SECOND LANGUAGE PROSODY:
TEACHING STUDENTS TO SOUND BETTER
Keiko Yamasaki
1．INTRODUCTION
English education in schools is being strongly promoted by the Japanese govern-
ment.For example,English lessons are being introduced in elementary schools,and the Ministry of Education and Science has designated certain schools Super English Language High Schools so that they can contribute to future improving English language education in Japan(Ministry of Education and Science,2002) . Moreover,
not only grammatical knowledge,which has long been emphasized in Japan,but also speaking skills, like pronunciation and fluency, are being focused on in English classes. However, intonation and prosody still tend to be ignored.
In this paper, I will discuss intonation and prosody in discourse:how important they are and how learnersʼabilities can be improved by teaching them.
1．1 INTONATION AND PROSODY IN DISCOURSE
Intonation and prosody have an important role in discourse. They are not only useful for distinguishing between interrogative and affirmative sentence. They are also related to grammar and are useful for showing the meaning or the focus of the discourse. In addition, they are associated with relations with other people and emotional expression: intonation has a lot of information that cannot be com-
municated by words alone. Significantly, persuasion and lucidity can be enhanced through intonation(Sugito1997).
A number of studies have indicated that structure of spoken discourse is reflected in prosody, for example pitch range and pause duration. These two are the most important and significant researched prosodic means for structuring utterances 東京女子大学言語文化研究（Studies in Language and Culture)17(2008）pp.69‑86
(Mayer, Jasinskaja, & Kolsch2006).
Pauses play an important role in discourse. It is necessary not only for speakers to draw a breath,but also for listeners to understand what speakers utter.They have a grammatical role in discourse at the same time: speakers pause unconsciously before beginning large grammatical segments (Sugito1997). The duration of silent pauses appears to have a similar effect to that of the degree of pitch reset,discussed below. Longer pauses appear before topic changes and the shortest pauses are seen between intonational phrases dealing with the same topic (Mayer, Jasinskaja, &
Kolsch2006).
Many studies have shown that a greater pitch range is associated with the beginning of paragraphs and topic shifting in discourse.Readers use a high pitch at the beginning of the first sentence of a new paragraph.This phenomenon is referred to as a topic reset. A good example of this phenomenon can be seen in news broadcasts,which have different news items or topics,and each of them is introduced with an extra high pitch(Wichmann2000) . Low pitch reset, in contrast, generally indicates topic continuity. Compressed pitch range, on the other hand, signifies the end of the paragraph or the closing a topic.
However, printed paragraphs only account for a small part of the readerʼs prosodic topic signaling(Wichmann2000) . According to Wichmannʼs research, the organization of the printed text does not always correspond with that of spoken performance. While in spoken performance, paragraphs tend to depend on the rhetorical relationships, but are related to time or place in printed paragraphs.
Paragraph breaks are not necessarily correlative even if one begins and ends with a shift in the temporal coherence.
1．2 TEACHING INTONATION
Intonation and prosody are not focused on in second language classrooms in Japan even though they are important in discourse. One reason is thought to be the difficulty of teaching intonation:
In speech of native speakers intonation is planned at a subconscious level, the units and contours being mapped out in the speakerʼ s mind before he/she decides
which words he/she will use.Native speakers respond intuitively to the intona- tion used by other native speakers. Foreign learners may need to discuss this seemingly complex phenomenon, and we must, therefore, find some way of enabling the learner to capture the fleeting moment and analyze it (Bradford
1998, p.2).
It is said that for teaching intonation, getting learners to consciously understand intonation patterns is effective and efficient (Okuda1975). Several attempts have been made to support teaching intonation to learners with electronic devices that give visual feedback on intonation. These enable learners to see how the pitch pattern develops and how parts of the sentence are related to parts of the pattern,
and to become aware of mistakes that they may not otherwise notice(Spaai &
Hermes1993).Munakata(2003)studied teaching intonation focusing on pitch reset.
Her study showed that teaching with visual feedback is effective for intonation learning.It however showed no clear effects of teaching pitch on the pause duration,
although she did not investigate this in depth. On the other hand, the research of Mayer,Jasinskaja,& Kolsch(2006),described above,suggests that it is related and that teaching about pauses may also be effective.
Another limitation of Munakataʼs study is that it only investigated one text.This single text was used for pre-testing,teaching and post-testing.She did not test to find out whether the intonation learning would be transferred when reading another text.
