Thinking Together in the UK and Mexico: a comparative study

Rupert Wegerif, Julieta Perez Linares, Sylvia Rojas Drummond, Neil Mercer, Maricela Velez.

Abstract

The Thinking Together educational programme was first developed in the UK to promote the use of exploratory talk in primary classrooms. It emerged out of a strong tradition of group work and the promotion of oracy in the UK curriculum. The approach was then adapted and applied in te very different context of Mexican state primary education. In this paper we compare the programme in Mexico with the programme in the UK. We explore the similarities and the differences in the implementation of the programme and in the research findings. This comparison leads to three main conclusions:

· that both quantitative results and the results of discourse analysis were remarkably similar in the two studies despite the marked differences in social and cultural context;

· that the relationship between teacher's normal practice and the proposed intervention was closer in the UK than in Mexico and this impacted on the implementation and the sustainability of the project.

· that there are some signs that the programme may have empowered girls in mixed gender group work in Mexico in a way that was not noted in the UK

Key words: Comparative study, primary, thinking skills, Mexico, discourse, oracy

Introduction

Increasingly educational approaches that have been developed in one country are being applied in other countries.  In this paper we present the results of one such transfer of an educational innovation. Through this we also investigate some of the factors that affect the transfer of educational innovations to new cultural contexts in general. Our focus is the Thinking Together educational programme. This has strong roots in educational traditions in the UK that go back to the nineteen sixties. It involves promoting a type of talk considered to be effective for thinking and learning that was first described by Douglas Barnes in the nineteen seventies and called exploratory talk (Barnes, 1976: Mercer 1995). Group work and oracy have been a part of the UK primary curriculum since the influential Plowden report of 1967. The Thinking Together programme was first developed in the UK to promote the use of exploratory talk as an integral part of teaching and learning within primary classrooms. The Mexican educational context is very difference. Historically there has been little interest in either group work or oracy. This is reflected both in the curriculum and in the physical layout of classrooms which all have desks in straight rows facing the front. To apply the Thinking Together programme to the Mexican context was therefore something of a challenge. In this paper we describe how the programme needed to be adapted to fit the Mexican context. We also explore similarities and differences in the impact of the programme. However before we can compare and contrast implementations of this educational programme we need to say a little about it.

Principals of Thinking Together approach

Behind the Thinking Together approach lies the educational theory of the Russian psychologist Vygotsky, that an important way in which children learn to think individually is through first learning to reason with others in dialogues. Exploratory talk is educationally effective talk in which participants pool ideas, opinions and information, thinking together aloud to create meanings, knowledge and understanding. Achieving exploratory talk depends on the willingness of all participants to adhere to some basic rules, or ‘ground rules’. In classroom situations, ground rules for talk can be created and agreed by the class.  These ground rules are then applied by groups talking and solving problems together in different areas of the curriculum. Collaborative learning supported by computers is one context for the use of exploratory talk within the curriculum. Each classroom develops its own set of ground rules but these are variations of the main ground rules suggested by the developers of the approach. These are that:

· All relevant information is shared openly.

· Each group member should be actively encouraged to contribute to the discussion.

· Everyone should listen to others attentively.

· Each suggestion should be carefully considered.

· Group members are asked to provide reasons for ideas and opinions.

· Constructive challenges to ideas are accepted and a response is expected.

· Alternatives are discussed before a decision is taken.

· The group works together with the purpose of reaching agreement.

The group, not the individual, takes responsibility for decisions made, for success achieved or for problems that may occur. (Dawes, Mercer and Wegerif, 2000)

Working closely with primary teachers, the UK research team produced a series of 'Talk Lessons' to teach these ground rules and to apply them within normal curriculum teaching and learning. The main focus is on developing children’s use of language as a tool for reasoning and constructing knowledge. The Talk Lessons encourage teachers to create a 'community of enquiry' in their classrooms in which children are guided in their use of language as a tool for both individual reasoning and collaborative problem-solving.

Background to Thinking Together in the UK and in Mexico

There have been several experimental implementations and evaluations of the Thinking Together approach in the UK. The first systematically evaluated and published study was part of the doctoral research of Dr Rupert Wegerif who was supervised by Professor Neil Mercer and worked in collaboration with teacher/researcher, Lyn Dawes[1] (Wegerif, 1996: Wegerif and Mercer 1997: Wegerif and Dawes, 1997). This work served as a pilot study for the larger UK government (ESRC) funded study called the Talk, Reasoning and Computers (TRAC) project which ran from 1996 to 1998. Since the end of that project there have been several more funded studies, RATTLS, 1998 to 2000 (funded by Milton Keynes Local Education Authority), Thinking Together in Maths and Science at Key Stage 2, 2000 to 2002 (funded by The Nuffield Foundation), Thinking Together at Key Stage 1, 2002 to 2004 (funded by Esmee Fairbairn) and Thinking Together at Key Stage 3, 2002 to 2004 (funded by Milton Keynes Local Education Authority). In addition to these evaluated projects the Thinking Together approach has been brought into classrooms through a book of classroom materials (Dawes, Mercer and Wegerif, 2000), numerous talks to educationalists and guidelines to teachers written for the government authority responsible for the UK curriculum (e.g QCA, 2002).

