STARN: Scots Teaching and Resource Network

Back to contents page

Gender, Speech Styles and the Assessment of Discussion

Chapter 4 - Data Analysis I: The Classroom Discussions

4.3 Overview of Discussions

In this section, I review the results of the seven analyses in terms of the distribution of certain linguistic features, and the significance of these distributions in the assessment process. The aspects of the discussions dealt with are the eight features considered in the discussion analyses: floor apportionment, back channel support, questions and tag questions, epistemic modality and hedging, topic development, lexical repetition, interruption and overlap, and simultaneous speech.

4.3.1 Floor Apportionment

This section deals with the distribution of the floor among members of each group. I have two aims in comparing the distribution of the floor across the seven discussions. Firstly, I consider whether the distribution was different depending on whether groups were mixed-sex groups, single-sex girls' groups, or single-sex boys' groups. Secondly, I consider the extent to which the distribution of the floor influences the assessment process. The distribution of the floor in mixed-sex groups

Groups I, III, and VI were mixed. The pattern which the literature suggests is most likely to occur in the context of mixed-sex group discussion is that in general the boys dominate the floor. This did not emerge from my data. In Group I, the three girls, G1, G2, and G3 all spoke considerably more than the two boys, B1 and B2, one of whom (B2) did not speak once during the discussion. In Group III, two of the boys (B6 and B7) spoke very little, while one boy (B5) spoke considerably more than any other group member. The only girl in the group (G4) spoke less than the most vocal boy, but still considerably more than either of the other two boys. She was the chair of the discussion, which may have been a factor affecting her contribution rate positively. Group VI consisted of two girls (G11 and G12) and two boys (B8 and B9). Although G12 spoke least in the group, G11 spoke far more than either of the two boys.

Thus no pattern emerged from this data to suggest that gender was a significant factor in determining floor distribution in these mixed-sex groups. The distribution of the floor in single-sex groups

There were three single-sex girls' groups (Groups IV, V, and VII) in my sample, and one single-sex boys' group (Group II). From the literature, the typical structure of all- girls' groups appears to be non- hierarchical, and therefore might be expected to have more-or-less equal participation from all group members, while boys' groups are more likely to establish a hierarchy in which the most dominant boy talks the most, and his subordinates talk less. Again , these patterns were not confirmed by my data.

There was only one all-boys' group, Group II, in my sample, so no great claims can be made on the basis of the boys concerned. The two participants, B3 and B4, both contributed relatively little. While B3 did talk more than B4, there were no signs of either boy competing for the floor as neither seemed keen to talk.

In the all-girls' groups, there was one example of profoundly unequal levels of contribution, and two of fairly equal levels of contribution. In Group IV, the four group members (G5, G6, G7, G8) spoke to vastly differing degrees. G5 and G6 spoke considerably more than G7 and G8; G8 hardly spoke at all. This group appeared to be hierarchically structured in this respect (and in other respects). G9 and G10 in Group V shared the floor with greater equality, although G9 did talk more than G10. In Group VII, G13, G14, and G15 shared the floor with a greater degree of equality than any of the other groups. The differences between the relative amounts of time speakers in these two groups held the floor were small, compared with other groups in the study.

There was one striking case of equal floor distribution amongst girls in Group VII. However, Group IV provided a strong counter-example, in which there was very unequal distribution of the floor amongst four girls, and the other discussions, mixed- and single-sex, did not provide evidence for boys competing for the floor to a greater extent than girls, or girls being more concerned with the equal distribution of the floor than boys. The influence of verbosity on the grades awarded by the S.E.B.

From looking at the six discussions on the Young People Talking film, it became apparent that the more they talked, the higher the grade awarded to a pupil. The table below shows the average number of words per minute of discussion uttered by each pupil assessed (the number of words each pupil spoke is given as words per minute rather than as a total for the whole discussion because all the discussions were different lengths, so the absolute numbers of words spoken were not comparable). The table also shows the average length in words of the pupils' turns, and the level and grade they were awarded in the S.E.B.' documentation accompanying the film.

speakeraverage no. words/1 minuteaverage length of turnGrade / level awarded
B1 7.48 19 6/ Foundation
B3 33.43 10.6 5 / Foundation
B5 97.82 13.5 4 / General
G5 76.47 13. 3 / General
G10 71.61 27.75 2 / Credit
G11 76.6 63.8 1 / Credit

As can be seen from the chart above, in the cases of the pupils assessed by the S.E.B., the two pupils assessed as foundation level, B1 and B3, speak markedly fewer words per minute than the pupils assessed as general and credit levels (7.48 and 33.43 words per minute respectively). The three girls G5, G9, and G11, who are awarded grades 3, 2, and 1 respectively, speak considerably more (76.47, 71.61, and 76.6 words per minute respectively). The only pupil whose average number of words per minute does not fit into this pattern is B5, who speaks considerably more words per minute than any other assessed pupil (97.82), but is awarded a relatively low grade (a 4 at general level). B5 is a markedly competitive speaker, as demonstrated by his verbosity and in other aspects of the analysis, and he is explicitly penalised for this by the S.E.B..

