Magyar Fonetikai Füzetek 24. - Tamás Szende: Phonological Representation and Lenition Processes (1992)
2. Systematic phonological representation - 2.2. Phonological representation in a functional perspective
the root is used as a family name (cf. Szende 1976b). Apparently, certain vowel harmony phenomena can also be interpreted in these terms. To introduce this digression, recall that a constant neuralgic point of the relevant literature, quoted earlier, is the issue of 'exceptions'. One set of these used to contain high back unrounded vowels (ijr 'write', Csík 'geographical name' [homonymous with csík 'stripe'], etc.). Their back-harmonic behaviour can therefore be historically explained although such an explanation is only acceptable as specifying the origin of the phenomenon. As a synchronic rule of phonology, however, it is not applicable — or only at the price of a 'false step'. Originally front-harmonic roots that have subsequently turned backharmonic (e.g. sir 'grave') remain unexplainable even on historical grounds. The back-harmonic behaviour of words like griindol 'found (a firm)' from German gründen 'id.' is also a puzzle in this perspective. Further items whose harmonic vacillation can be said to be irregular, like fotel 'armchair' or farmer 'blue jeans', and which are claimed by Kontra and Ringen to prefer back vs. front-harmonic suffixation depending on the harmonic setting they occur in (cf. Kontra—Ringen 1987, Ringen—Kontra 1989), resist all attempts at a phonological explanation. On the analogy of proper name/common noun homonyms (cf. sík 'flat', back-harmonic vs. Sík 'family name', front harmonic) it is possible that such types of exceptions can be explained by the 'quotation mark effect' : language users will remove such roots from their lexically natural categories and signal their specific origin or meaning by disharmonie suffixation. Vacillating stems would occupy an in-between position in this respect. Such rationalization intended to go beyond mere phonological facts could only be taken seriously, however, if an undoubtedly 'foreign' nature could be proved for all such items, or else that there is some sharp meaning component to set them apart (for sir 'grave', this would be the sacral character of its reference to death). On this point, it is advisable to return to firm phonological ground. The phonological property corresponding to highlighting is 'prominence' — a term that collectively refers to marked values of suprasegmental factors in the text (cf. Lehiste 1970, 2; Hyman 1975, 203). Prominence peaks of a word form cannot be predicted by exclusively grammatical or, in general, by communicative factors. Their primary task, however, is exactly to implement such properties (cf. the role of Nuclear Stress Rule in English com-