JEastAsianLinguist(2014)23:71–111 DOI10.1007/s10831-013-9119-x The nominal pitch accent system of South Kyungsang Korean · Hyunjung Lee Jie Zhang Received:13August2011/Accepted:11January2013/Publishedonline:19December2013 ©SpringerScience+BusinessMediaDordrecht2013 Abstract ThecurrentpaperinvestigatesthenominalpitchaccentsysteminSouth KyungsangKoreanthroughanacousticstudyandpresentsaphonologicalanalysisfor thesystembasedontheacousticresults.ThedatawerecollectedfromfourmaleSouth Kyungsang speakers by recording monosyllabic and disyllabic nouns with various typesofsuffixes.Thepitchresultsconfirmedtheaccentdistinctionsreportedinearlier worksonthelanguage,andwefoundthatsuffixesmayalsobearcontrastiveaccent. We analyzed the pitch accent system as having three pre-linked accents and one defaultaccent.Twoofthepre-linkedaccentscountfromtheleftandarelinkedtothe initialandpeninitialmorasoftheroot,respectively,andbothspreadonemoratothe right.Theotherpre-linkedaccentcountsfromtheright(penult)anddoesnotspread. ThisanalysisfitsinwiththeculminativetonetypologyestablishedinEvans(Typesof tonal culminativity in language of Sichuan and elsewhere, 2009). An Optimality Theoreticanalysisthatderivesthesurfacetonepatternsforboththedefaultandpre- linkedaccentsisproposed,andtheproposaliscomparedwithearlieranalysesofpitch accentsinNorthandSouthKyungsangKorean. Keywords South Kyungsang Korean · Pitch accent · Culminativity · Tone spreading · Optimality Theory 1 Introduction The current study examines the nominal pitch accent system in South Kyungsang Korean. Unlike Standard Seoul Korean, which does not distinguish words using H.Lee·J.Zhang DepartmentofLinguistics,UniversityofKansas,1541LilacLane,Lawrece,KS66044,USA H.Lee(&) DepartmentofLinguistics,UniversityofChicago,1010E59thStreet,Chicago,IL60637,USA e-mail:[email protected] 123 72 H.Lee,J.Zhang tonal contrasts, North and South Kyungsang dialects of Korean may use tonal differences in a word to cue meaning differences. We refer to the systems as pitch accent systems as the use of tone in these languages is different from that in a canonical tone language (e.g., Chinese) in that tone is restricted in its distribution (see, e.g., Yip 2002). The North and South Kyungsang dialects share many of the pitch contrasts, as shown in (1). But South Kyungsang preserved the rising pitch accent (R) from Middle Korean (15th century), while North Kyungsang did not. This R accent1 is realized as a rising tone on monosyllabic roots and as LH on disyllabic roots in South Kyungsang whereas R has merged with H(H) in North Kyungsang,asshownin(2).DuetodifferenthistoricaldevelopmentbetweenSouth and North Kyungsang, while North Kyungsang has only one LH pattern, South Kyungsang has two types of LHs that are distinguished morphophonemically, as shown in (3) (Kenstowicz et al. 2008). The Kyungsang data are transcribed according to the Yale Romanization (Martin 1992). (1) South and North Kyungsang: HL ka´c`ı ‘kind (n.)’ HH ka´c´ı ‘branch’ LH ka`c´ı ‘eggplant’ (2) South Kyungsang: North Kyungsang: R nwuˇn H nwu´:n ‘snow’ LH sa`la´m HH sa´:la´m ‘person’ (3) South Kyungsang: North Kyungsang: LH-L pa`la´m-`ı LH-L pa`la´m-`ı ‘wind-nom.’ LH-H sa`la´m-´ı HH-L sa´:la´m-`ı ‘person-nom.’ WefocusonthenominalpitchpatternofSouthKyungsangKoreaninthisstudy,but we also refer to previous studies on North Kyungsang for analytical insights (e.g., KenstowiczandSohn1997;Kim1997;Chang2005;Lee2008).Westartbyreviewing thedescriptivepatternsofpitchaccentinSouthKyungsangandproposingthegoalsof thecurrentstudy.WethenreportanacousticstudyoffourSouthKyungsangspeakers ontheiruseofpitchaccent,followedbyaformalanalysisofthepitchaccentsystem. 