In this paper we will focus on the first government funded study, the TRAC project, but we will also look at some data from the very similar, if smaller, doctoral study by Rupert Wegerif. This is necessary because the Mexican implementation of the Thinking Together project used the same reasoning tests as Wegerif (reported in Wegerif, 1996) rather than those used in the TRAC project (reported in Mercer, Wegerif and Dawes, 1998). This means that, while we can compare the implementation of the Mexican study and that of TRAC, and we can compare the qualitative results of the Mexican study with those of TRAC, we cannot directly compare the quantitative reasoning test results. Wegerif’s doctoral study used only one target class and one control class while the TRAC project used three target and three control classes all in different schools. In all other respects essentially the same methods were used therefore it is reasonable to include the pilot with the main study.

In 1995 the British Council agreed to support a research link between Professor Neil Mercer in the UK and Professor Sylvia Rojas-Drummond in the psychology faculty of UNAM in Mexico City. As part of this link Sylvia Rojas-Drummond, and her team, decided to adapt the Thinking Together approach in the UK and apply it in Mexico. To support this the Thinking Together book of materials (Dawes, Mercer and Wegerif, 2000) was translated into Spanish and adapted to the cultural and educational context.

Key details of the two implementations are given below. These are taken from the TRAC study in the UK (Mercer, Wegerif and Dawes, 1999: Wegerif, Mercer and Dawes, 1999) the doctoral pilot study in the UK (Wegerif, 1996) and the study in Mexico (Rojas-Drummond , Perez, Velez Gomez and Mendoza, 2002)

Comparing the implementation of Thinking Together in the UK and Mexico

Design

A: In the UK

The study was designed as a field experiment. Each of three experimental classes, each in a different school, was matched with a control class of the same age group in another local state school. There were pre-tests, an intervention lasting 12 weeks in total and then post-tests. The control classes had the same observations and tests as the experimental classes but no intervention programme.

The pilot study had the same design but it was smaller with only one experimental class matched with one control.

B: In Mexico

The experimental design was similar to the UK study. The intervention lasted five months (22 weeks) and both the experimental classes were in one school.

Participants

A: In the UK

The TRAC study involved about 180 children in total, aged 9 and 10 years old (year 5), in three state middle schools in and around the city of Milton Keynes. This is an area of high mobility. Of the original 90 in the experimental classes there were only 60 children remaining at the post-test. The three control classes had about 90 children and, in a similar way, this reduced to 64.

For the purpose of the tests and many of the exercises in the intervention programme the children were required to work together in mixed gender groups of three (plus one or two groups of two if numbers did not divide by three). These groups were organised by the class teacher, so as to include a range of ability in each. There were in total 23 target groups and 25 control groups in the TRAC study. In each class one group was selected as representative of the class by the class teacher. This group was video-recorded.

In the pilot study there were 33 children in the experimental class and 18 in the control class. They worked together in 9 groups of three in the experimental class and 6 groups of three in the control class. Three groups, selected by the teacher as representative of the class, were video-recorded in the target class and in the control class.

B: In Mexico

In Mexico 84 children were involved in the study. These were from 10 to 12 years old (fifth and sixth grades), in two public primary schools in Mexico City, Mexico. The schools were nearby and equivalent in socio-economic status, corresponding to low to middle-class. One school was randomly assigned to an experimental and the other to a control condition (with 42 children in each). For one of the tests (see below) the class was divided into groups of three. As in the UK these were mixed gender and mixed ability groups. In each condition three groups of three were selected as representative of the class and video-recorded. (One of the focal groups in the target classes turned into a group of three girls because the boy who was supposed to be in the group left the school just as the project was starting up)

Intervention

A: In the UK

The TRAC intervention programme is described fully in a book for teachers that was written out of the project, Dawes, Mercer and Wegerif (2000). In the project nine lessons were delivered by teachers each designed to last for about one hour and to focus on one or more of the ground rules of exploratory talk which were outlined earlier. The first few lessons deal with skills such as listening, sharing information and co-operating, while later lessons encourage children to make critical arguments for and against different cases..

The use of the ground rules was taught to the children through explicit modelling by the teacher, coaching their use in whole group and small group discussions and giving opportunities for their use by the children working in small groups without the teacher. The explicit modelling phase involved the teacher at the front of the class illustrating the ways in which she (The teachers in the UK were all women) wanted the children to talk together. Asking ‘why?’, using ‘because’ to give reasons for statements, asking other children what they think, reaching agreement before making a final decision. A key lesson in the programme involved eliciting the ground rules for the children in their own words. This was the third lesson after the children had all had some practice in collaborative activities. In a guided discussion the teacher drew from the class the kind of rules that they think should be used in group work. The list that resulted was then put on the wall in large letters. Although each of the three classes studied produced a different set of ground rules they were all similar to the ground rules for exploratory talk A key principal of the programme was for the teachers to use these ground rules as much as possible in all their teaching with the class. 

B: In Mexico

A series of educational materials and procedures were designed to create a training program to encourage children to use exploratory talk. The core program was developed by adapting the program “Thinking Together” designed by Dawes, Mercer and Wegerif (2000). The programme included extra materials, games, texts, objects to carry out experiments, activity cards, answer sheets and some software. This programme was delivered in one lesson a week over a period of five months. On the whole the lessons were delivered by members of the research team working with the class teachers.

Setting

A: In the UK

All teaching was done in normal classrooms by normal class teachers. These teachers had one day of training provided by the university research team.