If in addition to the number of words spoken per minute, the average length of speakers' turns is taken into account, the correlation between quantity of speech and grade becomes more emphatic. The speaker with the longest average turn is awarded the highest grade: G11, who is graded as 1 at credit level, has an average turn length of 63.8 words. The speaker with the second longest average turn length is awarded the second highest grade: G9, awarded a 2 at credit level, has an average turn length of 27.75 words. G5 and B5, both assessed as general level, have similar average turn lengths: 13 words and 13.5 words respectively. B3, who is graded the second lowest, at grade 5 foundation level, has the lowest average turn length at 10.6 words. The gap between B3's average length of turn and that of G5 and B5 is not very great, but the number of words B3 says per minute is much lower than the number said by G5 and B5.

Thus, when a profile is constructed taking into account the average number of words uttered by the pupil per minute, and the average length of the pupil's turns, there is a strong correlation between quantity of talk and the grades awarded by the S.E.B.

4.3.2 Back Channel Support

This aspect of linguistic behaviour is not explicit in the GRC, not even as 'listening' or 'being supportive', although 'encouraging others to have their say' might be invoked if the assessor wanted to acknowledge a pupil's skilful use of back channel support. I have argued that in more co-operative discussions, speakers use a greater proportion of back channel support items. The ratio of back channel support items to words uttered for the seven discussions is shown below.

GroupRatio of back channel responses to words uttered
Group I 1:50
Group II 1:26
Group III 1:68
Group IV 1:69
Group V 1:97
Group VI 1:82

These results show very variable rates of back channel support between the seven discussions. Group VII and Group II stand out as having at least twice as many items of back channel support as the others discussions, although in the case of Group II, B4 uses considerably more back channel support than B3, and appears to do so rather than have to make his own contributions, and both boys speak very little; thus the apparently high frequency of back channel response may be a consequence not of co- operativeness but reluctance to talk in their case. The lowest comparative rate of back channel support is in Group V between G9 and G10, with Group VI the second lowest. These figures bear out the overall profiles of these groups as predominantly non-co-operative and competitive. The back channel support in Group VII between G13, G14, and G15 is four times as frequent as in Group V, bearing out the overall profile for this group as co-operative. Groups I and IV were both analysed as having participants who were selectively co-operative towards certain group members, and non-co-operative and competitive in their treatment of other group members. The ratios for back channel support for these two groups place them between the most co- operative discussion and the more non-co-operative / competitive groups. The only discussion which does not correspond to this trend of the more co- operative the group, the higher the ratio of back channel items is Group III, which was decidedly competitive in most respects, but has a ratio for back channel support almost identical to that of Group IV.

Another important aspect of back channel support which has already been considered briefly in the group profiles is whether there is a causal link between receiving the greatest number of back channel support items and speaking the most in a group. Do powerful speakers receive more back channel support, or do the speakers who receive more back channel support dominate the discussion?

In all discussions in the study, the speaker who talked most received most back channel support (although this was only just the case in Group IV, where G5 and G6 speak to similar extents, and Group V, where the difference in back channel items received by G9 and G10 is only one). In many of the groups, the person who speaks least is given the least back channel support; this is true of Groups I, II, III (B6 and B7 both receive only one item of back channel support) IV, V, and VII (in Group VII, G14 receives just one less item than G15).

There also appears to be a correlation between talking less and giving more back channel support. This shows up in Group I, where G1 talks least of the three girls, but gives the most back channel support. It also shows up in Group II, in Group III, where B7 talks least but gives five items of back channel support (although G4 actually gives the most back channel support in that discussion: six items), in Group V, and in Group VI, where B9 gives the most items of back channel support while talking less than anyone except G12.

Finally, the speakers who speak most frequently produce relatively little of the back channel support in the discussion. This is the case in Groups I, II, III, and VI.

These figures suggest that in many of the groups in the sample, a hierarchy operates, according to which the person who speaks most in a group is supported most; the person who speaks least is supported least; those who speak most frequently do not support others very much, and those who speak little are frequently very supportive of other speakers. These trends show up strongly in groups I, II, and III, and in the other groups to a lesser degree.

4.3.3 Questions and Tag Questions

The features of questions and tag questions proved particularly thorny to analyse, as the range of functions realised in these forms was exceedingly varied, and generalisations could not be made, based on such disparate data. Questions were used to invite the opinions of others, to challenge others' ideas, to pass the responsibility of speaking on to someone else, to gain access to the discussion, and rhetorically, to retain the floor. Tag questions were used to hedge areas of doubt or possible contention, to coerce others into agreeing with the speaker, or to elicit their support.

There were no striking differences in the ways girls and boys used these forms, co-operative and competitive functions being demonstrated by pupils of both genders. It is worthy of note however, that B1 used questions as a means of joining in Group 1, which is a use of questions associated with female speakers rather than male in the literature. As mentioned in the analysis of Group I, the relative status of the speaker appears to be more significant than gender in this respect.

