Durational Ratios of Long Vowels against Short Vowels in Silozi

Tsutomu Sato

Abstract This is the second interim report of an ongoing project called Durational Contrasts in the Worlds Languages, which tries to compare durational ratios between short and long vowels of six different languages. This paper focuses on one of African languages called Silozi or Lozi spoken in Zambia, which has a phonemic or semantic opposition only, marked by the vowel contrast, like Japanese language. The durational ratio is found to be rather small, compared with those of Japanese and Kiribati Sato, 2009b.

0. Introduction

This paper is the second report of annual project currently engaged in by the author. The main purpose is to compare durational ratios of segments of six different languages, particularly vowel durations. Remarks on the previous studies such as Lehiste 1970 and main reasons for starting this project were stated in the first report on Kiribati, i.e., Sato 2009b. The initial work prior to the series of fieldworks was Sato 2004 which treated Japanese short and long vowels. The first fieldwork in this project was carried out on the Christmas Island in May of 2009, i.e., Sato 2009b on Kiribati, published after two previous pilot works, Sato 2006 on the short and long vowels as well, and Sato 2009a , the main topic of which was an opposition of consonants, i.e., short and long nasals. The second fieldwork was conducted in Zambia, Africa, in July, and reported 15
in Sato 2010a , this paper. The third one was done in Mongolia, in September, and reported in Sato 2010b in this volume. The fourth data collection was held in India, in October, and the analysis is under way at the time of December, the result of which will be presented in Sato 2010c, forthcoming. The fifth fieldwork is planned to be carried out in Nicaragua on Miskito, in January of the year 2010. The detailed background information on language families of these five languages is described in Katzner 1995 and was summarized in Sato 2009b. Although these languages do not always share exactly the same linguistic features such as prosodic features and mora or syllable structures, they do have short and long vowels. Therefore, the main task of this project is to clarify the acoustic reality or linguistic facts of their vowel durations and compare them one another, leaving aside other phonetic and phonological issues such as stress and tone for the time being. As the next step, attempts will be made to relate these complex issues to main findings of this annual project.

1. Zambia and Silozi

Zambia is located in the right middle of south part of Africa, as shown in Map 1 URL 1. Fieldwork of this study was carried out in July of the year 2009, at a visitors center called the Jungle Junction on Bovu Island next to Nandavu village, near Livingstone, south west of the country. Zambia is a multi-lingual county. Webb and Kembo-Sure 2000: 320 estimate that the number of languages and/or dialects spoken in Zambia is 40, but the numbers shown in more recent references on the internet are over 70, 72 in URL 2, and 73 in URL 3, for example. The geographic distribution of the languages is presented on their Map 82 in Asher and Moseley 1994. Silozi or Lozi is one of the languages and belongs to Bantu Branch Yukawa 1993 in Benue-Congo subgroup of Niger-Congo family Katzner 1995. It is mainly used in southwestern Zambia, around Zambezi River 16
The location of Zambia in Africa and detailed map inside the country
plain, and in surrounding countries such as Botswana, Namibia, and Zimbabwe. Approximate number of the speakers in Zambia is around 500,000, six percent of the population URL 4, 5, and 6. Detailed descriptions of its definition and historical development are given in URL 7 and 8, entitled Barotseland, Silozi Language, and Lozi, respectively. Silozi is supposed to have five short vowels like Japanese and Kiribati, i.e., /i, e, a, o, u/, and corresponding long vowels, as indicated by Fortune 1977: 4. OSullivan 1993: page x has a slightly different view towards the long vowels but does mention short and long oppositions. This study follows Fortune in assuming that Silozi has five short and five long vowels.
2. Experimental Procedure
The following 12 pairs of phonemically contrasting words of Silozi were prepared as test tokens for production experiment of this study. 1 a. tisa bring tiisa hold tightly
b. muki someone who draws water mukii someone who locks c. fela come to an end feela but d. le this, here e. ngwe leopard f. kwa to, from g. luta teach h. loba break i. yoya live j. lundu tree k. mungi taker l. litu fruit lee this one ngwee money kwaa spit lutaa pea looba the day before yesterday yooya breathing luundu big hills muungi founder lituu seeds
In selecting these tokens, online Silozi-English dictionary in URL 9 18