1 Someotherstudies(Utsugi2009;LeeandDavis2009)usetheLtonedescriptioninsteadofR.We adopt the R tone description for the following two reasons. First, the R description better reflects its phoneticpropertiesthantheLdescriptionasitbettercaptures(1)thefactthattherimedurationofthis classofwordsislongerthanthatoftheHightoneclassastherisingcontourrequiresalongerdurationto implement(Zhang2002),and(2)thelackofdifferenceintheF0peakvaluesofthisclassfromtheHigh toneclass(Chang2007).Second,theRtonedescriptionbettermatchesthetonesystemofMiddleKorean (fifteenthcentury).ThewordsinthemonosyllabicRanddisyllabicLHclassesusedtobeintheRandRL classesinMiddleKorean,respectively(Ramsey1975;Kenstowiczetal.2008).Therefore,usingtheR tonedescriptionallowsustodirectlyindicatethedifferenthistoricaldevelopment ofthepitchsystem betweentheSouthandNorthKyungsangdialectsofKoreanandthusallowsustobeinabetterpositionto tracethehistoricalchangesoftheKyungsangaccentsystem. 123 PitchaccentofSouthKyungsangKorean 73 Comparisonsofouranalysiswithotherapproaches,directionsforfutureresearch,and otherconcludingremarksareprovidedattheend. 1.1 Descriptions of tonal patterns in South Kyungsang Korean In monosyllabic words, South Kyungsang Korean has pitch accent minimal pairs suchasnwún(H)‘eye’versusnwǔn(R)‘snow’andmál(H)‘horse’versusmǎl(R) ‘speech’(Ramsey1975;KenstowiczandPark2006;Kenstowiczetal.2008;Chang 2007). The behavior of monosyllabic noun roots under suffixation indicates that there are in fact two different types of H accents (Ramsey 1975; Kim and Schuh 2006; Schuh and Kim 2007; Chang 2007). For example, although both nwún ‘eye’ and mún ‘door’ have an H pitch in isolation, when the suffix -i (nominative case marker)isadded,thepitchpatternfor‘eye’isnwún-í(H-H)whilethepitchpattern for ‘door’ is mún-ì (H-L). In addition, when expanded with the disyllabic suffix -mankhum ‘as much as’, the pitch patterns are nwún-mánkhùm (H-HL) and mùn- mánkhùm (L-HL). Finally, when suffixed with -imyen ‘if’, the pitch patterns are nwún-ímyèn(H-HL)andmún-ìmyèn (H-LL). Wetentativelywrite thetwodifferent H accents as H1 and H2. There is a consistent pitch pattern for the R accent under suffixation:thesuffixwilltakeontheHtonewhiletherootsyllabletakesanLtone. Table 1 summarizes the pitch patterns for H1, H2, and R on monosyllabic nouns under different types of suffixation. For disyllabic nouns, we have already seen in (1) that South Kyungsang Korean has three pitch patterns: HL, HH, and LH. But expanding the nouns with suffixes indicates that there are in fact two LH classes. For example, although both sàlám ‘person’ and pàlám ‘wind’ have an LH pitch pattern in isolation, they pattern differently when suffixed: when the suffix -i (nominative case marker) is added, ‘person’ patterns as sàlám-í (LH-H) while ‘wind’ patterns as pàlám-ì (LH-L); when the suffix -imyen ‘if’ is added, ‘person’ patterns as sàlám-ímyèn (LH-HL) Table1 PitchpatternsforH1,H2,andRinmonosyllabicnounsundersuffixation.Datawereadapted fromKimandSchuh(2006)andChang(2007) H1 H2 nwu´n H ‘eye’ mu´n H ‘door’ nwu´n-´ı H-H ‘eye(nom.)’ mu´n-`ı H-L ‘door(nom.)’ nwu´n-ma´nkhu`m H-HL ‘asmuchaseye’ mu`n-ma´nkhu`m L-HL ‘asmuchasdoor’ nwu´n-´ımye`n H-HL ‘ifeye’ mu´n-`ımye`n H-LL ‘ifdoor’ R nwuˇn R ‘snow’ nwu`n-´ı L-H ‘snow(nom.)’ nwu`n-ma´nkhu´m L-HH ‘asmuchassnow’ nwu`n-´ımye´n L-HH ‘ifsnow’ 123 74 H.Lee,J.Zhang while ‘wind’ patterns as pàlám-ìmyèn (LH-LL). We