1．3 RESEARCH QUESTION
The purpose of this study is build upon the work of Munakata.I will research the effectiveness of teaching pauses and pitch on intonation, both by themselves and in combination. The research questions for this study are followings.
1．Is teaching intonation by focusing on pitch reset alone effective?
2．Is teaching intonation by focusing on pauses alone effective?
3．Is teaching intonation by focusing on both pitch reset and pauses effective?
4．Which of the three is the most effective?
5．Is intonation learning transferred when reading another text?
2．METHOD 2．1 SUBJECTS
English native speaker (ENS)speech samples were obtained from three British women. They had been in Japan for between three and four years, and all were teaching English at Tokyo Womanʼs Christian University.
The English learner subjects were fifteen native Japanese speakers. They are all seniors at Tokyo Womanʼs Christian University and aged between 21and23years old. They had studied English for between eight and sixteen years.
2．2 MATERIAL 2．2．1 SHORT STORY
Two short, intermediate-level stories were used in this study (Hill1967). They have approximately150words each, and both stories have four paragraphs. There were three criteria for selecting these two stories. First was that both stories consisted of prose and dialogue. I expected that intonational features of prose and those of dialogue might be different.The next was that the two short stories had the same number of paragraphs and approximately the same number of words so that the data from both stories would be able to be compared simply. The last one was that the stories did not have unfamiliar words. This was so that the subjects could read with no difficulties and focus on the content. The texts were:
Text1
［0］Mr Black gave his wife money every Friday evening,［1］but she always spent it before the next Wednesday,［2］so that for the next three days she had none.
［3］Every Tuesday evening Mr Black asked her,［4］ʻBut what did you spend all that money on?ʼ［5］and she always answered, ［6］ʻI donʼt know.ʼ
［7］One Friday Mr Black brought home an exercise book and a pencil and［8］
gave them to his wife with her money.［9］ʻNow look!ʼhe said to her.［10］When you get money from me,［11］write it down on this page, ［12］and on the opposite page write down what happens to the money.ʼ
［13］When Mr Black came home the next Tuesday,［14］his wife came to him and showed him the book.［15］ʻI have done what you told me,ʼshe said happily.
［16］On one page she had written［17］ʻFriday,28 June.［18］I got￡18from John;ʼ［19］and on the opposite page,［20］ʻ Tuesday,2 July.［21］I have spent it all.ʼ
Text2
［0］Mr Blackʼs wife was very ill,［1］and at last she died.［2］After a few months,［3］Mr Black married again.［4］ His new wife was a widow.
［5］Exactly seven days after he married her,［6］she had a baby.
［7］Mr Black at once hurried away to the market［8］and bought some paper, some pencils, some pens and some childrenʼs books.［9］Then he hurried back home again with these things［10］and put them beside the baby. ［11］His new wife was surprised.［12］ʻWhat are you doing?ʼshe said. ［13］ʻThe baby wonʼt be able to use those things for a long time. ［14］Why are you in such a hurry?ʼ
［15］Mr Black answered,［16］ʻYou are quite wrong,my dear.［17］Our baby is not an ordinary baby.［18］It came in seven days instead of nine months, ［19］
so it will certainly be ready to learn to read and write in a few weeks from now.
2．2．2 EQUIPMENT
All the speech samples were recorded using Speech Analyzer (SIL International
2005),a computer program for the acoustic analysis of speech sounds.This software can display prosodic features like wave forms and pitch contours, allowing pause duration and pitch to be measured.Measurements were made at the numbered points in the texts(above).
2．2．3 TEACHING MATERIAL
The teaching material consisted of screen captures of the ENSsʼpitch contours with topic resets circled.Figure1shows an example.For Group A,which was taught
only about pauses, no marked pitch contours were prepared.
2．3 PROCEDURE
First, the three ENSs were asked to record the short stories. They had time to practice before recording,and then they were asked to read as naturally as possible.
The fifteen Japanese learners of English were randomly divided into three groups.
The general procedure was as follows:
1．Practice(Text1) 2．Initial recording(Text1) 3．Teaching(Text1) 4．Final recording(Text1) 5．Practice(Text2) 6．Recording(Text2)
Each group was asked to read the same two short stories as the ENSs.They also had time to practice, approximately10minutes per story and were allowed to ask how to pronounce words and about their meanings. After practicing, they recorded
themselves reading the story.
