ABSTRACT
Up to the 1960s the term "vowel harmony" (VH) was applied to various African languages, with an emphasis on the description of this relatively newly discovered phonological phenomenon. A perennial problem in the morphological parsing of an utterance is that it crucially depends on the results of the phonological parsing of that utterance. One might therefore be tempted to refer to categorial morphological information and claim that once a candidate VH domain and process have been identified, the candidate morpheme types "noun stem" and "noun-class suffix" are also almost automatically identified. Some languages have more than one VH process, sometimes even in opposite directions. A further distinction seems to be relevant to the typology of VH processes: whether the target vowel would be pronounceable without the harmonizing element. Some languages have processes that seem to run contrary to VH, and some languages have partially self-contradictory VH processes.
