Using long − jump exercise as a medium , we examined differences in teachers ' verbal interaction by comparing classes With high scores ( abbr . HSC group) with those showing low scores ( abbr . LSC group) . Using Kobayashi ' s Physical Education Inventory , teachers ' attitudes were measured twice for fifth and sixth graders , constitUtilg seven classes in various elementary shools , in September and November 1996. After the first measurement , aU of the teachers ' spoken words and actions

more »

... recorded by VTR in four lessons ( 2, 5, 8and ll sessions )as the core of the fourth teaching stage each of the seven classes . The teaching plan in aU the classes was the same . All elelnents of the teacher ' s speech and behavior were classi 丘ed into parts of speech as minimum element sentences , and their frequencies were counted in each 45 ・ minute session ( abbr , etc . ) , Using results obtained by Takahashi et al. and Urneno et al. , the other method employed the fre − quencies of typi 〔 ral parts of speech that were connected with the teacher ' s behaVior used for learning development( attitude scores ) . Namely , the words of Meishi were classified into five types ( spacing , timing ...and physical parts of words , technical terms related to long − jump ...and personal names ) . The words of Keiyoshi were divided into two types( positive − negative words and binomina 皿 y oppositional words ) . The words of FUkushi were also diVided into two types ( interrogative words and spacing or timing words ) , The results obtained were as follows : 1) The frequencies of ordinary parts of speech in the HSC group had significantly higher values ( P く 5% )than 廿lose in the LSC group for eight parts of speech except Kandoshi and Keiyodoshi . The above differences were especially significant ( P < 1%)for Fukushi and Zyoshi words . 2) For the typical parts of speech , the frequencies of Meishi ( technical tenns and timing words )and Fukushi ( interrogative words and spacing or ti 皿 i 皿 g words )in the HSC group had sig ・ nificantly higher values ( P < 1%)than those in the LSC group . The frequency of Keiyoshi ( binomina1 − ly oppositional words )also showed a similar tendency( P く 5%) . It was considered that these words