B: In Mexico

For each school, the study took place in a multiple-purpose room. In the experimental school, this room was adapted to suit the needs of the intervention program. The room was fitted with modular furniture and equipment, including computers.

Tests

A: In the UK

Members of both target and control classes were given Raven's Coloured Progressive Matrices (CPM), which is made up of similar non-verbal problems, as individuals both before and after the intervention programme.

In addition target and control classes were divided into groups of 3, and in those groups the children attempted to solve the full Raven's Standard Progressive Matrices. They were given this test before the intervention and then again after the intervention had been completed in the target school.

Raven's Progressive Matrices (RPM) consists of graphical puzzles. They are widely used in education and psychology as a test of 'non-verbal' reasoning. Raven's tests are particularly appropriate for exploring the link between language practices and the non-culturally based tradition of research in cognitive development, because they correlate well with other similar tests of reasoning and with measures of academic achievement (Raven, Court and Raven, 1995, Richardson, 1991, p 129).

The UK team used two similar Raven's tests, the Standard Progressive Matrices (SPM), consisting of 60 problems, and the Coloured Progressive Matrices (CPM) consisting of 36 similar problems. The CPM is appropriate for younger children (up to the age of 12) while the SPM is appropriate for any age. Test scores from CPM and SPM can be translated into a common scale through the use of a table provided (Raven, Court and Raven, 1995, p 64). The SPM were used for children working together in groups of three, giving each group a single book and a single answer sheet and encouraging them to talk together in solving the problems. The same children were given the CPM exactly three days later, asking them to work individually and following the guidelines for the administration of the test given in the manual (Raven, Court and Raven, 1995).

Both of these tests proved problematic for different reasons. The CPM test is designed for young children and those with learning difficulties. It proved a little too easy for some children and a 'ceiling effect' was noted. Essentially this means that they did so well in the pre-test that there was not much room for improvement in the post-test. The SPM test took too long for the groups to complete using exploratory talk. The ground rules of exploratory talk involve asking questions, considering alternatives and reasoning about every problem. After the thinking together programme groups take longer to complete any task because they talk more about it. This meant that the target children in the post test became bored of using exploratory talk for all the sixty problems of the SPM and tended to switch back to cumulative talk just to save time.

The pilot study (Wegerif, 1996) used an adapted version of the Raven’s test described below in the section on the Mexican study. The English researchers recommended a reversion to this shorter simpler constructed test after the problems encountered with the full CPM and SPM.

B: In Mexico

All participants were administered an adapted version of the Raven’s Test of Progressive Matrices. The adaptation consisted of dividing the original into two tests of equal difficulty, and using one as an individual test and the other one as a small-group test (Wegerif, 1996). Each test consisted of a booklet and an answer sheet.

The original test consists of 60 logical-perceptual problems where subjects select one out of six to eight figures to complete a progressive matrix. Problems are divided into five scales from A to E, which increase in degree of difficulty. From this original version, two parallel, half-sized formats were prepared by assigning each successive problem to either format 1 or 2. Each parallel format was used as either a pre- or post-test. At the same time, during each testing period both versions of the test were administered, one version as an individual test and later the other as a small-group test (in triads).

Discourse analysis

A: In the UK

In each of the three target classes, one 'focal group' of children was video-recorded as they worked on the Raven's SPM test, before and after the intervention.

The video-tapes of these focal groups were used for qualitative discourse analysis of the kind developed and described by the present authors and associates (Edwards & Mercer, 1987; Mercer & Fisher, 1993; Mercer, 1996b; Wegerif & Mercer 1997b). In addition transcripts of the talk of the children were made and a computerised text analysis concordancer was used to search for pre- and post-intervention differences. This method (described in more detail in Wegerif and Mercer, 1997b) involves integrating qualitative analysis of the full transcript with the abstraction of 'key words in context' in order to generalise significant features and compare different transcripts. Although essentially qualitative, this approach also facilitates linking qualitative evaluation to quantitative descriptions of texts. For example detailed qualitative analysis of the deliberations of all the groups revealed that the word ‘think’, as in ‘I think’, was being used to put forward reasons. This word was then included in a search for key features in the talk of the focal groups which indicated exploratory talk and so could be used to provide a quantitative measure for comparison with other transcripts.

B: In Mexico

The Mexican team adopted a different approach coding the talk elicited by each problem. The coding was based on types of talk analysis (Mercer, 1995: Wegerif and Mercer, 1997a). The three main types of talk are cumulative (sharing but uncritical talk); disputational, (unconstructive conflict) and exploratory talk (described above). When analyzing the data the Mexican team saw the need to make a further distinction between what they call Incipient Exploratory and Elaborated Exploratory Talk. The first category is for exploratory talk that is 'not very consistent nor very prominent' in the way children talk, whereas the latter indicates exploratory talk that is 'more consolidated and sophisticated' (Rojas et al, 2001).

Summary and discussion of the differences and similarities in the implementations of the two studies

The summary is presented in table 1.

Table 1: Comparing the implementation in the UK and Mexico

Factor	UK	Mexico
Age of children	9 and 10	10,11  and12
Length of intervention	9 weeks	5 months
Lessons	9 lessons integrated in curriculum	10 similar lessons in Spanish and additional activities but not integrated in curriculum.
Delivery	Class teachers	Research team with class teacher.
Setting	Normal	Special

Oracy has had a place in the UK curriculum since the Plowden Report (1967). 'Speaking and Listening' is a part of the 'English' strand in the national curriculum. There is no such equivalent in the Mexican curriculum.  It was therefore difficult to integrate the Thinking Together approach into the Mexican curriculum.