and OSullivan 1993 were fully consulted. During the recordings, it was pointed out that looba in 1h and luundu in 1j could not be pronounced appropriately without adding a word ma as in ma looba and ma luundu, and these forms with ma were adopted in the recordings. However, in analyzing the data, it was reconsidered that adding ma changed the number of moras and affected the duration of the vowels. Therefore, these pairs of 1h and 1j were eliminated from the analysis and only other 10 pairs were taken into consideration. This problem should have been treated in a better way, but it could not have been helped. 13 Silozi male speakers took part in the data collection and all of them resided in Nandavu village at the time of recordings. In this interim report, the result of six participants will be treated, i.e., GK 34 years old , AL 53 , GM 24 , AK 26 , CK 38 , and DK 36. They were paid for their participations in the production experiments. The recordings were held in a hut in the village which was ideally quiet enough for the data collection. The repetition of reading the tokens were six times and last five were taken for the analysis. In this way, 900 tokens in total 12 pairs 2 words 6 dummies 5 repetitions 6 participants were recorded and the token list is shown in Appendix 1 at the end of this paper. As mentioned, relevant 600 tokens only, based on the 10 pairs, were put into analysis. The data were recorded into a linear PCM recorder, the PCM-D50 by SONY, with 44.10 kHz, 16 bit settings. At the same time, a portable Minidisc Recorder, the MZ-N920 by SONY was used for the backup recordings. Then, the recorders were connected with a PC and the data were automatically digitized into sound files.
3. Results and Discussion
A speech analysis program called Sugi Speech Analyzer was used and the measurements of vowel parts were conducted with reference to their waveforms and wide-band spectrograms. The examples are shown in the figures, DKs fifth readings of the pair, tisa and, tiisa in 1.1 and 1.2. The 19

Figure 1.1

The waveform and wide-band spectrogram of tisa by DK

Figure 1.2

The waveform and wide-band spectrogram of tiisa by DK
Durational Ratios of Long Vowels against Short Vowels in Silozi Table 1 GK tisa/tiisa muki/mukii fela/feela le/lee ngwe/ngwee luta/lutaa kwa/kwaa yoya/yooya mungi/muungi litu/lituu Ave loba/looba lundu/luundu 1.06 0.76 1.07 1.09 1.09 1.18 0.9 1.11 1.61 1.3 1.117 0.79 1.66 Average durational ratios GM 1.45 1.43 0.9 1.12 0.99 1.16 1.01 1.07 1.22 1.1 1.145 0.81 0.83 AK 1.24 0.91 1.06 0.0.91 1.11 1.21 1.04 0.92 1.033 1.07 0.98 CK 0.86 0.97 1.1.19 1.1.31 1.43 1.29 1.129 0.99 0.73 DK 1.1 1.25 1.02 0.85 1.14 0.92 1.02 1.04 1.11 0.87 1.032 1.13 0.58 Ave 1.281667 1.106667 1.035 0.988333 1.093333 1.31 1.011667 1.153333 1.32 1.19 1.149 0.943333 0.95

AL 1.98 1.32 1.15 0.94 1.15 2.46 1.03 1.18 1.51 1.66 1.438 0.87 0.92
duration of short vowel /i/ in Figure 1.1 is 84 ms, while the one of long vowel /i:/ in Figure 1.2 is 109 ms. The resulting durational ratios of the six participants and average values are shown in Table 1. Durational ratios are given when the duration of long vowel is divided by that of corresponding short vowel. Just for reference, the results of discarded pairs of loba/looba and lundu/luundu are given here. The average durational ratio of the relevant 10 pairs is 1.149. This value is relatively smaller, compared with the ratios of Kiribati, 1.86 as shown in Sato 2009b and that of Japanese, 2.66, as in Sato 2004.