4.3.4 Epistemic Modality and Hedging

Epistemic modality and hedging are not recognised explicitly by the GRC, although a teacher / assessor might acknowledge them under 'allowing / encouraging others to have their say' or 'using language suited to the listener(s)' .

These forms appeared to varying degrees in the data, and again are used by speakers of both genders. B1, a tentative speaker, hedges to a considerable extent when he speaks. The two boys in Group II use hedges, apparently to signal lack of commitment to their propositions. In groups I and VII, the girls use hedges and epistemic modal forms where there is doubt or potential conflict, and not otherwise. In Group IV, the use of hedging appears to be linked to relative status of the speaker within the hierarchy: G7, the most subordinate speaker, uses the greatest number in proportion to how much she speaks, while G5, the most dominant speaker, uses very few hedges. Very few hedges and epistemic modals appear in Group VI, and those which do appear not to be triggered by the need to negotiate potential conflict, but more as purely epistemic forms, to indicate degrees of doubt. Almost no hedges appear in Group III's discussion, where conflict is explicit, and assertions are frequently aggravated. The same is true to a lesser degree of Group V, where G10 uses hedges to mitigate her propositions and her dissent from G9's suggestions, but G9 makes a number of aggravated assertions.

The use of epistemic modality and hedging appears to be consistent with other aspects of the groups' behaviour regarding co-operative and competitive talk.

4.3.5 Topic Development

Topic development in my data operates under the constraints of the set tasks. To an extent abrupt shifts of topic are built into the activity of classroom discussion. Pupils have a list of questions or other brief to complete, and these will to a large extent determine the development of topic in a discussion. If one pupil in the group is told by the teacher to chair the discussion, that pupil may take responsibility for moving the group through a list of set questions, thus influencing the pattern of topic.

Nevertheless, there were differences between groups in how they dealt with topic shift. Group III was competitive in this respect, with sudden topic changes being introduced by G4, the chairperson and B6, in an attempt to wrest the floor from B5. In other groups, topic shift was by consensus only, in Group VII, between the girls in Group I, and between G5, G6, and G7 in Group IV. Thus there were marked differences between competitive and co-operative topic shift even within the restrictions of the classroom task.

The construction of joint turns seemed to be particularly sensitive to speakers' relationships with one another. While in some groups, joint turn construction was very frequent, in others it did not occur at all. Joint turn construction occurred less between boys than girls, and not at all in the most competitive group (III) and non- co-operative groups (V and VI). There was one instance between the two boys in Group II, and many between the girls in Group I, G5, G6, and G7 in Group IV, and in Group VII. However, even in Groups I and IV where joint turn construction was frequent, speakers appeared to use it selectively, constructing turns jointly with one speaker and not with another. Only in Group VII did all each of the three group members construct turns jointly with the other two.

4.3.6 Lexical Repetition

This is only recognized as a foundation level skill by the GRC. However, I have tried to demonstrate in these analyses that lexical repetition can be a sophisticated and effective way of supporting another speaker and developing a discussion.

Again this was a form used co-operatively more widely by girls than boys in this data. Girls in Groups I, IV, and VII used lexical repetition regularly to develop ideas and give feedback.

The form was also used competitively in other groups, to challenge other speakers. This occurred in Groups III, V, and VI. Both boys and girls used lexical repetition to disagree with a previous speaker.

4.3.7 Interruption and Overlap

Interruptions are not frequent in the data except in Group III, where B5 interrupts to a considerable extent, and other speakers also interrupt him, in order to be able to speak at all. Otherwise only occasional interruptions occur - G2 interrupts B1 in Group I; G8 interrupts G7 in Group IV; G10 interrupts G9 in Group V. Interruptions are not explicitly named as behaviour to be penalised by the GRC, but it is specified that pupils should 'allow/ encourage others to have their say'. The pupils interviewed and cited in Chapter Two gave one of their criteria for participating well in a discussion as 'don't interrupt'.

From this sample, it seems rare that pupils, boys or girls, do actually interrupt each other intentionally.

Overlapping speech occurs very frequently in some discussions, and not at all in others. Groups in which it occurs most frequently are I, IV, and VII, between girls in each case. Boys do not appear to use this form to anything like the same extent as girls from this data, and neither does it occur in the more competitive or non-co- operative discussions ( Groups III, V, and VI).

This seems to be an important feature about which to raise awareness, as some of the teachers cited in Chapter Five who viewed the film of Group VII's discussion, perceived G14 as interrupting the other speakers. G14's turns were treated within the discussion as supporting other speakers, not violating the rights of others to the floor, and this would be a very important distinction to make when assessing.

4.3.8 Simultaneous Speech

The co-operative feature of simultaneous speech is not recognized by the GRC. This did not occur to a great extent in any of the discussions in my data. The two occasions on which it did occur were between girls who were very co-operative in their interaction with each other, but less so with other group members: between G2 and G3 in Group I, and between G5 and G6 in Group IV. It may be that this feature is particularly sensitive to context, and therefore is less likely to appear in the environment of classroom discussion.