Postalveolar assimilation and its exceptions from the point of view of Hungarian language history

The author examines the different types of postalveolar assimilation in the different imperative forms of Hungarian verbs ending in -t (eg. szeret -> szeress, tanít -> taníts, pronounced [tanícss]). Generative phonology has presented several proposed solutions for these phenomena in the last decade. It has become understood that a long vowel before a final -t becomes a short vowel when the consonant is followed by a -j. In CV phonology and X-level phonology, these consonants became represented – somewhat abstractly – as C or X units. The author tries to prove in his article that these results actually are in complete agreement with explanations from earlier language history, and an etymology of these words (and hence their actual morphological structure) explains these results. Therefore it is unfortunate that the modern direction of the field in large part hardly considers the results of language history: it would be desirable for both sides to exhibit more affinity for the results of the other, and thus enriching their research. 1. The question of postalveolar assimilation in the imperative form of the verb with word final -t has long occupied many Hungarian linguists. It is well known that the two types of variations that occur here can be distinguished: in one version the final -t and the imperative -j become /s:/, while in the other -t + -j yields /cs:/. Both types include two groups of verbs. (1) Verbs with s-assimilation (a) All verbs with word-final -t, in which the final -t is preceded by a short vowel. For example, szeret + j -> szeress ‘love!’; köss ’tie!’ (b) The verbs lát ‘see’, bocsát ‘forgive’, and lót(-fut) ’run about’: lát + j -> láss; bocsáss; lóss(-fuss) (2) Verbs with cs-assimilation (a) All verbs with word-final -t in which a consonant appears before the -t (the consonant can only be a sonorant: l, r, j, m, n). For example bont + j -> bonts ‘break!’; márts ‘dip!’; tölts ‘fill!’; onts ‘shed!’ (b) All verbs with the derivational suffix -ít, as well as the verbs fűt ‘heat’, hűt ‘cool’, (dűt) ‘topple’, műt ‘operate’, szít ‘instigate’, tát ‘open (mouth)’, and vét ‘fault’. For example tanít + j -> taníts; fűts, véts. The different assimilation processes of the groups of the type in (1) are easily explained on the basis of the different phonological makeup of the two groups, and if anything problems illustrate the different assimilation features of the group in (2). Based on their differing behaviors during assimilation, they constitute two explanatory attempts. According to Lotz (1960; 1976, 173 in Hungarian), Deme (1961, 103), and Papp (1966, 143), as well as Abondolo (1988, 146), cs-assimilation is a rule-governed process and s-assimilation is not rule-governed. According to this view, stems with a final syllable of two mora in the coda (i.e. a short vowel stands before the -t) participate in this assimilation, while stems with more than two mora in the final syllable (i.e. a long vowel or consonant stands before the -t) take part in csassimilation. Lately, generative phonology has suggested several solutions to clarify the phenomenon, cf. Szépe (1969), Vago (1980; 1987; 1991), Siptár (1990; 1994; 1995), Olsson (1992) and Zsigri (1997). In this research what is critical is whether the segment before the -t is a vowel or a consonant, and as such for them the forms like láss are classified as regular forms together with the verbs ending with a short vowel. I quote here a part of the argumentation from Vago: “A hagyományos felfogás rövid és hosszú magánhangzó közötti megkülönböztetése nem egészen kielégítő: A tényeket ugyan leírja, de magyarázatot nem tud hozni arra, hogy (pár kivétel) a mássalhangzók és a hosszú magánhangzók egyformán viselkednek a t-végű igék felszólító módjában. Ezzel szemben nyilvánvaló, hogy ha a hosszú magánhangzókat legalább is részben mássalhangzónak elemezzük egy mélyebb fonológiai szinten, akkor a márt és a fűt féle igék egységes viselkedése a fut felékhez viszonyítva rögtön kiugrik. De ez csak akkor lehetséges, ha a fonológiai elmélet több leíró szintet ismer el. Pontosan ezzel az érvvel, többek között természetesen, járult hozzá a hatvanas és hetvenes évek generatív teóriája a fonológiai tudomány előrehaladásához. [The traditional approach’s distinction between short and long vowels is not entirely satisfactory: the facts are described in such a way without an explanation being provided, that (with a few exceptions) consonants and long vowels behave identically in imperative verbs ending in -t. On the other hand it is clear that if we analyze long vowels at least in part as consonants on a deeper phonological level, then the uniform behavior of the márt and fut type of verbs in relation to the fűt type of verbs jumps out relatively immediately. However that is only possible if the theory of the phonology admits additional levels of description. It is exactly for this reason, amongst others naturally, that in the 60’s and 70’s generative theory made its ascent to the forefront in the discipline of phonology.]" (Vago 1991, 683). Vago also presents different solution attempts, which generative phonology believes solve the question of the postalveolar assimilation. He presents three different solutions, which chronologically follow one another. 1.1. Generative phonology Generative phonology mistakes the long vowel before -t as a short vowel-consonant sequence. Thus the same rule applies in the imperative of the verbs márt and fut: if a consonant precedes -t, then the -t and the imperative symbol -j become written -ts, spoken -cs, and in the end the consonant before the t disappears and the preceding vowel lengthens [Konsonanten in text is an error]. Schematically: (3) füCt + j (C = consonant) füCts fűts (cf. Vago 1991, 683; also Szépe 1969, 458-9) Vago (as cited) next examines which consonant is behind this "phantom consonant", which disappears in the above-mentioned case and exerts an effect on the preceding vowel. Since this consonant cannot break any phonotactic rule, he examines material the Reverse-Alphabetized Dictionary of Hungarian (Papp 1969) and states that (with exception of s and sp) at the end of the stem or derivational suffixes and before the -t only j, l ,m ,n and r can appear. One can not accept the l, n, r, and m (see e.g. the imperative forms of the tilt, int, irt, teremt, in which the consonant before the -t remains), hence only the -j remains. That is, verbs such as szít and fűt can be described in generative phonology as szijt and füjt, from which it follows that the changes in the imperative of these verbs can be described uniformly, depending on whether a short or long vowel or a consonant is postulated before the -t (see Vago 1991, 684).