Why do we compare? (1) Goals and methods of science • to advance our understanding of the world • answering " what? " (observation) is not sufficient, we also need to ask " how? " and " why? " (explanation) • the hallmark of the scientific method is the artificial experiment , but such experiments are often not possible or ethical in many fields, especially those who study the past and unique events: evolutionary biology, archaeology, but also epidemiology, ecology, sociology, etc. • one alternative to the controlled experiment is the " natural experiment " , or " comparative method " : manipulation is replaced by comparison with other attested cases • comparison has a long history and has been a standard method of enquiry in anatomy for centuries We consider that it is worth while to examine many animals of different species because what is often more condensed or more concealed in one species Nature displays more clearly and openly in another. (Regnier de Graaf 1663, Disputatio medica de natura et usu succi pancreatici, cited in Cole 1944: 12) (2) In linguistics too, comparison is an invaluable tool 1. typological comparison: search for the universal laws, tendencies, and possible variations of language (see the presentation by M. Shimoji) 2. historical comparison: explanation of languages by reference to their evolutionary changes, as revealed by comparison with related languages 1 The historical comparative method in linguistics (3) Motivation and method • usually, we cannot observe ling. changes, only their results • there are usually few historical records, so that we have to rely on comparison with related languages, which constitute as many pieces of the original puzzle • we infer (reconstruct) an earlier state (proto-language) from which all languages can be plausibly derived • regularity of sound change and arbitrariness of form-meaning associations allow to rule out coincidences (4) Alternatives to the traditional comparative method? • mass comparison, lexical diffusion, allofams, typology-based classifications, etc., have not and will never produce any reliable etymology nor explain any irregular inflection • i.e. they do not solve interesting problems