In Step3,the three groups received different treatments.Group A were instructed only about pauses, Group B about pitch only, and Group C about both pauses and pitch.In all groups the material used for teaching was Text 1.Each group was given instruction in its own specific prosodic feature and then got visual input which showed the native speakersʼpause durations and /or pitch contours in Text1.They were then able to see their own pause durations and /or pitch contours on computers,
so they could compare them with those of the native speakersʼ.At the same time,the features of the English native speakersʼpatterns found in my analysis were pointed out to the subjects.
After the teaching session, the subjects rerecorded themselves reading Text1. After this they were given Text2, and after time for practice and to ask questions about pronunciation and meaning, they recorded themselves reading Text 2. This was done in order to investigate whether learning gains would transfer to a new text.
3．RESULTS
3．1 ENGLISH NATIVE SPEAKERS 3．1．1 PAUSES
Since there was variation in the ENSsʼspeech rates and pause lengths,total pause duration was normalized to the approximate group mean, 12.5seconds with Text2
and10seconds with Text2,in order to make the speakers more easily comparable.
Individual and mean pause lengths with Text1are shown in Figure2. Sentences 3,7,9,13,and15had especially long pauses,with the pauses at3,7,13,and15over 1.0second long.Since new written paragraphs appear at Sentences3,7,and13,this shows that longer pauses tended to occur before new paragraphs /topics. More variations were found in pause length in Paragraphs 3and4(after pause7),which contain more conversational text,than in the other paragraphs,which consist mainly of prose.
Overall patterns with Text2(Figure3)were similar to those found with Text1. Longer pauses appeared before new paragraphs/topics also in Text2.Three pauses (5,7,and15)out of the five longest pauses(5,7,9,11,and15)occurred at paragraph
boundaries.Of these three pauses,pauses7and15were over1.0second.The pattern of variation was also similar to Text1 , with pause length variations greater in Paragraphs3and4(after Pause6), which have more conversational text.
3．1．2 PITCH
Figure4shows the pitch at the each segment produced by ENSs with Text 1.The ENSs started new paragraphs with a high pitch:the beginning of new paragraphs,at
Sentences0,3,7, and13, were300to357Hz. ENS3made particularly high pitch resets(over400Hz)at Sentences7and13 . In contrast, during topic continuations,
for example at Sentences1,2, and21, pitch is lower, around260Hz. The highest pitch was at0and was357Hz on average.This is166Hz higher than the lowest pitch in this text.
Although extra high pitch resets were seen at the beginning of new paragraphs with Text1,they were not with Text2(Figure 5).With Text2,only Sentence0was over300Hz.The other three(5,7,and15)were around 250Hz.On the other hand,
some sentences started with very high pitch resets in the middle of topic continuity.
For example,Sentence12,which was380Hz on average,is in the middle of the third paragraph in this text. It is assumed that this is because of differences between spoken and written paragraphs(Wichmann 2000). ENSʼs pitch range was also greater.Pitch resets ranged from199Hz to 380Hz,with the highest pitch reset about
180Hz greater than the smallest one.
3．2 JAPANESE LEARNERSʼ 3．2．1 INITIAL READING
Total pause durations produced by the fifteen Japanese learners were normalized to14seconds, their approximate mean total pause length in the initial recording of Text1.The mean pause durations are shown in Figure 6.Only two points,13and16,
are over1.0second. Sentences3and7, at the beginning of new paragraphs, have relatively long pauses. However, they are below 1.0second.
The ratio of total pause length to total speech length(pause:speech ratio)was also low compared to the ENSs.With the ENSs,this ratio was 23.4%.Their total pause length was12.5seconds out of a total speech length of 41.09seconds. For the Japanese learners the ratio was19.3%;13.86seconds of pauses out of a total speech length of71.5seconds.
Figure7shows the pitch in Text1by Japanese learners.Although some high pitch resets of more than300Hz are seen in the ENS data,only pitch at Sentence0is over
300Hz in the Japanese learnersʼspeech. Pitch was higher at the beginning of new paragraphs(at Sentences0,3,7and13) , but these were only about50Hz greater
than other pitch resets. The ENSʼs topic resets were100Hz to200Hz greater than other pitch resets.
