The Sonority Hierarchy in Hungarian (original) (raw)

The classical "mirror rule" of traditional grammars subsumes three, logically independent observations: (1) If PQ is a possible syllable onset (P, Q arbitrary consonants), then QP is not. (2) If PQ is a possible onset, then QP is a possible coda, and conversely, if RS is a possible coda, then SR is a possible onset. (3) If PQ is a possible coda, then QP is not. Of course, if (2) holds, (1) and (3) are equivalent-but there might well be languages where (2) turns out to be false, but the other two statements are true. In fact, every language where consonant clusters are disallowed as codas but permitted as onsets is a counterexample to (2), and the same holds for those languages that allow complex codas but do not allow complex onsets. Before turning to the investigation of the mirror rule in Hungarian, let me add a further clause, (cf. Clements-Keyser 1983:47-48) which I will call Hjelmslev's Law: (4) If PQR is a possible onset, then so are PQ and QR, and similarly for codas. (5) If PQ and QR are possible onsets, then so is PQR, and similarly for codas. This last requirement (the converse of Hjelmslev's Law) and (4) have the effect of extending (1) and (3) to arbitrarily long consonant clusters: in Hungarian, the longest cluster that we will encounter contains three consonants. If the notion "Sonority Hierarchy" (in the sense of Jespersen 1897-99) has any validity, then the statements (1-5) will follow automatically. Suppose that phonemes are 0 I would like to thank Lászlo Kálmán,Ádám Nádasdy, and Péter Siptár for their kind help with the manuscript. The 1985 lectures of Nick Clements on syllable structure at the Salzburg International Summer School have had a decisive influence on my treatment of this material. It is a pleasure to acknowledge my indebtedness.