In UK primary schools group work is common. How to teach with small group work is part of normal initial teacher training and in-service training. This is not the case in Mexico. In Mexico teachers found the methods used difficult to integrate with their normal teaching.

Despite the fact that some group work is advocated in the Mexican national curriculum, talk between pupils in Mexican classrooms is still largely seen by teachers as a sign of insufficient discipline. According to research most teachers in Mexican state primary schools have a mainly transmissional view of teaching and learning (Rojas-Drummond & Alatorre, 1994; Peon M; Rojas Drummond S in press; Mercado del Collado,R 1996). The Thinking Together intervention programme is based on a social constructivist view of teaching and learning that is widely held by UK teachers and was certainly shared by the teachers involved with the project.

.The research team in the UK was based in a School of Education. The university researchers either had experience as school teachers themselves or considerable experience of working closely with school teachers as teacher trainers. These researchers therefore found it easy to establish a close working relationships with the teachers on the team who were referred to as 'teacher-researchers'. This model of research-partnership is described in Mercer (1995, chapter 6). In Mexico the research team was based in a psychology faculty and the team were mainly research students. Some of the team had experience as school teachers but most did not. Nonetheless a research-partnership was established with at least two teachers who appreciated the goals and methods of the project.

The classrooms in Mexico were physically laid out with desks in rows. This made it difficult to support group-work in the normal classroom. The research team found it necessary to equip and run a separate classroom within the school area. This classroom had tables and many resources for group work. However the physical separation of the space in which the project was run symbolised the separation of the project from the normal teaching and learning in the school.

Several of the key differences between the two implementations indicate a programme that has roots in the history and culture of education in one country, the UK, being transplanted to a very different environment where it does not have the same natural supports. In the next section we will compare the results of the evaluation of the Thinking Together programme. In the light of the differences we found when looking at the implementation it is perhaps surprising and certainly interesting that the results of the project in Mexico closely mirror those in the UK. 

Comparing results of the Thinking Together programme in the UK and in Mexico

Quantitative test results

The TRAC study employed two standard psychometric tests of non-verbal reasoning, Raven’s CPM (for individuals) and Raven’s SPM (for groups). As we mentioned earlier there were some problems associated with using these tests. Because of this experience the Mexican study reverted to the two non-standard tests used in the pilot study. As described above these were created by dividing an SPM test into two tests of equal difficulty using information provided in the Raven's manual. The tests used in the pilot study and the Mexican study were very slightly different. In the UK pilot the last three most difficult questions had been dropped off the test in response to the class teacher’s concern not to intimidate the children. While with hindsight this was an error it had very little impact as almost no children of this age range can understand the problems in the second half of the last section of the Raven's SPM. However even random guessing would have produced an average extra score of 0.375 had these questions been included. Another small point is that the score reported in Wegerif (1996) did not include the first two questions which were used to explain the test to the children, while the scores in the Mexican study did include these questions. These two factors mean that the reported UK pilot study total scores had to be adjusted by the addition of 2.37 to bring them up to equivalence with the Mexican results.

We have included the results of another recent study which used exactly the same tests as the Mexican study. This study was with over 100 children aged 9 and 10, in the UK. The Thinking Together approach was used over an entire school year.

Look up tables provided in the Ravens' manuals allowed us to convert CPM to SPM scores [explain]. SPM scores could be divided in half to bring them to equivalence with the tests scores in the Mexican study. In theory it is therefore possible to compare the scores on all the tests. This is what we have done in Tables 2 and 3.

Table 2 mean scores of the experimental triads in the UK pilot and the Mexican Study

	Pre-test	Post-test	Gain
UK TRAC	20.72	22.74	2.02
UK Pilot	17.81	22.70	4.89
UK Nuffield	20.08	23.62	3.54
Mexico	20.64	24.29	3.65

Table 3 mean scores of the experimental individuals in the UK pilot and the Mexican study

	Pre-test	Post-test	Gain
UK TRAC	19	21.5	2.5
UK Pilot	16.95	18.25	1.29
UK Nuffield	16.26	18.90	2.64
Mexico	18.12	19.44	1.32

These studies ran over different time periods. This means that the effect of normal increase in individual score over time needs to be taken into account. This can be done using tables for age-related norms in the UK and USA provided by the Ravens manual. This shows an average 4 SPM point increase between age 9.5 and 10.5 and an average 2 point increase between age 11 and 12.

Table 4 mean scores of the experimental triads in the UK pilot and the Mexican Study (adjusted for normal increase over time)

	Pre-test	Post-test	Gain
UK TRAC	20.72	22.43	1.71
UK Pilot	17.81	22.39	4.58
UK Nuffield	20.08	22.62	2.42
Mexico	20.64	23.87	3.23

Table 5 mean scores of the experimental individuals in the UK pilot and the Mexican study (adjusted for normal increase over time)

	Pre-test	Post-test	Gain
UK TRAC	19	21.19	2.19
UK Pilot	16.95	17.94	0.99
UK Nuffield	16.26	17.90	1.52
Mexico	18.12	19.02	0.90

In looking at these tables we need to recall that the original tests in the TRAC project were different from those in the other studies and look-up tables are not a fully adequate way of relating them. The adjustments made for age-related gains are also a fairly crude instrument with a margin of error. Another factor is that the Mexican children were older than the UK children (average age 11 as opposed to 9.5). According to USA norms this difference should translate as an extra score of 2.5 on each of our half SPM tests. In fact this greater score in Mexico is not so evident, but Ravens norms are often different by a few points between cultures and countries.