tentatively label the two accents as LH1 and LH2, respectively. The HL and HH accents have consistent behavior under suffixation: the suffixed syllables will take an L tone while the HL and HH tones remain on the root. The pitch patterns for HL, HH, LH1, and LH2 on disyllabic noun roots under different types of suffixation are summarized in Table 2. Wecanobservethefollowingparallelsbetweenthemonosyllabicanddisyllabic pitch patterns. (1) H1 in monosyllables and HH in disyllables behave in parallel in having an H tone on the first and second syllables if there is a second syllable available. (2) R in monosyllables and LH1 in disyllables behave in parallel in having an H tone on the second and third syllables if they are available; otherwise theHdockstotherightedgeofthemonosyllabletocreatearisingtone.(3)Unlike theH1/HHandR/LH1toneclassesthathaveaconsistentpitchpatternregardlessof the suffix, the H2 and LH2 classes both have different pitch patterns depending on the suffix. When the words in these tone classes are suffixed with -mankhum ‘as muchas’,theHtonealwaysoccursonthefirstsyllableofthesuffix;butwhenthey aresuffixedwith-i(nom.)or-imyen‘if’,theHtoneoccursonthelastsyllableofthe root,andtherestofthesyllablestakeanLtone.Theseparallelsaresummarizedin Table 3. They seem to indicate that, descriptively, there are four different pitch patterns in South Kyungsang Korean nouns: HL, which only surfaces on disyllabic roots,H1/HH,R/LH1,andH2/LH2.AnadditionalreasonfortheparallelbetweenR inmonosyllablesandLH1indisyllablesisthattheLH1classwasalsoderivedfrom therisingpitchaccentinMiddleKorean(Ramsey1975;KenstowiczandPark2006; Kenstowiczetal.2008).Theaccentpatternsoftrisyllabicandquadrisyllabicnouns will be discussed in Sect. 5 after our formal analysis for the nominal pitch accent pattern is proposed. Table 2 Pitch patterns for HL, HH, LH1 and LH2 in disyllabic nouns under suffixation. Data were adaptedfromKimandSchuh(2006)andChang(2007) HL HH ke´wu`l HL ‘mirror’ ku´nu´l HH ‘shade’ ke´wu`l-`ı HL-L ‘mirror(nom.)’ ku´nu´l-`ı HH-L ‘shade(nom.)’ ke´wu`l-ma`nkhu`m HL-LL ‘asmuchasmirror’ ku´nu´l-ma`nkhu`m HH-LL ‘asmuchasshade’ ke´wu`l-`ımye`n HL-LL ‘ifmirror’ ku´nu´l-`ımye`n HH-LL ‘ifshade’ LH1 LH2 sa`la´m LH ‘person’ pa`la´m LH ‘wind’ sa`la´m-´ı LH-H ‘person(nom.)’ pa`la´m-`ı LH-L ‘wind(nom.)’ sa`la´m-ma´nkhu`m LH-HL ‘asmuchasperson’ pa`la´m-ma´nkhu`m LH-HL ‘asmuchaswind’ sa`la´m-´ımye`n LH-HL ‘ifperson’ pa`la´m-`ımye`n LH-LL ‘ifwind’ 123 PitchaccentofSouthKyungsangKorean 75 Table3 Parallelsinpitchpatternsbetweenmonosyllabicanddisyllabicnouns Monosyllables H1 R H2 Isolation H R H Mono_suffix:-i H-H L-H H-L Di_suffix:-mankhum H-HL L-HH L-HL Di_suffix:-imyen H-HL L-HH H-LL Disyllables HL HH LH1 LH2 Isolation HL HH LH LH Mono_suffix:-i HL-L HH-L LH-H LH-L Di_suffix:-mankhum HL-LL HH-LL LH-HL LH-HL Di_suffix:-imyen HL-LL HH-LL LH-HL LH-LL 1.2 Problems posed by suffixes for H2/LH2 As noted above, the pitch pattern for words in the H2 and LH2 classes varies dependingonthetypeofsuffixaddedtotheroot:disyllabicsuffixes-mankhumand -imyen elicit different pitch patterns on these nouns. Based on her preliminary phonetic examination, Chang (2007) made the generalization that the tone pattern difference is determined by whether the disyllabic suffix is vowel-initial or consonant-initial.Specifically,whenamonosyllabicwordintheH2class(analyzed as an M-class by Chang) is followed by a monosyllabic or vowel-initial disyllabic suffixsuchas-i(nom.),-to‘also’,-un(topic),or-imyen‘if’,theHtoneisplacedonly in the