4. Remaining Issues

The result was based upon the six participants out of 13 in total, and so the remaining data should be analyzed for better comparison. That said, the tentative resulting value was found to be rather low, i.e., 1.14. Therefore, a statistic test such as the Welchs test must be run on the data 21
to see if the difference between durations of short vowels and those of long vowels is significant. Needless to say, prosodic features such as tone patterns must be taken into consideration as the next step of this Silozi analysis.
Acknowledgements First of all, I am very grateful to Brett Murray Saunders of the Jungle Junction on the Bovu Island, for letting me record his stuff members as informants of production tests of this research and his constant encouragements. I would like to show my deep gratitude to Godfrey Kalemba, one of his stuff, for helping me to find the informants and introducing me to local people of the Nandavu village. Without his precious support, the data collection could not have been completed. I would like to thank Aggrey Sianga Lubinda, a community school teacher, Nandavu, for selecting appropriate words for final version of the token list in Silozi. Harriet Malambo also helped me to pick out Silozi words for making the list and I would appreciate her greatly. Evelyn Row kindly let me use her dictionary during my whole stay at the Jungle Junction, which was very helpful. Special thanks goes to Kazunori Yumoto of the Keikyu Travel Service for his travel arrangement for me to visit Zambia. Finally, I was inspired by the following two works by Peter Ladefoged, a late great phonetician, which made me feel like carrying out this fieldwork in Africa: A Phonetic Study of West African Languages, published in 1964, and Phonetic Data Analysis: An Introduction to Fieldwork and Instrumental Techniques in 2003.
References Asher, R. E. and Christopher Moseley eds. 1994 Atlas of the Worlds Languages. Routledge. Fortune, George 1977 An Outline of Silozi Grammar. Bookworld Publishers. Katzner, Kenneth 1995 The Languages of the World. Routledge. Lehiste, Ilse 1970 Suprasegmentals. The MIT Press. OSullivan, Owen 1993 English-Silozi Dictionary. Zambia Educational Publishing House. Sato, Tsutomu 2004 The Production and Perception of Japanese Identical Vowel Sequences and their Acoustic Characteristics in Connected Speech. Ph.D. Thesis. University of London. Sato, Tsutomu 2006 Toward Cross-linguistic Analyses of Vowel Durations of World Languages: the Durational Differences between Short and Long

Durational Ratios of Long Vowels against Short Vowels in Silozi Vowels in Kiribati. The Journal of English and American Literature and Linguistics, Meiji Gakuin University, 118, 129 136. Sato, Tsutomu 2009a The Pilot Study on Long Nasals in Kiribati. The Journal of English and American Literature and Linguistics, Meiji Gakuin University, 123, 143 150. Sato, Tsutomu 2009b Durational Contrasts in the World Languages: Short and Long Vowel Oppositions in Kiribati. The Journal of English and American Literature and Linguistics, Meiji Gakuin University, 124, 153 165. Sato, Tsutomu 2010b, in this volume Intransitive and Transitive Distinction Marked by Short and Long Vowels in Mongolian. The Journal of English and American Literature and Linguistics, Meiji Gakuin University, 125, 25 33. Sato, Tsutomu 2010c, forthcoming The Derivative Markers by Long Vowels in Hindi. Webb, Vic and Kembo-Sure eds. 2000 African Voices: An Introduction to the Languages and Linguistics of Africa. Oxford University Press. Yukawa, Yasutoshi 1993 Bantu Lauguages. In Kamei, Takashi, Mutsuro Kawano, and Eiichi Chino eds. Gengogaku Daijiten, 394 397. Sanseido. URL 1. http://www.worldatlas.com/webimage/countrys/africa/zm.htm 2. http://en.wikipedia.org/wiki/Zambia 3. http://www.mapsofworld.com/zambia/culture/language.html 4. http://en.wikipedia.org/wiki/Lozi_language 5. http://www.everyculture.com/Africa-Middle-East/Lozi-Orientation.html 6. http://www.panafril10n.org/wikidoc/pmwiki.php/PanAfrLoc/Lozi 7. http://www.barotseland.com/sil-eng1.hhttp://www.barotseland.com/ silozi1.htmtm 8. http://www.mnsu.edu/emuseum/cultural/oldworld/africa/lozi.html 9. Silozi-English dictionary http://www.websters-online-dictionary.org/translation/Lozi/