Initially it looks as though the Mexican study produced greater gains than the TRAC UK study for the group tests and poorer gains for the individual tests. However this appearance may be an artefact of the problems with the tests used in TRAC described above. When the Mexican study is compared with the adjusted pilot study scores the results look more comparable.

The gain in scores of the small group condition is more marked in the UK pilot but this was with one class and with the class teacher who was the main originator of the teaching programme and a researcher on the project (Lyn Dawes). The high gain in the triad condition of the pilot study may have been anomalous. This is confirmed by the results of the Nuffield funded study.

Overall the results of the Mexican study are similar to those of the UK studies and suggest that, when natural increase with age is discounted, the Thinking Together approach reliably leads to gains on reasoning tests of between 5 and 10% for individuals and between 10 to 15% for groups. This effect currently appears to be independent of the length of the intervention. This is perhaps not surprising as the initial ten lessons and ground rules taught have been the same for all the intervention studies regardless of the total length of time of the project.

Discourse analysis

In each country some triads were selected as representative of the class and video-recorded talking together around the tests. However, as described above, quite different methods were used to analyse changes in the talk of those groups. In order to compare the two sets of findings we have applied the Mexican method of analysis to some of the UK data and the UK method of analysis to the Mexican data.

An illustration of the UK method of discourse analysis applied to the Mexican data

The discourse analysis in the UK always began with an episode of talk, usually an episode in the post test compared to one in the pre-test. To illustrate taking similar approach to the Mexican data we will explore two episodes: pre-test talk around one problem in section E which the triad failed to solve followed by the same triad’s post-test talk around a very similar problem which they managed to get right. In the UK study groups did the same test before the intervention and again afterwards. In the Mexican study the groups did one half of the divided SPM Raven's test in the pre-test and the other, equally difficult, half in the post-test. This design meant that they did not encounter exactly the same problem in the post-test as in the pre but the underlying logic of question next to each other in the test was the same. This similarity can be seen by looking at the illustrations of questions E4 and E5 below.

Pre-test: Matrix E4

E4
L: Ahora tú piénsale, Georgina G: Tu estupida... L: (Ríe) G: A ver // ¿el trecx? L: El tres G: [ ¿El tres?	Luis: Now you think, Georgina Georgina: You stupid... Luis: (laughs) Georgina: Let’s see, number three Luis: Number three Georgina: Three?

Post-test E5 -

E5 / Post-test

Spanish

English

G: Aquí le quitan los puntitos y la esta // la cruz. (señalando el primer renglón) M: No pero pérate, no, no queda. L: No. Aguanta. M: No. G: Vamos viendo la secuencia: aquí tiene así, quitan la equis y los puntitos (señalando de nuevo el primer renglón). Aquí ya no, aquí (señalando). L: Aquí le quitan sólo los circulitos. G: Ajá, los circulitos // y esa parte, como la estrellita, nada más (señalando la primera columna). M: Sería este, míralo(señalando). L: Lo que le quitaron. M: Sería este, porque mira va así, así (señalando la opción 1). G: ¿Pero cómo?, si no tienen puntitos. Luis: No tiene puntitos. Quedaría nada más la cruz. G: Ajá // sí, por lo que le quitaron.

G: Here they take out the little dots and this // this cross (pointing to the first row). M: No, but wait, no, no, doesn’t fit. L: No. Hold on. M: No. G: Let’s look at the sequence: here it is like this, they take out the ‘x’ and the little dots (pointing again to the first row). Here there are no more, here (pointing). L: Here they take out only the little circles. G: Yeah, the little circles // and this part, like the little star, nothing else (pointing to the first column). M: It would be this, look at it (pointing). L: What they took out from it. M: It would be this, because look, it goes like that, like that (pointing to option number 1). G: But how?, if they don’t have little dots. L: It doesn’t have little dots. There would remain nothing else but the cross. G: Yeah // yes, because of what they took out from it.

Commentary: In the pre-test we see Luis assumes a certain leadership in telling Georgina that it is her turn to think now. He is implying that she has been taking a back seat. Georgina responds with an expletive but she does as she has been told and gives a suggestion as to the answer. There is no explicit reasoning here. Georgina points at the solution on the page. Maurice points at an alternative. Without discussion or reasoning they agree on Georgina's suggestion. Luis comments that they are thinking well. This comment suggests that he is assuming the role of group leader.

The questions in section E of Ravens SPM are the hardest in the test and cannot be solved so easily. The answer agreed by the group is, in fact, the wrong one in this case.

In the post-intervention talk around a very similar problem, question E5, we see as change in roles. Georgina spontaneously takes the lead with explicit reasoning about the problem. She acts as a kind of tutor to the others throughout, pointing them to the features of the problem that they need to focus on. Together they work out the solution. This time they get the answer right.

