1Syllable Weight*weight assume that weightemploy a uniform weight crit - PDF document

1Syllable Weight*weight assume that weightemploy a uniform weight criterion. Anextensive survey of weight-sensitive phonological phenomena shows the very opposite, however:weight criteria are within a given language, but there is weight because of thethatare in turn grounded in other such as the inventory of codaconsonants. classes of syllables, even if this means notweight criteria, governing The 2sets of constraints. Within individual phenomena, constraints referring to the phonetically moston the firstweight-sensitive. For example, in many tonelanguages, syllables differ in terms of the range of tonal contrasts which they may support. Thus,areKeyser 3models is that they assume representations which are projected from properties of the underlyingb. Skeletal slot /taÉ/ /tat/ /ta/ /taÉ/ /tat/ /ta/ t aÉ s m m t a t s m m t a s m R t a XX s C t X O N R t a X N s X O R t aÉ XX N s X O 4to arise from a prohibition against associations between more than one tone and a single timingon which to be languages with Forexample, the Kiowa restriction against contour tones on CV[-son] and CV follows if one assumes but are uniform for within the same 1985,Prince 1986, 1995, Zec 1988, Hayes an heavy and CV as that weight is a property ofothersshows that the stress system, the system of and the 5stringent, however, at both stages. In Early CV[+son] are heavy, while in CV[+son] � CV. Crowhurst the Lhasa is also of the basic weight as a than a process-drivenseems worthwhile to explore Under this view,weight-basedrather than to 6between addressing how and why weight between patterns languagesobserves a hierarchy, where 7���(1) CVV CV[+son] CV[-son] CVa given phenomenon, languages draw points between heavy and 1 discussion; are ToneMetricsMinimal 21017 CVC heavy31680 CV[+son]heavy300 8metric systems. Stress systems are almost equally split between the CVV heavy and the CVV, languages with the those with the CVV, CV[+son] heavy criteria. means of a chi-squared test, which assesses pairs ofcriteria. In summary, comparison of the weight are those differing cross-weight criteria, since any weight 9metrics, and a given criterion, a finding which stronglyis the comparison of stress andfor analysis of the stress and to observe criteria). 10the frequent the CVV,CV[+son] tone; see Housedefines a voicing contrast is the segments lack aor hence its tonalprofile) is cued not only by the but also by harmonics, which occur atfrequencies which are multiples of the fundamental. Thus, a signal with a fundamental frequency800Hz, at 200Hz those of 11 contrast to provide no cues to voiced obstruents, harmonics above the very to supporting tonal information.2 A further factor the The types to carry tone can be made more vivid byconsidering a narrowband of voiced segments in Figure 2. a m z harmonics and harmonics, though the sonorant 12 tonal types offers anexplanation for the tone earlier.which may bear contour tones: will focus on provides an I will argue that, unlike tone, isalso suggests that weight is guided not only by 13rimes, CVC, as heavy syllablesin which long voweled syllables and those vowels followed by is one proposal that is fully with my survey a starting point in the discussion of complexity, it is useful to consider in 2the number of languages 43 heavy4 5Yimas Non-high V heavy3Komi JazÕva Short-central V light15 141964) and Chickasaw 3 own weight unit in virtue of their4 CVV heavy b. CVV, CVC heavy c. Non-high V heavy Heavy= Heavy= X] R +syllabic [X X] R R [X ] +syllabic -high d. CVV, CV[+son] heavy e. Low V heavy f. Heavy= X] R +sonorant R [X ] +syllabic +low R [X ] +syllabic the set of heavy syllables in a serve toMost of the other Figure 3 are straightforward with the possible 15I thus offer the following a working hypothesis: A weightthe rime to be both [+syllabic].5 The6be made for other will argue that languages choose their weightorder of among the weight 16weight of the typedivisionlearn than on Furthermore, oneasier to deploy, since they phonetic Phonetic and (or conversely, lack ofdistinctness) have been argued to play an important role in phonology in such diverse areas as thesegment which are astrive to preserve or or segments(Flemming 1995).were chosen which cross- 17ÉiuaARimes /iwere not measured in Khalkha due to confounds the voweluset for measurement, Short vowels werea set of codaobstruents (if the 18 the present study, duration and energy were forinvestigation because they are closely linked to the realization of stress in many languages. It is a and for 19 the each segment in the waswas to test all the A 55 weightinFinnish are voiceless. The tested weight based on several the and other syllables was tested forChickasaw. 