Proceedings Fonetik 2009 - Institutionen för lingvistik

Proceedings, FONETIK 2009, Dept. of Linguistics, Stockholm Universitypossible to analyse (see table 1 for the numberof recordings for each speaker). Two of thespeakers (Sp2 and Sp4) were recorded twice.For four of the 24 speakers no useable recordingswere made. Of the remaining 20 speakers,8 speak the non-tonal dialect and 12 the tonaldialect.The maximal fundamental frequency wasmeasured in each copy of the reduplicatedwords using the Praat analysis program.Results and discussionThe results are shown in table 1.Concerning question (1) above, the resultsshow that most speakers follow the pattern thatthe first copy of the reduplicated adjective hashigher F0 than the second one. 14 of the 20speakers use this high–low pattern in all theirproductions. These are the 5 non-tonal speakersSp1, Sp3, Sp5, Sp6, Sp10 and the 9 tonalspeakers Sp2, Sp4, Sp13, Sp16, Sp20, Sp21,Sp22, Sp24, Sp25. Two speakers, Sp9 (nontonalmale) and Sp17 (tonal female) use a completelydifferent tone pattern, low–high. Theremaining speakers mix the patterns, Sp8 andSp18 use high–low for blia-blia but low–highfor thaw-thaw, and the two speakers Sp11 andSp23 seem to mix them more or less randomly.As seen in table 1, the difference in F0 betweenthe first and second copy is statistically significantin the majority of cases, especially forthose speakers who always follow the expectedhigh–low pattern. Some of the non-significantresults can probably be explained by the largevariation and the small number of measurementsfor each speakers.The second question is whether or not thetonal speakers retain the tone difference in thereduplicated form. In the last column in table 1,we show the difference between the mean F0values (on both copies) of the productions ofthaw-thaw/tháw-tháw and blia-blia/plìa-plìa. For11 of the 12 speakers of tonal dialects, F0 was,on the average, higher on tháw-tháw than onplìa-plìa, but only 2 of the 8 speakers of nontonaldialects had higher F0 on thaw-thaw thanon blia-blia. An exact binomial test shows thatthe F0 difference is significant (p = 0.0032) forthe tonal speakers but not for the non-tonalones (p = 0.144).One might ask why the majority of nontonalspeakers have higher F0 on blia-blia thanon thaw-thaw. One possible reason is that bliabliawas always recorded before thaw-thaw, andthis may have led to higher engagement whenblia-blia was recorded than when thaw-thaw wasrecorded just after that; see House et al.(forthc.) for the role of the speakers’ engagementfor Kammu intonation.ConclusionThe results show that the great majority of thespeakers we recorded used the expected fixedpattern, high–low, for intensifying reduplication,independent of their dialect type, tonal ornon-tonal. Furthermore, the speakers of tonaldialects retain the contrast between high andlow lexical tone when they apply this fixed tonepattern for adjective reduplication.AcknowledgementsThe work reported in this paper has been carriedout within the research project Separatingintonation from tone (SIFT), supported by thebank of Sweden Tercentenary Foundation (RJ),with additional funding from Crafoordskastiftelsen.ReferencesDuanmu, San (2000) The phonology of StandardChinese. Oxford: Oxford UniversityPress.House, David; Karlsson, Anastasia; Svantesson,Jan-Olof and Tayanin, Damrong(forthc.) The phrase-final accent in Kammu:effects of tone, focus and engagement. Papersubmitted to InterSpeech 2009.Svantesson, Jan-Olof (1983) Kammu phonologyand morphology. Lund: GleerupSvantesson, Jan-Olof and House, David (2006)Tone production, tone perception andKammu tonogenesis. Phonology 23, 309–333.83