Our interpretation of the difference between these two episodes is informed by our knowledge of both the video tapes and our reading of the full transcriptions of the pre and post intervention talk of this group. Applying the UK method we sought features in this episode that could be used as a basis for a comparison of all the pre and post test talk of this group and, perhaps, be generalisable to the other groups in the Mexican study.

Key features

At first sight the features in the language that make the post-test talk more successful are similar to those that were found to be important in the UK (see Wegerif and Mercer, 2000). The use of long utterances made up of chains of clauses in order to provide evidence and reasons. We also see the use of ‘porque’ with explicit reasons. However when we looked further we found Spanish terms being used for problem-solving whose English equivalents were seldom used by the children in the UK. We found that 'entonces' was used in the post-intervention talk by all the groups in Mexico. The English translation 'therefore' or 'thus' was not used at all in the UK. The word 'so' was used infrequently to carry the same meaning but not often enough to be used as a key indicator of exploratory talk. The term 'pero' was also used more in Mexico than its English translation, 'but' was used in the UK.

In the analysis of the UK data of children talking around Raven’s test the measure was set at utterances which when transcribed produced 100 or more characters of ascii text. When w tried to apply this measure there were too few instances to make it a useful way of discriminating between speakers. Instead we decided upon the measure of 70 characters or more.

Table 1: analysis of indicators of exploratory talk for Group 1

	pre					post
	turns	Long turns	Entonces	Pero+	Porque+	turns	Long turns	Entonces	Pero+	Porque+
G	81	2	0	0	4	99	8	3	2	7
L	103	2	0	0	3	124	5	1	3	5
M	87	1	0	1	4	120	3	0	1	7
Totals	271	5	0	1	11	343	16	4	6	19

The method applied in the UK is flexible. It can follow up any hypothesis suggested by an interpretation of episodes. In this case it seems that the only girl in the group is being contained by the two boys in the pre-intervention and her role switches to become more dominant in the post-test. Computer analysis of the text can be used to test this hypothesis. Breaking down key features in terms of who was speaking reveals some evidence for the hypothesis of a changing role. Although the number of times Georgina speaks remains proportionately fewer than her two male partners in the post test, she has more long utterances. These long utterances contain explicit reasoning about how to solve the problem. Looking at the context of these long utterances reveals that she was indeed taking a lead in the post test in suggesting and justifying solutions adopted by the group.

The change in Georgina’s role was associated with a marked shift toward more explicit exploratory talk in the group as a whole and also towards greater success in solving the problems. The score of this group increased from 21 in the pre-test to 27 in the post test.

Focussing in on each of the 6 questions that this group got right in the post test, having failed to solve very similar problems in the pre-test, revealed that Georgina's change in role was significant to the greater success with the problems. In four of the six problems it was her explanations (using long utterances and reasoning terms such as entonces) that led the group to the solve the problem.

Each group of children can, to some extent, establish their own culture in which particular words may be used more than others. Nonetheless we applied the key indicators that we had found for the first Mexican group, GLM, to the other two groups to see if they held.

Table 2: analysis of indicators of exploratory talk for Group 2

	pre					post
	turns	Long turns	Entonces	Pero+	Porque+	turns	Long turns	Entonces	Pero+	Porque+
As	124	0	1	0	4	104	1	3	6	5
L	121	1	1	1	4	77	0	0	0	2
A	124	0	0	1	7	93	0	0	1	3
Total	369	1	2	2	15	274	1	3	7	10

Table 3: analysis of indicators of exploratory talk for Group 3

	pre					post
	turns	Long turns	Entonces	Pero+	Porque+	turns	Long turns	Entonces	Pero+	Porque+
T	131	25	0	3	11	112	39	9	9	15
M	46	0	0	0	1	73	2	2	4	0
I	64	0	0	0	2	28	0	0	1	1
Totals	241	25	0	3	14	213	41	11	14	16

Table 5. Group test scores

	pre	post
Grp1	21	27
Grp2	20	24
Grp3	26	27

Group 2, like group1, had two boys and a girl. Here there is little hard evidence of any increase in the indicators of exploratory talk and yet their joint score on the test increased by four points out of thirty. However, as with group 1, the indicators suggest that the one girl in the group, ‘As’, shifted her role to take more of a lead in the problem solving. When we applied the UK method to focus on the four questions that they failed to solve in the pre-test when they solved similar questions in the post test this shift in role was confirmed. In the pre-test there was more disputational talk when trying to solve these four problems. At times answers were disputed by ‘As’ but written in by the boy who had the pencil and the answer sheet. As with the first group this disputational style did not favour the girl. In the post-test there was less dispute and more reasoning contributions from the only girl (As) in the group. Although the quantitative evidence was not strong the qualitative analysis therefore did suggest that a shift in the style of talk, connected to a shift in the role of the girl, lay behind the increase in the test score.

The third group was made up of three girls. One girl, T, did most of the talking in the pre and in the post test. She already used reasoning words and indicators in the pre-test and the number of these indicators increased in the post-test. The score did not increase much partly because it was already high in the pre-test.

The Thinking Together programme is intended to encourage more able and articulate children to tutor the others. A key ground-rule is to ask other members of the group 'what do you think?'. In this way a leadership role in the group can become the role of creating space for others to talk - a sort of group facilitator role. In most, but not all, of the UK groups the question 'what do you think?', coached in the programme is found in the talk of the children. This was not the case in the Mexican data (we searched all groups for questions with ' piensas' or ' opinas' and found nothing at all). Certainly the programme in Mexico did not manage to convert T's dominance into help for others in her group.