3．3 DEVELOPMENT
Japanese learnersʼpause and pitch features after teaching are describing in this section.
Pause durations in each treatment group are shown in Figures8to10.Change in
Groups A and C are clear. After teaching, pauses were longer at the beginning of each paragraph, over1.0second. In group B, the pause at 7 (the beginning of Paragraph3)became longer,but at the beginning of the other two paragraphs(3and
13), they did not and, at3, pauses became shorter.
A great change is seen in the pause:speech ratio in two of the groups.Total pause length accounts for22.3% of total speech in Group A,.16.3 % in Group B,and23.9%
in Group C. Compared with the initial recording, this shows a remarkable increase
in Groups A and C.
Figure11shows pitch before teaching and after teaching. The bars show mean pitch before teaching, and the line graphs each treatment groupʼ s mean pitch after teaching.
In Group A, which was taught about only pauses, pitch at the beginning of new topics(0,3,7and13)became higher.(Pitch at 13was only1Hz lower than before teaching.)However,pitch range did not widen by much,and was still within 60Hz.
Group B,taught about only pitch,made extra higher pitch resets at the beginning of new topics than before teaching. The pitch resets were from 250Hz to310Hz, but the range was within about60Hz, similar to Group A.
Topic resets also became higher in Group C.Especially high topic resets were seen at Sentences3,342Hz, and13,332Hz. The big difference between this group and others is the difference between topic resets and other pitch resets. There is about
100Hz difference between topic resets and other pitch resets, which are roughly
between340Hz and240Hz.
Figure12shows mean pause durations in Text2 (normalized to a total pause length of10seconds,which was the ENSʼs approximate mean )for the ENSs and for the Japanese learners after teaching.At the beginnings of new paragraphs (5,7,and
15), Groups A and C tended to produce longer pauses than the ENSs. Group B, however, produced shorter ones.
With Text1,the pause:speech ratio of16.3% in Group B was clearly lower than the other groupsʼ. This pattern was also seen with Text 2. Group Bʼs total pause duration was11.02seconds, the shortest of all the groups, and total speech 63.76
seconds,for a ration of16.9%.In contrast,Group Aʼs pause duration accounted for 21.0% of total speech, and Group Bʼs22.6%.
Figure13shows the mean pitch produced by Japanese learners in Text2.Compar- ed to before teaching,pitch ranges are clearly wider,although Group A had a much narrower pitch range than the others. Pitch resets by Group A were within 67Hz,
and those by Groups B and C within about85to90Hz.In Groups B and C,these pitch resets are distinguishable from other pitch resets, but are now much higher than ENSʼs topic resets.
4．CONCLUSIONS 4．1 FINDINGS
In this study, Japanese learners were asked to record readings of two different texts. First, they read Text1without teaching. In that reading, their pauses were much shorter than the ENSʼs.Even before new paragraphs,no pauses were over 1.0
seconds in contrast to the ENSs.Pause duration accounted for only19.3% the total speech time compared with23.4% in the ENS data.As for pitch,Japanese learnersʼ
pitch range was relatively flat before teaching, and there were no remarkable distinctions between topic resets and other pitch resets.
The three teaching treatments had different effects.This study showed that pause teaching has an effect on intonation learning as well as pitch teaching.Both Group A and C, which are taught about pauses, took longer pause when the paragraph /
topic changed. Moreover, total pause length also was longer after teaching than
before teaching,when it was extremely short compared to English native speakersʼ. Learnersʼpitch resets are not strongly influenced by pause teaching,but their topic resets appear to be. Pitch teaching also affected pause features, leading to longer pauses before new paragraphs. However, total pause length remained significantly shorter than those of native speakersʼ. Giving information about both pause and pitch had a remarkable influence on pause duration,pitch reset,topic reset,and pitch range. This group only produced a large pitch range after teaching.
These teaching effects transferred to the reading of a new,unseen text.However, it is likely that learners tend to focus on what they learned too much when reading new texts.
All these findings considered, teaching pause and teaching pitch have a great influence on intonation and prosody learning, and they are related each other.
Teaching pause and pitch by themselves has an effect, but teaching them together was the most effective for learning intonation and prosody.
Before the recording,I asked the subjects about their English learning experience.
