃ/, however, then we need postulate only one such independent change in French. 1.1 Grammatical evidence of genetic relationships A further important source of evidence for genetic relationships comes from what what Campbell (1995:350-2) terms shared aberrancy or submerged features in the morphological make-up of two or more languages. Example: English good/better/best and German gut/besser/best. A similar and highly arbitrary morphological alternation, and so unlikely to be a chance resemblance. 2.0 Issues in linguistic reconstruction Linguistic reconstruction is a deductive technique in which known sounds/forms are used to establish a hypothetical earlier (and no longer extant) stage of a language or languages, from which actually occurring forms are proposed to be systematically derived. The Comparative Method is standardly (although not exclusively) used as the basis for linguistic reconstruction. Linguistic reconstruction serves two aims: (1) It allows us to acquire knowledge of the earliest stages in the development of a language; (2) It provides a means of addressing inconsistencies in existing historical data. But, in the overwhelming majority of instances the study of language history is hampered by the unavailability of such evidence (Fox 1995:2). For example, most of world s languages do not have a written form. This poses a particular problem when the goal is to compare certain non-indo- European languages, such as African or American Indian languages, since there is essentially no historical data to work with. University of Cambridge 2
One area of debate in the literature: Also, even when there exists documentary evidence of earlier stages of a particular language, it is unlikely to reveal its full historical development. In the latter case, in order to determine the full historical development of a language, it is necessary to investigate its prehistory, as well. Not all researchers agree that it is necessary, or even desirable, to consider reconstructed forms historically accurate. Two views (Fox 1995:9): Formalist Reconstructions assumed to have no psychological validity; are simply helpful abstractions that can aid linguistic study. Realist Admit possibility that reconstructed forms represent actual once-spoken forms. 3.0 Theoretical Developments in Linguistic Classification In this section, we will consider certain wellrecognized limitations of the Stammbaum or family tree model of language relationships, as discussed in Fox (1995:122-44). In particular, it has been variously claimed that certain of the basic assumptions of the model are flawed, and that, therefore, it presents an inaccurate and/or unrealistic representation of language change and language relationships. Assumptions inherent in the family tree model a. Languages have only one parent, with language change proceeding in terms of languages or dialects cleanly splitting away from the parent language. It is well known, however that language contact is one of the major factors influencing language change. Such contact may result in language mixing or convergence, phenomena that cannot be represented in the standard family tree. The sources of Modern English: Representation 1 (Katičić 1970, cited in Fox 1995:124) English is a prime example. The form of Modern English is the result of a number of influencing factors, one of which is the direct contact that took place in earlier stages of the language between speakers of English and those of Celtic, Norse, and Norman French. [IE: Proto-Indo-European; ME: Middle English;ModE: Modern English; OE: Old English; PG: Proto-Germanic] University of Cambridge 3
The sources of Modern English: Representation 2 (Katičić 1970, cited in Fox 1995:124) [IE: Proto-Indo-European; ME: Middle English; ModE:Modern English; OE: Old English, PG: Proto-Germanic] Another criticism of the family tree model: b.changes take place independently within a branch, with no reference to other branches. This problem was recognized quite early on and was addressed by Johannes Schmidt (1872), who proposed the wave model of language relationships. In this conception of language change, linguistic innovations spread out among language and/or dialects like waves radiating on the surface of water after a pebble has been thrown into a pond. Schmidt s (1872)Wave Model Shared features in Indo-European languages (Fox 1995:132) A further criticism of the family tree model: c. It only admits the possibility of a single, uniform, proto-language. However, early work in dialect geography, and more recent work in sociolinguistics, has revealed linguistic variation to be widespread and pervasive. There is, therefore, every reason to believe that dialectal variation was attested in proto-languages, just as it is in the case of modern-day descendant languages. But are the above criticisms valid? Not necessarily. For example, Fox (1995:137-42) claims that it is unfair to criticize the model for too narrowly representing language relationships when, in fact, it is intended to highlight only those aspects of language change that suggest the existence of a genetic relationship. University of Cambridge 4
Diachronic isogloss maps (Southworth 1964; cited in Fox 1995:237) Furthermore, there have been some noteworthy attempts at modification and/or updating of the graphic representation of the family tree. Example:, Southworth (1964) offers a representation that combines tree-like schematics with isogloss divisions. 4.0 Alternative theories of language relationships Punctuated Equilibrium Model (Dixon 1997;2002) Dixon claims that the Comparative Method, while helpful in the study of Indo-European languages, is much less so in determining the genetic relationships that exist between the indigenous languages of Australia. On the basis of his study of the Australian languages, he proposes a general theory of language change, termed punctuated equilibrium. For the most part, linguistic communities exist harmoniously in periods of equilibrium, during which languages change slowly and largely by borrowing from one another. Linguistic areas are created that cross genetic families. These long periods of equilibriumare interspersed with short periods of punctuation, where harmonious coexistence is disturbed by, for example, invasion, migration, technological development, etc. It is only during periods of punctuation that languages will split in the manner depicted on the family tree. Use of the Comparative Method for determining language relationships is therefore appropriate only when this type of change has taken place. University of Cambridge 5
Reading recommendations But see Bowern (2006) for an alternative account of the Australian languages and for a detailed critique of the punctuated equilibrium model of language change. Campbell, L. 2004. : An Introduction. 2 nd edition. Edinburgh University Press. (Chapters 5 & 6 on the Comparative Method and linguistic classification. Chap. 13 on distant genetic relationships.) Fox, A. 1995. Linguistic Reconstruction: An Introduction to Theory and Method. Oxford University Press. (Chapter 6: Issues in comparative reconstruction & Chapter 9: Reconstructing Language Relationships ) Crowley, T. 2000. An Introduction to Historical Linguistics. 3 rd edition. Oxford University Press. (Chap. 5: The Comparative Method & Chap. 11: Problems with the Comparative Method ) Joseph, B.D. & R.D. Janda(eds.) 2003. The Handbook of. Blackwell Publishers. [See Chap. 1: The Comparative Method (R.L. Rankin) ; Chap. 2: On the limits of the Comparative Method (S.P. Harrison); Chap. 4: How to show that languages are related:methods for distant genetic relationship (L.Campbell)] Dixon, R.M.W. 2002. The Rise and Fall of Languages. CUP. Bowern, C. 2006. Another look at Australia as a linguistic area. In Y. Matras, A. McMahon & N. Vincent (eds.) Linguistic Areas: Convergence in Historical and Typological Perspective. Palgrave Macmillan. (Downloadable from course website.) See the course website for further (and more advanced) references. University of Cambridge 6