Applying the Mexican approach to the UK data

In Mexico a coding analysis was applied to the talk used around each problem. This showed an increase in ET and a decrease in CT. This method was useful because it was used to demonstrate a 'Zone of Proximal Development' in the talk. The shift to exploratory talk mainly occurred in those problems that were neither too easy not to need it nor too difficult for it to make any difference (Fernandez,Wegerif, Mercer and Rojas-Drummond 2002)

We applied the same approach to analyse a sample of the UK data. We chose two groups from the UK pilot study because the tests used in the pilot were very similar to those used in Mexico. We could not use all the three groups video-recorded in the UK pilot study because one group changed one member between the pre and the post test. In a group of three a change of one person means that the post-test talk was not comparable with the pre-test talk.

Figure 1: Results of UK pilot: percentage change in types of talk by problem

This graph is very similar to that produced from the Mexican study (Rojas et al 2001). It shows a large percentage shift from cumulative talk in the pre-test to exploratory talk in the post test.

Figure 2: Results of Mexican study: percentage change in types of talk by problem

Discussion

The Raven’s test score results of the projects in Mexico and the UK look similar. There were also similarities in the use of some key words such as ‘porque’ in Spanish and ‘because ‘ in English. However some of the language used was different. ‘Entonces’ (therefore), which is a common word in Mexican Spanish was used by the children in the post-test in Mexico as a tool to help them solve the problems. No equivalent was found being used in the UK data. Whereas ‘because’ works back from a conclusion to refer to the evidence for it, ‘therefore’ does the opposite, working from evidence to a conclusion. However it was not apparent from the transcripts that these different word choices implied different strategies for solving the problems.

We found no evidence in the Mexican data of the children asking each other ‘what do you think?’ This simple rule proved very important for mixed ability group thinking in the UK. The fact that it is not found at all in the Mexican data suggests that there might be aspects of the teaching programme in Mexico that should be revised.

The qualitative analysis of data in the Mexican study suggested that gender was an issue. In the two boy and one girls groups the programme appeared to encourage a shift in the role of the one girl in the group from being a little subordinated towards taking on a more leading role in the joint problem solving. Both cumulative and disputational talk focuses on social identities more than on the task or on the shared reasoning. Gender identity is therefore very important in these types of talk. In Mexico, perhaps more than the UK, gender identity suggests that boys are more likely to see themselves in a leadership role than girls. Exploratory talk focuses away from identity onto the quality of shared reasoning (Wegerif and Mercer 1997a). Our hypothesis is that the ground rules designed to support a shift in focus from identities to shared reasoning made it easier for girls to speak and for boys to listen to them and to follow their lead.

A close search of all the UK data did not find a similar effect. In the only group in the recorded data with two boys and a girl there was no effect of this kind. The girl was dominant in the pre-test and in the post-test. However we were surprised that all the other recorded groups in the UK were two girls and a boy. This suggestion from Mexico has already led to a further research project in the UK to explore if there is a gender effect in the Thinking Together approach.

This comparison of the two projects has provided an overview of the Thinking Together approach from which some practical suggestions can be drawn.

Firstly the finding that all Thinking Together projects have approximately the same effect on Raven’s test scores regardless of the length of programme should make us more aware that, even in longer programme, we are teaching the basic ground rules of exploratory talk and then practising these. An alternative on longer programmes would be to follow-up teaching the ground rules with lessons teaching more complex ways of using language as a tool to help solve problems.

Secondly the almost total dominance of one individual in one of the Mexican groups, T in group three, suggests a need to think seriously about the role of the dominant and the very able in the Thinking Together approach. The simple ground rule of asking others what they think has proved effective in similar groups in the UK but it may be that more that this is needed. Perhaps the more able should be identified and encouraged to take on a modelling and facilitating role in the use of exploratory talk for the group.

Thirdly the strong suggestion that gender was important to the Mexican data indicates a need for more research on the impact of this factor. It also indicates a need to take gender more into account in developing the teaching programme. If gender roles are having a negative impact on group thinking in some contexts it might be useful to address this directly in some of the content of the Thinking Together teaching materials.

Finally it is clear that the success and sustainability of the Thinking Together approach in the UK owes much to the enthusiasm of the teachers involved in the project. One way forward for the Mexican team would be to work on finding ways to involve teachers more in the project.

Summary and conclusion

Given the very marked differences in the implementations of the Thinking Together programme in Mexico and the UK it is surprising that the results, both the quantitative test results and the qualitative discourse analysis, were so similar. One major source of the differences in the implementations was the difference in the educational systems. Another important factor connected with this was the difference in the culture of primary teachers. These differences may make it difficult to sustain the innovation after the initial funding runs out in Mexico, although, as always with questions of sustainability, much depends on the political climate.

As a result of this comparative the team in Mexico are working to make the training programme more sustainable. A more systematic teacher training programme is being developed so that the effective implementation of the programme can be less dependent on researchers.  One further strategy they have adopted is to make sure that those responsible for setting the curriculum are aware of their findings and to argue from this evidence for the need for changes in the curriculum.