20/, were mostbe the most distinctions. Because they are notnumber of tokens, means were used to xyxy.for each weight in and Lower WilkesÕ 21in languages with this weight Gordon 1999, 2001a forand the to theare For example, the first two the column of/ heavy) and 22only the are superior to at least one of the simple the are listed in order of SCOMPLEX W-lp-val.W-lp-val. { /aÉ/ heavy/aÉ,uÉ/ heavy100100.657425.657425.0000.0000} { /,iÉ/ heavy80.6.603375.0000} VV, a[+son] heavy73.3.581391.0000 VV, a.612845.0000 V, hiV[+dor] light71.6.796293.0000 67.8.799441.0000 VV, V[+son] heavy64.8.661233.0000 VV, V[+voi] heavy56.3.760122.0000 VV, V[+cont] heavy55.9.747150.0000 .934154.0006 +low V heavy17.7.974586.0351 As can be seen, the two the simple 23The Chickasaw data also provides the complexity. If not play a role in the phonology of weight, one wouldÉ SCOMPLEX W-lp-val.W-lp-val. VV, a.634865.0000 VV, a[+lab] heavy99.1.707726.0000 48.1.948532.0047 VV, V[+son] heavy43.9.878960.0000 VV, V[+voi] heavy38.8.905707.0001 VV, V[+cont] heavy13.9.988034.1769 .990921.2398 +low V heavy11.9.991083.2441 VV, V[-son] heavy2.9.999471.7770 other rimes is theheavy and CVV, a[+lab] heavy; however, both of these 24 SCOMPLEX HeavyDiffW-lp-val.HeavyDiffW-lp-val. { V, V[+dor] light100.431361.431361.0000.0000 } light99.5.603583.0000 VV, V[+son] heavyVV, V[+voi] heavy62.662.6.554805.554805.0000.0000 VV, V[+cont] heavy57.2.628306.0000 VV heavy53.8.835684.0000 .735541.0000 .985190.1161 +low V heavy1.1.999879.8872 in weight than stress turn out to be 25sonorous duration, while weight in weight may be a their inventory of codaconsonants) influences phonetic properties, which in turn influence the choice of weight criterion.One be considered in section this 26in turn lead towhich number of obstruent codas the number of sonorant codas. The reasons for example, if one assumes that all codas areweighted equivalently whether they occur in 10 words or 100 words. Thus, what is claimed to beinventory of coda consonants, which arejjp, t, ts, tS, kjb, g] 27mn, N, r, [7segments along these dimensions are energy.Sonorants characteristically have greater energy than obstruents and voiced sounds typically haveexception, sonorancy and voicing are two of the best, if not the best, features forare than in Finnish, since Khalkha has both a and a mr, lgmr, voiced, (m, r, lstvoiced voiceless consonants and in the type frequency of sonorants obstruents. 28 Khalkha Finnish 100 200 300 0 As predicted, CVC is closer in energy to CV than CVV in Khalkha, whereas CVC is closer to 29that weight is large part by coda inventory. The only chi-c2 As far as the present research isas an weight criteria, as the proposed which is in terms of the 30can beincorporated into a formal theory of weight. The discussion here will focus on stress (see Gordona the formal analysis of weight obtains for other with the dimensions along which term calls contrasts in the the tests simple weight the and only8 31 (4) SRlanguage, that rank the SR all the other stressthe basis of The constraints on stress are otherthan one ranked highly enough in the[XX]R and the constraintthe next section, a sample analysis the STRESS 32(5) Yana stressa. sibu!mk'ai ÔsandstoneÕb. ÔHat Creek IndiansÕ,tsiniya!É ÔnoÕc. p'u!diwi ÔwomenÕall CVC syllables be stressed is highly ranked in Yana. A secondNE STRESSfirstamember of the ALIGNsLIGN (s!, A (s!, (s!, R, PrWd), which requires that(8) ONE STRESS�� SR�� ALIGN (s!, L, PrWd) (s!, 33(9)Inputsuk'oÉniyaÉONESTRESSSTRESS[XX]RALIGN (s!,L, PrWd) + a) suk'o!ÉniyaÉ** b) suk'oÉni!yaÉ**! ** c) suk'oÉniyaÉ*! ** d) suk'o!Éniya!É*! **** e) suk'oÉniya!É****! f) su!k'oÉniyaÉ**! AsSTRESS. STRESS Input:sibumk'aiONESTRESSSTRESS[XX]RALIGN (s!,L, PrWd) + a) sibu!mk'ai** b) sibumk'a!i***! c) si!bumk'ai**! (s!, L,(s!, 34(11)Input:p'udiwiONESTRESSSTRESS[XX]RALIGN (s!,L, PrWd) + a) p'u!diwi b) p'udi!wi*! ALIGNss/.(12)Input:p'udiwiALIGN (s!,L, PrWd)ALIGN (s!, +p'u!diwi ** b) p'udiwiÚ*! the stress onvowel, otherwise the If ONE Sranked below the presented tone and stress systems showing that 35and differ in their theyweight not by which filters out more criteria. This aare Finally, thethe 36& Cohn primary andsecondary Baitschura, Uzbek Untersuchung der , Beckman, Mary (1986). [ University.Susan Chen & phonology and phonetics. Phonology 37Phonology Foley, William A. (1991). The Yimas language of New Guinea. Stanford: Stanford weight: phonetics, typology. PhD.appear). A typology of contour tone restrictions. Studies inLanguage. Extensions and Chicago Press. the 1996 Milwaukee Linguistics 38 9Roman Jakobson, Selected Writings I, 117-136. The Hague: Mouton. I Kenstowicz, Michael (1994). Sonority-driven stress of human Phonology, Stanford University Press.McCarthy, John in Semitic phonology and morphology. dissertation, MIT.Rutgers University. Rutgers ROA-7,http://ruccs.rutgers.edu/roa.html.]McCarthy, John & morphology. In John Goldsmith (ed.) 39Moore, Brian (1995). In Peter MacNeilage 13 40 4 ,A grammar of Kiowa. Lincoln: University of Nebraska Press. . PhD dissertation, MIT.papers from UCLAprosodic PhD dissertation, StanfordUniversity [Published by Garland 1994]. 41NOTES *1CV is actually phonetically the phonemic long vowel in CVV[-son] is quite short2there is the three languages in the survey (Hausa, Musey, and 3 4central vowels should be languages in which are 567h 8