UC Berkeley Phonology Lab Annual Report 2007 PHONETIC vs PHONOLOGICAL LENGTHENING IN AFFRICATES Anne Pycha University of California Berkeley pycha berkeley edu ABSTRACT root is suffixed as in vason va5on ironSUPERESSIVE We can also compare the duration of 5 when it is singleton versus when it is geminate In Hungarian the instrumental case suffix al triggers gemination when it attaches to a noun root baj trouble bajjal trouble INSTR vas iron vassal iron INSTR etc To get an idea of how phonological lengthening operates then we can compare the duration of 5 when it is singleton with plain suffix as in vason va5on with its duration when it is geminate with geminating suffix as in vassal va5 al We might reasonably expect phonetic lengthening to be smaller and more variable than phonological lengthening which would suggest that the two processes lie along a continuum of sorts If we introduce affricates into the picture however we have the potential to uncover deeper differences This is because affricates consist of two portions stop closure and frication While there is disagreement as to how these portions are best represented underlyingly 2 all theories agree that the structure of affricates is more complex than that of simple segments The null hypothesis is that phonetic and phonological lengthening treat the complex structure of the affricate in the same manner for example by lengthening both portions to a certain degree If on the contrary phonetic lengthening specifically targets one portion of the affricate while phonological lengthening targets another this suggests that the former is not merely a gradient version of the latter To pursue this question the present study analyzes phonetic and phonological duration changes in stops fricatives and affricates in Hungarian Affricate consonants consist of two portions stop closure and frication Can these portions play different roles in phonetic and phonological processes In this study I address the question by probing the behavior of Hungarian affricates under lengthening I measure the duration changes that affricates undergo in two types of lengthening processes first a phonetic process of final lengthening and second a phonological process of gemination I show that these two processes alter the internal structure of affricates in very different ways The results suggest that the difference between phonetic and phonological processes is in fact deeper than a mere difference between gradient and categorical effects Keywords affricates geminates Hungarian 1 INTRODUCTION Consonants can lengthen as the result of their position in a syllable word or phrase This finding has been demonstrated for a number of different languages 1 For example consonants at the beginning or end of a word tend to be longer than those which are medial Such processes are typically considered to be gradient and phonetic because they do not alter the phonemic status of the consonant Consonants can also lengthen as the result of morpho phonological conditioning For example a suffix that triggers gemination will lengthen the final consonant of the root to which it attaches Such a process is typically considered categorical and phonological because the phonemic status of the consonant changes from singleton to geminate Both lengthening processes affect the edges of roots or words which means that we can study them in tandem Consider the Hungarian root vas va5 iron which may occur in isolation or with suffixes To get an idea of how phonetic lengthening operates we can compare the duration of 5 when it is word final when the root is bare with its duration when it is word medial when the 2 METHODS Stimuli from 3 were constructed using a set of monosyllabic and disyllabic Hungarian noun roots ending in one of two affricates ts t5 or one of 269 UC Berkeley Phonology Lab Annual Report 2007 three simple consonants t s 5 The root final consonants occurred in four different syllable types CVC CVNC CV C and CV NC giving rise to a total of eight root shapes The vowel of the final syllable was always a just before the release burst Fricatives and fricative portions of affricates began at the onset of aperiodic energy and ended at the cessation of aperiodic energy Breathy offset at the end of words was not included in segment durations We will be concerned with two comparisons To examine final lengthening effects we will compare the target consonants in word medial versus word final environments kat5on versus kat5 To examine gemination effects we will compare the target consonants in singleton and geminate environments kat5on versus katt5al Table 1 Root shapes used in stimulus construction CV CVC CV CVNC CV C V C CV CV NC Monosyllable kat5 mant5 a t5 ga nt5 Disyllable pamat5 parant5 tana t5 zoma nt5 3 RESULTS Note that vowel length differences do not correlate with consonant length differences in Hungarian Stress is always on the first syllable The root list was initially designed to contain 5 root final consonant types x 8 word shapes x 2 repetitions of each shape 80 roots Because of gaps in the lexicon of Hungarian for example there is no monosyllabic noun root with the shape Ca n5 the final number of roots was 63 To construct the stimuli each noun root was placed in three environments bare unsuffixed suffixed with a plain case ending Superessive on or suffixed with a geminating case ending Instrumental al 3 1 Overall results suggest that final lengthening targets the fricative portion of an affricate Figure 1 shows mean durations for stops fricatives and affricates in word medial and final positions In this and subsequent graphs T represents stops or stop portions of affricates S represents fricatives or fricative portions of affricates s and 5 are pooled together and TS represents affricates ts and t5 are pooled together Paired t tests p 0 05 indicate that the fricatives lengthen significantly in the word final environment mean increase 39 3 ms as do the affricates 36 2 ms Stops however do not Table 2 Three environments for roots Bare t5at va5 kat5 Plain suffix on t5aton va5on kat5on Final lengthening Geminating suffix al t5attal va55al katt5al Figure 1 Mean duration of stops T fricatives S and affricates TS in medial m and final f environments in milliseconds 250 Each word 63 roots x 3 environments 189 was embedded in a quoted phrase within a carrier sentence Marika azt mondta hogy X gyorsan Marika said X quickly For bare roots this context should trigger both phrase final