Several interesting new findings emerged in the Mexican study. One of these was the significance of gender and the possibility that the Thinking Together approach is particularly empowering for girls in Mexico. This finding is already stimulating more research on gender and exploratory talk in the UK. More research on this topic is also needed in Mexico.

In summary form the conclusions of this comparison are:

· that both quantitative results and the results of discourse analysis were remarkably similar in the two studies

· that there are some signs that the programme may have empowered girls in mixed gender group work in Mexico in a way that was not noted in the UK

· that the relationship between teacher's normal practice and the proposed intervention was closer in the UK than in Mexico and this impacted on the implementation and the sustainability of the project

References

Barnes, D. & Todd, F. (1978) Discussion and Learning in Small Groups. London: Routledge & Kegan Paul.

Barnes, D., From Communication to Curriculum. Penguin, Harmondworth (1976)

Dawes, L., Mercer, N., & Wegerif, R. (2000). Thinking together: A programme of activities for developing thinking skills at KS2. United Kingdom: Questions.

Edwards, D, and Mercer, N. (1987) Common Knowledge: The development of understanding in the classroom.  London:  Methuen/Routledge.

Fernandez, M.,Wegerif,R.,Mercer,N., Rojas-Drummond, S (Fall 2001-Spring 2002) Journal of Clasroom Interaction. 2, 40-54

Mercado del Collado, R. El habla en el aula como vehiculo del proceso de ensenanza-aprendizaje en una escuela primaria en Mexico. Madrid, 1996: Tesis Doctoral en Psicologia Evolutiva y de la Educacion.

Mercer, N. (1995). The Guided Construction of Knowledge: talk amongst teachers and learners. Clevedon: Multilingual Matters.

Mercer, N. (1996b) Socio-cultural perspectives and the study of classroom discourse. In C. Coll & D. Edwards (eds.) Teaching, Learning and Classroom Discourse . Madrid: Infancia and Aprendizaje.

Mercer, N. and Fisher, E. (1993) How do teachers help children to learn? An analysis of teachers' interventions in computer-based activities. Learning and Instruction Vol. 2, pp 339-355.

Mercer, N., Wegerif, R., y Dawes, L. (1999). Children’s talk and the development of reasoning in the classroom.

Peón, M., & Rojas-Drummond, S. (in press). Promoción de habilidades argumentativas en niños de primaria. In R. Barriga (Ed.), Mitos y realidades de la escuela primaria. El Colegio de México. México. 

Plowden Report (1967) Children and their primary schools. Vol 1. Report. London: HMSO.

Raven, J., J. C. Raven, and J. Court. (1995) Manual for Raven's Progressive Matrices and Vocabulary Scales. Section 1: General Overview. Oxford: Oxford Psychologists Press.

Richardson, K. (1991) Reasoning with Raven - in and out of context. British Journal of Educational Psychology 61 (2):129-138.

Rojas-Drummond, S, Peon, M., Martinez, M (2001) Talking for reasoning among Mexican primary school children. Paper given at the 9^th European Conference of Research on Learning and Instruction, Fribourg. Switzerland.

Rojas-Drummond, S., & Alatorre, J. (1994). The development of independent problem solving in pre-school children. In N. Mercer, & C. Coll (Eds.), Explorations in Socio-cultural Studies, Vol. 3: Teaching, learning and Interaction, Madrid: Infancia y Aprendizaje. 161-175.

Rojas-Drummond, S., Fernández, M., & Vélez, M. (2000). Habla exploratoria, razonamiento conjunto y solución de problemas en niños de primaria. In La Psicología Social en México, Vol. VIII, México: Asociación Mexicana de Psicología Social, 403-410.

Wegerif, R. & Mercer, N. (2000). Language for Thinking: A Study of Children Solving Reasoning Test Problems Together. In H. Cowie, & G. Aalsvort (Ed.), Social Interaction and learning and instruction: The meaning of discourse for the construction of knowledge. United Kingdom: Earli and Pergamon, 179-192.

Wegerif, R. (1996). Using computers to help coach exploratory talk across the curriculum. In M. Kibby & J. Hartley (Eds.), Computer Assisted Learning: selected contributions from the CAL 95 Symposium. Oxford: Pergamon.

Wegerif, R. and Dawes, L. (1997) Computers and Exploratory Talk: An Intervention Study. In Wegerif, R. and Scrimshaw, P. (Eds) Computers and Talk in the Primary Classroom. pp 226-239. Clevedon: Multilingual Matters

Wegerif, R., & Mercer, N. (1997a). A dialogical framework for researching peer talk. In R. Wegerif & P. Scrimshaw (Eds.), Computers and Talk in the Primary Classroom, . Clevedon: Multi-lingual Matters.

Wegerif, R., & Mercer, N. (2000). Language for thinking: In H. Cowie& G.van der Aslsvort (Eds). Social Interaction in Learning and Instruction (pp. 179-192). Oxford: Pergamon.

Wegerif, R., and N. Mercer. (1997b) Using computer-based text analysis to integrate quantitative and qualitative methods in the investigation of collaborative learning. Language and Education, Vol. 11. 4

Wegerif, R., Mercer, N., & Dawes, L. (1999). From social interaction to individual reasoning: An empirical investigation of a possible socio-cultural model of cognitive development. Learning and Instruction, 9, 6, 493-516.

---

[1] Lyn Dawes went on to do a doctorate and is now a senior lecturer in education at De Montfort University