root. However, when a word in this class is followed by a consonant-initial suffixsuchas-pota‘than’,theHtoneoccursonthefirstsyllableofthesuffixbutnot ontheroot,andtherestofthesyllablestakeLtones:mún-ìmyèn(H-LL)versusmùn- pótà(L-HL).Likewise,whenawordinthedisyllabicLH2class(analyzedasanLM- classbyChang)isfollowedbyamonosyllabicorvowel-initialdisyllabicsuffix,theH toneoccursonlyintherootbutnotinthesuffix.Butwhenthewordisfollowedbya consonant-initialdisyllabicsuffix,theHtoneoccursinthesuffix:pàlám-ìmyèn(LH- LL) versus pàlám-pótà (LH-HL). This suffix-determined tonal pattern led Chang (2007)tocontrolthesuffixtypeinherphoneticstudybyincludingonlymonosyllabic andvowel-initialdisyllabicsuffixesbutnotconsonant-initialdisyllabicsuffixes. SchuhandKim(2007,pp.16–17),ontheotherhand,arguedthat“Particlesbear H only when they are associated with H that is part of a lexical tone pattern. Otherwise, particles bear L tone.” This means that H tones in suffixes must come from underlying H tone specifications. In other words, the different pitch patterns between the vowel-initial suffix -imyen ‘if’ and the consonant-initial suffix -pota ‘than’ discussed in Chang (2007) should come from the difference in their underlying pitch accent, not from the difference in initial segment type. This is evidenced by the fact that a consonant-initial disyllabic suffix -tul-i (pl. nom.) induces the H-LL pattern on monosyllabic nouns just like vowel-initial suffixes 123 76 H.Lee,J.Zhang while -mankhum induces L-HL. The implication, then, is that -mankhum and -tul-i belong to two different underlying pitch accent classes. The exact role of the suffix in the nominal pitch pattern of South Kyungsang, therefore, is controversial. We believe that the clarification of the following two issuescanbenefititsunderstanding.First,manyKoreansuffixeshaveallomorphsthat differ in the initial segment. For example, the topic marker is -un following a consonant-finalstembut-nunfollowingavowel-finalstem;thesuffixmeaning‘if’is -imyen or -lamyen2, also depending on whether the final segment of the stem it attachestoisaconsonantoravowel.Ifthenominalpitchpatternisindeeddetermined by the initial segment of the suffix, we would expect different pitch patterns when differentallomorphsareselected.Butifthepitchpatterndifferencecomesfromthe pitchaccentofthesuffix,thenwewouldnotexpecttheallomorphytoinfluencethe tones.Second,werecognizethatthedisyllabicsuffix-tul-iusedinSchuhandKim’s (2007) study is in fact a combination of two monosyllabic suffixes. The question, then, is whether there is indeed a contrast in disyllabic suffixes in their underlying pitchaccent.Weaimtoclarifythesetwoissuesinourownstudy. 1.3 Goals of the current study Thecurrentstudybroadlyaimstounderstandtheunderlyingpitchaccentsystemin SouthKyungsangKoreannounswithdifferenttypesofsuffixes.Tothisend,wefirst conductedanacousticstudywithfourmaleSouthKyungsangspeakersbyrecording monosyllabic and disyllabic nouns with various types of suffixes. Although acoustics studies of tone in South Kyungsang Korean already exist (Chang 2007; Lee 2008; Kim and Jun 2009), they did not focus on the tone patterns under different types of suffixes as their research goals were to clarify the tonal descriptions (Chang 2007), determine the location of the main accent (Lee 2008), and examine the interaction between lexical tone and focus prosody (Kim and Jun 2009).Ouracousticstudywillprovideafullpictureofthenominaltonalpatternsof South Kyungsang Korean. In addition, it will also address Chang (2007)’s