Nyelvtudományi Közlemények 91. kötet (1990)

Tanulmányok - Kornai, András: The sonority hierarchy in Hungarian 139

144 ANDRÁS KORNAI 'Serbian', gróf 'peer', burg 'castle in Vienna', drót 'wire', kard 'sword', tárgy 'object', törzs 'tribe', zri 'rumpus', borz 'badger', szárny 'wing'; Majs, fájsz 'hurt 2nd.sg.pres.indef, fjojxl 'id.' dölyf 'haughtiness', selyp 'lisp', sztrájk 'strike', Svájc 'Switzerland', sejt 'cell', ejts |ejc| 'drop 2nd.sg.imp.indef, ölyv 'buzzard', cajg 'calico', majd 'then', pajzs 'shield', rajz 'drawing'; snájdig 'neat', pikáns 'piquant', sznob 'snob', pasziánsz 'solitaire', knockout |knokaut| L\á.\ fánk 'id.', fánk 'doughnut', tánc 'dance', pont 'dot', kilincs 'doorknob', rokonszenv 'sympathy', gnóm 'gnome', rang 'rank', gond 'worry', pénz 'money'; sláger 'hit', fals 'out of tune', szláv 'Slavic', félsz 'fright', flaska 'bottle', golf 'id.', plakát 'poster', talp 'sole', klassz 'groovy', halk 'silent', pole 'shelf, bolt 'shop', kulcs 'key, vlach 'id.', nyelv 'tongue', blúz 'shirt', glória 'halo', rivalg 'whoop', föld 'earth', völgy 'valley'; smaragd 'emerald', szmötyi 'sediment', tömsz 'stuff 2nd.sg.pres.indef, tromf 'retort, kolomp 'bell', teremt 'create', teremts |mc| 'create 2nd.sg.imp.indef, hamv 'ash', domb 'hill', nyomd 'push' 2nd.sg.imp.def, tornzs 'lode', nemz 'beget'; dzeta 'id.', gerezd 'clove', küzdj |zj| 'fight 2nd.sg.imp.indef; idősb |z6| 'elder', Pünkösd |zd| 'whitsun', esdj |zj| 'beg' 2nd.sg.imp.indef; kedv 'mood'; yacht 'id.'; sztyepp 'prairie', hagysz |cs| 'let 2nd.sg.pres.indef, Batyk; steril 'sterile', Detk, barack 'peach '; skála 'scale', voks 'vote'; spicli 'informer', taps 'clap', szpáhi, 'spahi', gipsz 'gypsum', copf 'pigtail', Ape; szféra 'shpere', szívsz |fs| 'suck 2nd.sg.pres.indef. The point of the exercise is that only a few gross statements about the sonority hierarchy appear to be incontrovertible. Voiceless consonants will precede the voiced ones, and obstruents will precede the resonants in every reasonable rearrangement of (6), but besides these (rather trivial) observations, little can be said with certainty. With that, the question becomes the following: what can we possibly gain by employing a theoretical construct (the sonority hierarchy) if, on the one hand, it is next to impossible to model the facts (i.e. to arrange the consonants on a scale) by it in an unambiguous manner, and if, on the other hand, the predictions (i.e. 1-3) made by the theory do not really fit the data? My answer is based on the well-known facts that syllables are psychologically real units of speach production (cf. e.g. Kim 1971) and of speech perception (Savin-Bever 1970). The sonority hierarchy makes it possible to factor out a large part of the linear precedence (LP, see Gazdar-Pullum 1982) information that must be encoded with every syllable node immediately dominating' a number of timing units. In fact, no LP information has to be stored with C*V and VC* syllables conforming to the hierarchy. In syllables containing Cs on both sides of the V it is sufficient to store only the fact that a given consonant precedes or follows the vowel (so that pit will not be confused with tip) — Nyelvtudományi Közlemények 91. 1990.