A typological study of reduplication as a morpho-semantic process: Evidence from five language families (Bantu, Australian, Papuan, Austroasiatic and Malayo-Polynesian)

This thesis involves a systematic study of reduplication across five language families (i.e. Bantu, Australian, Papuan, Austroasiatic, and Malayo-Polynesian). It deals with 89 languages regarding parts of speech, semantics, category change, phonological patterns, and the form-function relation. It is only concerned with reduplication which has its unreduplicated form, or a stem. Also four major parts of speech (i.e. nouns, adjectives, adverbs, and verbs) are treated for the investigation.;A set of common characteristics which are shared among all of the five language families as well as characteristics particular to a specific language family have been found. It is demonstrated that no candidates for universal (or necessary) properties for reduplication are found except for the near-universal property: if a language has reduplication it should include verb reduplication. As for non-universal (or possible) properties of reduplication, a number of candidates are found. With regards to the semantic properties, the most frequent meanings across language families are plural and totality for nouns, intensity for adjectives and adverbs, repetition/continuation and intensity for verbs. As far as category change goes, nouns, adjectives, verbs can change into four major parts of speech while adverbs cannot. Concerning phonology, both total and partial reduplication are identified as the basic processes. Also, it is noted that partial reduplication tends to be prefixed and that the basic phonological modification is vowel changing. As for the form-function relation, there is a tendency that category change is given priority to be expressed in a totally reduplicated form if the language in question has both total and partial reduplication.;On the basis of generalizations about the sample, typological predictions on reduplication can be made in two ways: general predictions and family-specific predictions. These predictions are tested on six languages which are not included in the original sample and it has been observed that the results of the tests are consistent with all of the predictions. This gives support for the findings on reduplication.