assumptionaboutthetonalvariationsconditionedbytheinitialsegmentofsuffixes andthereforeresolvethecontroversyregardingthebehaviorofthesuffixes.Wealso paid attention to potential variations in the tonal patterns to see whether any disagreements in earlier descriptions are due to such variations and whether any systematicity emerges in the variations. Based on the acoustic results, we then provide a formal analysis couched in Optimality Theory (Prince and Smolensky 1993/2004) for the nominal pitch accent system of the language and see how the analysis works in longer nouns. We compare it with theoretical alternatives (Ramsey 1975; Kenstowicz and Sohn 1997; Kim 1997; Schuh and Kim 2007; Lee 2008; Lee and Davis 2009) and see how the system fits in with our current understanding of the typology of culminative tone systems (Evans 2009). 2 Wewouldliketocommentthat-imyenisallomorphicwith-myen,and-lamyenisallomorphicwith -ilamyen.Thesuffixes-imyen/-myenand-lamyen/-ilamyenhavethesamelexicalmeaning,‘if’.Sinceeach oftheallomorphsdiffersintheirnumberofsyllables,weselectedthedisyllabicsuffix-imyen/-lamyenand treated them as if they are allomorphs to test the relevance of the initial segment of the suffix to the surfacetonepattern. 123 PitchaccentofSouthKyungsangKorean 77 2 Acoustic study 2.1 Methods 2.1.1 Speech materials Ouracousticexperimentrecordedmonosyllabicanddisyllabicnounsindifferenttone classesasreportedinearlierliterature.Therecordingincludednounsbothinisolation andwithvarioustypesofsuffixes.Thereweresevenconditionsfortoneclasses:three for monosyllabic nouns (H1, H2, R) and four for disyllabic nouns (HH, HL, LH1, LH2). For each monosyllabic tone class, we recorded five words, and for each disyllabictoneclass,werecordedtenwords.ThesewordsaregiveninTables4and5. Inordertotesttheissuesregardingsuffixes,thefollowingfactorswereconsidered in the experiment: (1) the number of syllables in the suffix (monosyllabic or disyllabic);(2)theinitialsegmentofthesuffix(vowel-initialorconsonant-initial);and (3) whether the suffix has allomorphs that differ in the initial segment. Table 6 summarizesthesuffixconditionsintheexperiment.Themonosyllabicsuffix-toand disyllabic suffix -pota do not have phonologically conditioned allomorphs. But the monosyllabic suffix -un/-nun and disyllabic suffix -imyen/-lamyen have allomorphs thatareeithervowel-orconsonant-initial,andtheallomorphyisdeterminedbythelast segment of the noun root: a consonant-final noun takes the vowel-initial suffix allomorph;avowel-finalnountakestheconsonant-initialsuffixallomorph. Withfifteenmonosyllabicandfortydisyllabicnouns,eachwithfivesuffixconditions (includingisolation),werecorded275wordtokensintotalfromeachspeaker. 2.1.2 Speakers FourmalespeakersofSouthKyungsangKoreanparticipatedinthestudy.Theirage rangedfrom27to67yearsold(mean=40,s.d.=18).Allspeakershadlived and gonethroughprimaryandsecondaryeducationintheSouthKyungsangregionwith parents who spoke the same target dialect for at least 20 years; three of them were from Pusan city, and one of them was from Masan city. Three speakers were graduate or undergraduate students at the University of Kansas; the 67-year-old speaker was a short-term visitor to the U.S. None of the speakers reported any speech or hearing disorders, and all of the speakers were literate in Korean. Table4 Monosyllabicnounsusedintheacousticexperiment H1 R H2 mul ‘water’ mal ‘speech’ mal ‘horse’ nwun ‘eye’ nwun ‘snow’ mun ‘door’ pam ‘night’ pam ‘chestnut’ pi ‘tombstone’ pi ‘rain’ tol ‘stone’ swul ‘alcohol’ son ‘hand’ ton ‘money’ son ‘guest’ 123 78 H.Lee,J.Zhang Table5 Disyllabicnounsusedintheacousticexperiment HL HH LH1 LH2 anay ‘wife’ ceksam ‘jacket’ angay ‘fog’ kamca ‘potato’ cangsik ‘decoration’ hinsayk ‘white’ imca ‘owner’ mati ‘joint’ kewul ‘mirror’ kulim ‘painting’ nongtam ‘joke’ menci ‘dust’ koli ‘ring’ kwulum ‘cloud’ papo ‘fool’ namu ‘tree’ kutay ‘thou’ moki ‘mosquito’ paychwu ‘cabbage’ palam ‘wind’ kwuli ‘copper’ mole ‘thedayafter popay ‘treasure’ namul ‘wild tomorrow’ vegetable’ meli ‘head’ nalgay ‘wing’ samul ‘object’ poli ‘barley’ mole ‘sand’ phali ‘fly’ salam ‘person’ sokum ‘salt’ napi ‘butterfly’ phwulmu ‘bellows’ tampe ‘cigarette’ tali ‘bridge’ yelum ‘summer’ swukcu ‘beansprout’ wusan ‘umbrella’ elgwul ‘face’ Table6 Suffixconditionsintheacousticexperiment Wordinisolation Monosyllabicsuffix word+-to‘also’ Nodistinctallomorphs word+-un/-nun(topicmarker) V-orC-initialallomorphs Disyllabicsuffix word+-pota‘than’ Nodistinctallomorphs word+-imyen/-lamyen‘if’ V-orC-initialallomorphs 2.1.3 Procedure All four speakers were recorded in an anechoic chamber in the Phonetics and Psycholinguistics Laboratory at the University of Kansas using a cardioid microphone (Electrovoice-RE 20) and a solid state recorder (Marantz PMD 671). The test words were randomized and presented toeach speaker instandard Korean orthographyonacomputermonitorusingPowerPointwithafour-secondinter-trial interval. The speakers read each word once without repetition. Along with the orthography, pictures were provided for the speakers to help distinguish segmental homonyms in the orthography. For example, pictures of an eye and of snow were usedtodistinguishnwun(H1)‘eye’andnwun(R)‘snow’,bothofwhicharewritten as 눈 in Korean orthography. The stimuli were recorded at a sampling rate of 22,050 Hz and analyzed using Praat (Boersma and Weenink 2005). 2.1.4 Measurements InordertolocatetheHpeak(s)inthetestwords,theF0peakvaluewasmeasuredforall 1,100recordedtokens.ThemeasurementsoftheF0peakaredescribedinFig.1,which showsthemeasurementsinPraatforthewordsnwun(H1)-pota‘thaneyes’andnwun (R)-pota‘thansnow.’InFig.1,thefirst,second,andthirdsyllablesareindicatedasσ1, 123 PitchaccentofSouthKyungsangKorean 79 nwun (H1)-pota nwun (R)-pota 1 2 3 1 2 3 F0 Peak1 F0Peak2 F0Peak1 F0 Peak2 Fig.1 F0peaksforthetestwordsnwun(H1)-pota(left)andnwun(R)-pota(right).Solid-linearrows indicate the point where the F0 shows the first local maximum;dashed-linearrows indicate the point whereF0showsthesecondlocalmaximum σ2,andσ3,respectively;F0Peak1indicatesthepointofthefirstF0maximum,andF0 Peak 2 indicates the point of the second F0 maximum. The F0 maximum values betweenF0Peak1andF0Peak2arecomparable.WiththeseF0peakmeasurements, weadoptedthefollowingsimplecriteriontodecidewhetheraparticularsyllablehas anHtone.Forexample,iftherewasanobservableF0maximumbeforetheendofthe firstsyllableandanotherF0maximumbeforetheendofthesecondsyllable,boththe firstandthesecondsyllableswereanalyzedascarryinganHtone,asinnwun(H1) -potainFig.1;iftherewasanobservableF0maximumbeforetheendofthesecond syllableandasecondmaximumbeforetheendofthethirdsyllable,boththesecond andthethirdsyllablewereanalyzedascarryinganHtone,asshowninnwun(R)-pota. Thisprocedurewasusedforall1,100tokenstodeterminetheirpitchpatterns. 2.2 Results3 2.2.1 Dominant accent pattern Our data from the four speakers provide general support for the nominal pitch patterns in South Kyungsang Korean previously reported in the literature. In 3 Therewerelexicalanomaliesforfivenouns:theLH1wordpaychwu‘cabbage’withamonosyllabic suffixhadanLH-LpatterninsteadofLH-HbySpeakers1and4;ceksam‘jacket’intheHHgroupwas pronouncedasanLH2wordforSpeaker4;theH2nounspi‘tombstone’andson‘guest’patternedlike theH1classforSpeaker2;pi‘tombstone’,mal‘horse’,andson‘guest’fromtheH2classwereineither theH1ortheRclassforSpeaker3. 