Nonconcatenative morphology

Nonconcatenative morphology is a term which is used for non-agglutinative root-and-pattern morphologies. In such systems, word formation processes generally do not take the form of linear affixation.

Example
in Arabic words are commonly formed on the basis of a triliteral root, i.e. a set of three consonants, between which sets of vowels are intercalated. This set of vowels sometimes itself signifies a grammatical category such as 'perfective active'. Furthermore, the structure of words within a single class (or binyan) is identical, in the sense that they have the same prosodic template. Compare the following stems which are formed on the basis of the root ktb 'write', and a particular set of vowels:

binyan	Pref.Act. Perf.Pass. Impf.Act. Tmpf.Pass I	katab		kutib		aktub		uktab II	kattab		kuttib		u-kattib	u-kattab III	kaatab		kuutib		u-kaatib	u-kaatab IV	?-aktab		?-uktib		u-?-aktib	u-?-aktab V	ta-kattab	tu-kuttib	a-ta-kattab	u-ta-kattab

McCarthy (1979, 1981) has convincingly shown that nonconcatenative systems can best be accounted for within the framework of Autosegmental Phonology/Morphology. Words such as katab and kutib consist of three morphemes which are represented at autonomous levels or tiers, viz. the root ktb, the perfective active morpheme a and the perfective passive morpheme u-i, respectively, and the binyan I template CVCVC. These words differ from kattab and kuttib in that the latter take the binyan II template CVCCVC.

Links
Utrecht Lexicon of Linguistics