Appendix 1

Token list
1. muki 2. lituu 3. loba 4. lutaa 5. buka 6. fela 7. tiisa 8. kwa 9. lundu 10. kata 11. ngwe 12. muungi 13. mukii 14. lee 15. mwana 16. luta 17. 19. ma looba ma luundu 18. litu 20. lii 21. yoya 22. mungi 23. tisa 24. kwaa 25. muutwa 26. feela 27. ngwee 28. le 29. yooya 30. mandu

Intransitive and Transitive Distinction Marked by Short and Long Vowels in Mongolian

Tsutomu Sato

Abstract This study is the third interim paper of a project on the comparisons of differences in durations of segments of worlds languages, especially vowel contrasts of six different languages, which was initially designed in Sato 2004 , a doctoral thesis by the author. Unlike Japanese and Silozi Sato 2010a, in this volume , Mongolian, a main subject of this paper, has a unique contrast shown by short and long vowels to distinguish an intransitive action from a transitive action. This paper attempts to clarify the average durational ratio of this contrast, based on the data collected from six male native speakers of Mongolian during a fieldwork conducted in September, 2009.

0. Introduction

This paper, Sato 2010b , is the third report of an ongoing annual fieldwork project of the academic year, 2009, which follows Sato 2009 on short and long vowels in Kiribati and Sato 2010a, in this volume on Silozi short and long vowels. The main topic of this study is basically on the same theme, namely, linguistic opposition made by short and long vowels in Mongolian, which is a Mongolian subgroup of Altaic family Katzner 1995. However, unlike the oppositions in Japanese treated in Sato 2004 and the ones in Silozi in Sato 2010a which only have phonemic or semantic contrasts, Mongolian has an opposition, what we call here, intransitive and transitive contrast, as will be explained in section 25
2. This paper concentrates on this contrast marked by short and long vowels in Mongolian and tries to make clear the relation between the short vowel duration and long vowel duration. Then, comparisons will be made with the previous findings in Japanese, Kiribati, and Silozi. In concluding this paper, another interesting linguistic usage contrasted by short and long vowels in Hindi is mentioned, which will be fully discussed in a forthcoming paper, Sato 2010c.

1. Mongolian

Mongolia is situated in the east and central Asia and its official language is Mongolian URL 1. The language belongs to Mongolian subgroup of Altaic family Katzner 1995 , and it has over five million speakers URL 2. Map 1 shows the countrys location URL 3. The data collection of this study was conducted in and around Ulan Bator, the capital, in September of the year 2009, and 13 male speakers and six female speakers, 19 in total joined in the production experiments, who speak the Khalkha dialect mainly spoken in Ulan Bator. paper, the result of six male speakers will be reported. In this interim
2. Transitivity Distinguished by Short and Long Vowels
It has been recognized that Mongolian has phonemic contrasts between short and long vowels. This study follows the view that seven short vowels have their corresponding seven long vowels, held by Campbell 1995: 346 , Nukushina 1998: 360 , Otgonpurev 2007: 286 and URL 2, although other scholars such as Maddieson 1984: 281 and Kuribayashi 1992: 493 have slightly different positions. More specifically, the relation between the orthographic letters and the seven short vowels are: /i/, /e/, a /a/, y /o/, o / /, / /, and Y /u/. The double letters indicate their corresponding long vowels, such as aa /a:/. While gathering Mongolian word pairs containing short and long 26

The location of Mongolia

vowels for making list of tokens before the fieldwork, it was noticed that there are pairs where a short vowel marks intransitive action, whereas a long vowel denotes transitive action, such as /asax/ catch fire and /asa:x/ make something burn, and /zo sox/ come to a stop and /zo so:x/ bring to a stop. This study focuses on this contrast of intransitive and transitive functions distinguished by a short vowel and a long vowel and tries to clarify the average durational ratio of this kind of pair in Mongolian.
3. Experimental Procedure
After referring to Mongolian dictionaries, Otgonpurev 2007 and
Tuvshintulga 2004 , 25 pairs of Mongolian words were found to be appropriate for making token list of production experiments. Out of the 25 pairs, 8 pairs are examples of the intransitive/transitive contrasts to be analyzed in this paper, while other 17 pairs of 34 words make phonemic contrasts producing semantic differences only, and they are to be analyzed in forthcoming papers. The relevant 8 pairs are: 1 a. acax/asax/ catch fire acaax/asa:x/ make something burn b. xa ax/xatax/ become dry xa aax/xata:x/ dehydrate c. xa ax/xalax/ become warm xa aax/xala:x/ heat up something d. ypax/nurax/ fall down ypaax/nura:x/ destroy e. x x f. g. x/holdox/ be frozen x/holdo:x/ make something frozen
orcox/zo sox/ come to a stop orcoox/zo so:x/ bring to a stop ox/notsox/ be lit 28
oox/notso:x/ set fire to h. xox pox/xohirox/ get hurt xox poox/xohiro:x/ cause damage The six male participants are GA 29 years old , US 42 , OT 19 , GN 22 , MY 21 , and DO 31. After practice reading for a few times, they were asked to read the token list of 50 words, which contains the relevant 8 pair of 16 words, with five-time repetitions, producing 480 vowel parts to be analyzed, i.e., 16 words 5 repetitions 6 participants 480. They were paid for their participations in the production experiments. The data were recorded into a linear PCM recorder, the PCM D50 by SONY, with 44.10 kHz, 16 bit settings. Then, the recorder was connected with a PC and the data were automatically digitized into sound files. At the same time, a portable Minidisc Recorder, the MZ N920 by SONY was used for the backup recordings.

4. Results and Discussion
Waveforms and wide-band spectrograms were referred to in measuring the short and long vowel parts in the data by the Sugi Speech Analyzer. Figure 1.1 shows the fifth reading of the word acax /asax/by DO, and Figure 1.2 is the example of the corresponding word with a long vowel, acaax/asa:x/. The duration of short vowel is 59 ms and the one of long vowel is 147 ms. The resulting durational ratios by the individuals and the average of the whole data are shown in Table 1. The average durational ratios differ individually and from tokens to tokens, especially acax/acaax and xohirox/xohiroox pairs being rather short. However, the values as a whole are rather consistent and the average of all the values is 2.48. This is the result of limited data, i.e., pairs of transitivity uttered by only six speakers out of all the 19 participants. However, a tentative comparison shows that the average is very close to that of Japanese data in Sato 2004 , i.e., 2.66, 29

Figure 1.1

The waveform and wide band spectrogram of /asax/ by DO

Figure 1.2

The waveform and wide band spectrogram of /asa:x/ by DO
Intransitive and Transitive Distinction Marked by Short and Long Vowels in Mongolian Table 1 The average durational ratio of intransitive and transitive pairs GA acax/aca:x xatax/xata:x holdox/holdo:x xalax/xala:x zo sox/zo so:x notsox/notso:x nurax/nura:x xohirox/xohiro:x Ave 2.75 2.62 2.57 3.1 2.19 1.89 2.75 2.42 US 2.54 3.12 3.47 3.44 3.18 3.58 3.98 2.38 OT 2.37 2.32 3.04 3.04 2.24 2.9 2.34 1.8 GN 1 1.77 1.73 2.04 2.35 1.85 2.15 2.14 MY 1.32 2.19 1.95 2.25 1.8 1.5 2.11 1.3 DO 2.83 2.83 3.42 3.9 3.17 3.33 2.59 1.86 Ave 2.135 2.475 2.696667 2.961667 2.488333 2.508333 2.653333 1.983333 2.487708
2.53625 3.21125 2.50625 1.87875

1.8025 2.99125

and larger than 1.85 of Kiribati in 2009 and 1.14 in Silozi in 2010a.

5. Further Studies

Along the lines of this annual project on short and long vowels in worlds languages, data collections of four languages including Hindi were completed and the durational ratios of three languages were already made clear, i.e., Kiribati, Silozi, and Mongolian, although the analyses are still under way, leaving some data untouched yet. Putting aside their prosodic differences, it was found that there might be variations of durational ratios among these three languages, and it may be possible to posit an order of durational ratios from high to low, i.e., 2.48 of Mongolian this paper , 1.85 of Kiribati Sato 2009 , and 1.14 of Silozi in this volume. Remaining data of the three languages will be analyzed in forthcoming papers. In conducting this project, it was noticed that the oppositions of short and long vowels are used to perform other linguistic functions than not only making phonemic contrasts, to produce semantic differences. In Sato 2009 , an attempt was made to see if durational ratio difference arises between singular/plural group and semantic difference group in 31

Kiribati, producing no such a big difference, 1.85 and 1.86. As illustrated in this paper, Mongolian has the transitivity function and the durational ratio was found to be 2.48. The data of other 13 participants must be analyzed to get more reliable result, first, and the durational ratio of another group of phonemic or semantic difference group of Mongolian, 17 pairs of 34 words, will be compared, as the next step. In the next paper of this project, Sato 2010c , another linguistic function by a short and long vowel contrast in Hindi will be the main topic, data collection of which was already conducted in October of 2009. The Hindi contrast not only denotes parts of speech such as the derivation of an adjective form from a noun form, or vice versa, but also it sometimes shows a relation between an agent and its actual action, verb form, such as /kanhari/ a boatman and /kanhari:/ piloting a boat. Other types of linguistic usages by short and long vowel contrasts are supposed to be found from more languages as future topics of this project and durational ratios will be clarified for comparisons. Again, this project will not be completed without proper treatment of prosodic aspects, such as Mongolian stress assignment, and syllable structures. As another interesting finding, URL 2 introduces examples of durational differences of long vowels in different positions in a word, from The Phonology of Mongolian by Svantesson et. al. 2005 , quoting that the average ratio of long vowel against short vowel in word-initial syllables is 2.08, as opposed to 1.27 in word-internal and word-final syllables. These kinds of effects by syllable positions are also important factors to be considered in continuing this project.
Acknowledgements I would like to show my deep gratitude to R. Gansukh, a freelance translator and tourist guide in Ulan Bator, for his valuable comments on the Mongolian token list before the production test and finding the participants. Needless to say, the data collection was impossible without his kind supports. Usukhsaikhan also helped me a lot to choose appropriate words for making the
Intransitive and Transitive Distinction Marked by Short and Long Vowels in Mongolian token list. I am grateful to Oyungerel Mandshir, the service manager of the Black Ibex Expeditions for her arrangement of my fieldwork trip in Mongolia. Last but not least, I would like to thank all the 19 Mongolian participants for their wonderful cooperation in the production experiments.
References Campbell, Geroge L. 1995 Concise Compendium of the Worlds Languages. Routledge. Katzner, Kenneth 1995 The Languages of the World. Routledge. Kuribayashi, Hitoshi 1992 Mongolian. In Kamei, Takashi, Mutsuro Kawano, and Eiichi Chino eds. Gengogaku Daijiten, 4, 492 498. Sanseido. Maddieson, Ian 1984 Patterns of Sounds. Cambridge University Press. Nukushina, Renzo 1998 Mongolian. In The Institute of Language Research, Tokyo University of Foreign Studies ed. A Guide to the Worlds Languages. Part 2: Asia and Africa. 358 373. Sanseido. Otgonpurev, Baatar 2007 Mongolian-Japanese Dictionary. Kokusai Gogaku sha. Sato, Tsutomu 2004 The Production and Perception of Japanese Identical Vowel Sequences and their Acoustic Characteristics in Connected Speech. Ph.D. Thesis. University of London. Sato, Tsutomu 2009 Durational Contrasts in the World Languages: Short and Long Vowel Oppositions in Kiribati. The Journal of English and American Literature and Linguistics, Meiji Gakuin University, 124, 153 165. Sato, Tsutomu 2010a, in this volume Durational Ratios of Long Vowels against Short Vowels in Silozi. The Journal of English and American Literature and Linguistics, Meiji Gakuin University, 125, 15 24. Sato, Tsutomu 2010c, forthcoming The Derivative Markers by Long Vowels in Hindi. Tuvshintulga, TS 2004 Mongolian-English Dictionary. Y. Solongo. URL 1. http://en.wikipedia.org/wiki/Mongolia 2. http://en.wikipedia.org/wiki/Mongolian_language 3. http://www.worldatlas.com/webimage/countrys/asia/mn.htm