A majority,60% of them,answered that they had focused on grammar.A minority, 30%, answered that had learned pronunciation, and only7%, one person, answered that she had learned about intonation.At the same time,most of them, 67%,believe that the most important thing for speaking English was pronunciation. A small number,13%, mentioned grammar, and another 13% intonation. In other words, it appears that, while they are aware of pronunciation, few grasp the importance of intonation and prosody. This may be a result of intonation and prosody not being taught in schools.
These kinds of belief might influence Japanese learnersʼlearning of intonation and prosody this study. Second language learners, particularly older learners, have strong beliefs and opinions on language learning and assume that a particular type or content of instruction is the best for them (Lightbown & Spada1999).From this point of view,learnersʼbeliefs and opinions might need to be influenced to encourage them to focus on intonation and prosody.
4．2 NEED FOR FUTURE RESEARCH
The subjects(Japanese learners )were asked to record themselves reading a new text soon after a teaching session and recording the teaching text.This means that information about pauses and pitch was fresh in their minds when they read the new text. A useful follow up to this study would be to investigate the long term effects of this kind of teaching. In other words, a study with a delayed post-test is needed now.
In addition,the effectiveness of visual feedback still needs further investigation.In this study, visual feedback appeared to as an effective teaching resource, as in previous research. However, how much it affected intonation learning is still not clear.
REFERENCES
Bradford,B.(1998).Intonation in context teacherʼs book.Cambridge:Cambridge University Press.
Hill,L.A.(1967).Intermediate stories for reproduction (Third impression).London:Oxford university press.
Lightbown, P & Spada, A.(1999).How languages are learned Oxford University Press.
Mayer,J.,Jasinskaja,K.& Kolsch,U.(2006).Pitch Range and Pause Duration as Markers of Discourse Hierarchy: Perception Experiments. Paper presented at Interspeech 2006＜http://www.interspeech 2006.org/＞, Pittsburgh, PA.
Retrieved June24,2007, from www.sfb632.uni-potsdam.de/publications/
A3/A3Mayer2006a.pdf
Ministry Education and Science(2002) .International Exchange and Cooperation.Retrieved July10,2007,from http://www.mext.go.jp /english/org/exchange/66a.htm Munakata, Y.(2003). Prosody of Japanese EFL learnersʼspeech: Teaching intonation.
Tokyo Womanʼs Christian University: Studies in Language and Culture,14, 68‑92.
Okuda,N.(1975).Eigo no intoneshion: kenkyu to shidou.［English Intonation: research and pedagogy］Tokyo:Eiwa Shuppan.
SIL International(2005).Speech analyzer.［Computer Software］Retrieved June,2007,from http://www.sil.org/computing/sa/index.htm
Spaai, G. W. G. & Hermes, D. J. (1993). A Visual display for the teaching of intonation.
CALICO Journal,10,19‑30.
Sugito,M.(1997).Akusento, Intoneshion, rizumu to pozu.［Accent,Intonation,rhythm,and pause.］Tokyo:Sanseido.
Wichmann, A.(2000).Intonation in text and discourse: Beginning, middles and ends. Harlow, Essex:Longman.
要旨
近年，日本では，文部科学省の推奨により，学校での英語学習でも発話能力に注目するように なってきている。しかし，イントネーションが学習過程で触れられることは少ない。理由として，
イントネーション教授の難儀性が考えられる。一方，イントネーション教授の試みもなされ，可 視化されたデータによるフィードバックの効果が示されている。
本論では，ポーズとピッチに注目したイントネーション教授法の効果を探った。ピッチに注目 した教授法がイントネーション学習に効果があるという先行研究を元に，ポーズという要素を加 えた研究を行った。教授にあたり，英語母語話者の音読分析の結果とその音調曲線が用いられ，
学習者はそれらと自ら発話した音調曲線の比較を視覚的に行った。ポーズの教授，ピッチの教授，
ポーズとピッチの教授の３種類の教授を行い，それぞれが学習者にもたらす影響を検証した。ま た，そこでの学習が他の文章の音読においても影響するのか調査した。
その結果，学習後のポーズとピッチに変化が見られ，それぞれの教授方法に効果が見受けられ た。また，ピッチのみの教授ではポーズに影響が少なかったものの，ポーズに関する教授は，学 習者のピッチにもある程度影響があることが分かった。教授方法としては，ポーズとピッチの両 方の教授に最も効果が見受けられた。そして，類似する結果が，他の文章の音読においても確認 された。