123 80 H.Lee,J.Zhang particular, Speaker 2’s pitch patterns are in line with earlier reports with few exceptions. Thus, we first report the accent patterns for Speaker 2 and then discuss variationsobservedbothbetweenandwithinspeakersinthefollowingsection.The F0 tracings for Speaker 2’s pitch patterns under suffixation are provided in Appendix 1. Speaker 2 (32 years old, Pusan) behaved identically to the previously reported patterns in Table 3 except for the monosyllabic word mal ‘horse’ (H2). The word mal‘horse’(H2)behavedlikeanH2wordunderdisyllabicsuffixationbutlikeanR word under monosyllabic suffixation (L-H instead of the expected H-L). With the exceptionofthisword,Speaker2’spitchpatternsareexactlyinlinewiththeearlier reports given in Table 3. Our acoustic data confirm that the neutralized H1-H2 and LH1-LH2 in isolation pattern differently when suffixes are added, implying separate underlying tone classificationsforthesestemclasses.Inaddition,thereisaparallelbetweenH1and HH,RandLH1,andH2andLH2.ForH1/HH,Hightonesoccuronthefirstandthe secondavailablesyllablesforthesestemclasses;forR/LH1,twohightonesoccuron the second and the available third syllables. For H2/LH2, different tonal patterns emergeunderdifferentsuffixtypes.AnHtoneoccursonlyintherootwith-imyen/ -lamyenforbothmonosyllabicanddisyllabicnouns,consistentwiththemonosyllabic suffix-toand-un/-nun.AnHtoneintherootwiththedisyllabicsuffix-imyenwasalso observedbyChang(2007).Withthesuffix-pota,ontheotherhand,anHtonealways occursonthefirstsyllableofthesuffix,andthereisnoHtoneinthemonosyllabicroot. ThesepitchpatternsintheH2andLH2classesindicatethatthedifferencesinpitch patternsarenotcausedbytheinitialsegmentofthesuffix,contraChang(2007),asthe V-initialversusC-initialsuffixallomorphyinvariablydidnothaveanyeffectonthe surfacetonesforanyofourspeakers.Giventhatweusedtwotrulydisyllabicsuffixes,- potaand-imyen/-lamyen,ourresultslendfurthersupporttoSchuhandKim’s(2007) suggestion that the suffixed-induced pitch differences are caused by different underlyingpitchaccentsonthesuffixes.4We,however,arenotinapositiontoaddress the question of whether all pitch accents can appear in a suffix or whether monosyllabicsuffixescanalsobearpitchaccents.Thesewillrequirefurthertestingof additionalsuffixes,andweleavethemopenastopicsforfutureresearch. 2.2.2 Inter- and intra-speaker variation WhilethefourspeakersgenerallyagreewiththeaccentpatternfortheHLandHH/ H1classes,Speakers1,3,and4showaccentvariationsfortheR/LH1andH2/LH2 accent classes. The three speakers’ accent patterns for R/LH1 and H2/LH2 are summarized in Table 7, and their tonal variations are indicated in bold. For Speaker 1 (35 years old, Pusan), the variations were mostly observed in monosyllabic R and disyllabic LH2 words. For the monosyllabic R class with the 4 Through his or her own elicitation with a Kyungsang consultant, one of our reviewers noted the discrepancybetweenC-initialandV-initialsuffixesin thatC-initialsuffixestend to carrya high tone whileV-initialsuffixesdonot,indicatingthatChang(2007)’sgeneralizationmaybevalid.Wedonot haveanexplanationfortheimbalanceinthesuffixaltones.Thepossibilityoftheinitialsegment-induced tonalallomorphysuggestedbysuchanimbalancestillrequiresfurtherinvestigation. 123
Description: