The interactive attribution of school success in multi-ethnic schools

The study shows how explanations for school success are expressed and dialogically constructed during teacher–parent conferences at school. Attribution theory is used to conceptualize the various explanations for school success that were expressed. However, instead of only looking at attributions as beliefs which individuals or groups ‘have’, the aim of this study is to show how attributions are part of co-constructed processes in which multiple partners impact upon each other’s attributions over the course of a conversation. The results indicated that in the conversations between teachers and minority parents, school performance is more often attributed to effort while in conversations with majority parents, psychological attributions were more common. Besides these differences in content, the process through which these accounts were constructed was different. While the diagnosis on what went wrong was more commonly constructed in case of the conversations with majority parents, they were more characterised by opposition or a passive position by the parent in case of the conversations with minority parents. The analyses show that instead of a simple mismatch between explanations of the home and the school, these explanations are interactionally co-constructed as both parents and teachers necessarily ‘re’-act on each other’s claims and understanding of school success. The results ultimately reveal how the interactive process impacted upon the construction of the attributions and the possibilities this creates for partnerships between parents and teachers to create an understanding of the child’s academic potential across home and school.


Attribution theory and the interactive explanation of school success
In many countries across the world, children with an ethnic minority background still generally do less well at school as compared to ethnic majority children (Centraal Bureau voor de Statistiek 2007;Martin et al. 2007;Ogbu 2003). One factor that has been brought forward in the literature to explain this achievement gap is the different explanations majority and minority populations have of how school success can be achieved. In the current study, we are using notions from the attribution theory, next to notions on the dialogical nature of cultural differences to gain further insight into how school success is explained in an interactive setting in a multi-ethnic school. More concretely, the study shows how explanations for school success are expressed and dialogically constructed during teacher-parent conferences at school. The aim of this study is to show how attributions are part of a co-constructed process in which multiple partners impact upon each other's attributions over the course of a conversation and beyond. Instead of only looking at attributions as beliefs or cognitions that different individuals or groups 'have', our focus of interest is on the social construction of inferences and on how the dynamics of the conversation and contextual factors impact on them. With this goal, the study contributes to the need expressed earlier by Potter and Edwards (1990) to study attributions in more naturalistic settings as compared to 'classic' attribution studies and in particular to the need to conceive attributions in their sociodiscursive context. Given the notion in the literature that school success is differently accounted for in case of minority children as compared to how this is done for majority children, the study specifically focusses on how school success is differently explained between teachers and parents with an ethnic minority background as compared to between teachers and parents with an ethnic majority background.
We will situate this paper into two different research traditions. First, we will give an account from the literature on how attribution theory can be applied to the explanation of school success and how ethnicity might play a role in attributive processes. Second, we show how this perspective needs to be complemented with a perspective from discourse theory. We focus in particular on how dialogues develop in inter-ethnic settings in order to understand how attributions are differently constructed in conversations with partners from the same (ethnic) background as compared to how this is done with partners from a different (ethnic) background.

Attribution of school success and ethnic diversity
Attribution theory, or the study of perceived causes of behaviours and events, has applications in many different settings, among which the educational setting. In achievement contexts, the level of performance is predominantly ascribed to some ability factor (i.e. 'intelligence') and/or an effort factor (i.e. how hard you try). The first attribution concerns an uncontrollable but stable cause, while the second refers to an unstable but controllable cause (Georgiou et al. 2002). Following the classical model of Weiner (1985), the kind of attributions people make are believed to have an impact on affect, level of expectancy and achievement behaviour. Empirical studies confirmed that attributions influenced expectancy and achievement behaviour, which in turn affected performance (Georgiou et al. 2002). According to Miller (1995), attribution not only applies to the explanation of the individuals' own behaviour but also to the behaviour of others. In the current study, not the attributions of children themselves but the attributions of the teachers and parents for the child's school performance will be examined. With respect to the attribution categories used, we based ourselves on a study by Graham (1991) who found that teachers can attribute to internal (i.e. ability, effort, psychological factors) and/or external factors (i.e. help at home, help at school, task difficulty, system).
If we focus on the educational setting and the attribution of school success as related to ethnic status, the literature shows that minority parents usually have higher expectations and aspirations (Coenen 2001;Müller and Kerbow 1993;Nijsten and Pels 2000;Phalet and Schönpflug 2001) as compared to majority parents and that teachers generally seem to have lower expectations for minority children than for majority children (e.g. Van Ewijk 2009;Weinstein et al. 2002;Hall et al. 1986). Although many studies focussed on teachers' and parents' differential expectations, possible differences in the attributions teachers (or parents) make for the performances of majority and minority children did not receive as much attention. However, it can be expected based on these studies that teachers will attribute school success of ethnic minority children less often to high ability (as compared to majority children) and that minority parents, given their higher expectations, will attribute school success more to (a lack of) effort. This lack of attention for attribution is remarkable, since attributions can give more substantive insight into the reasons behind differential expectations (i.e. as attributions were theorized to underlie expectancy levels). For instance, it is important to know what role ability attributions play as compared to effort attributions, or how (individual) effort plays a role as compared to social circumstances in how educators and parents account for the child's success. This study fills in this gap by studying if teachers and parents employ different attributions for the school success of minority children as compared to majority children.
Attributions from a discourse analytical perspective and the establishment of 'common ground' Making use of a discourse analytical perspective (see, e.g. Gee 2003;Jaworski and Coupland 2001;Fairclough 1992;Gee and Green 1998), we are focussing not just on the description of individual attributions of the different partners involved in the conversation but as well on how these attributions are jointly constructed in the flow of the conversation. Moreover, we focus on how the attributions of one partner cannot be seen as (fully) independent of the other partner of the conversation. Attributions are part of the ongoing discourse in which both partners are actively involved and are bound to re-act on each other's moves in the conversation. At the same time, the positions of the conversational partners, in this case, their beliefs on how the school success of the child is accounted for, are also both the input and the background against which these conversational moves take place. Attributions, as any other psychosocial construct, are thus double-natured in this respect: They are both the input of the conversation as well as (defined by) the process of how they are used. As such, they are both beliefs that reside in the minds of people and subjected to and formed by 'language in action'.
Given our focus on how (the construction of) attributions differ depending on ethnic differences between participants, we are drawing from theoretical notions that explain how partners with a different frame of reference interactively (re)produce these frames while they coordinate their common actions. Taking the position that language is a form of joint action, being involved in discursive interactions demands constant coordination by the participants. From an intercultural communication perspective, successful coordination depends on the establishment of common ground, a frame of reference based on shared knowledge, beliefs and assumptions (Clark 1996). In the classic accounts on intercultural communication, it is presumed that the less ground is shared by conversational partners, the more they have to exert themselves to communicate in a successful manner (Jacobs 2002;Jones et al. 2009). However, cultural differences do not just influence conversation from the outside but they are thematized or purposively neglected by the interlocutors. For instance, the partners in a conversation use interactional strategies to move closer to each other, or to do the reverse and emphasize interpersonal or cultural differences. Moreover, the idea of opposing and diverse frames of reference cannot explain what happens in the conversations between teachers and migrant parents. The reason for this is that participants in these conversations constantly shift their positions depending on how the dialogue with the other partner develops. Building on views that acknowledge the fundamentally dialogical nature of culture and the idea that cultures are continuously produced, reproduced and revised in dialogues with their members (Mannheim and Tedlock 1995;Clifford 1988), we argue that cultural meanings and cultural differences are not 'given' but are reconstructed in dialogic encounters. As Mannheim and Tedlock argue, cultural events or dialogical encounters between members of cultural groups are not the sum of the actions of individual participants. Instead, individual productions of culture reflect former dialogical encounters and shared constructions of culture are the key sites of the production of culture. In this view, every cultural notion tastes of the lived encounters in which these cultural notions were articulated analogous to the Bakhtinian view that 'each word tastes of the dialogical encounters in which it has lived its socially charged life' (Bakhtin 1981, p. 293).
We think this perspective is useful for the present study as it allows us to study how attributions for the child's school success might be impacted by possible 'a priori' differences in 'common ground' between parents and teachers of a different ethnic background as well as by their attempts to deal with or make use of (the lack of) common ground. From this perspective, we expect the conversations to develop differently, depending upon this level of correspondence as well as on the particular strategies parents and teachers employ in bridging these differences. In addition, we expect that the attributions that are expressed in these conversations and the kind of understandings of how school success works that they create are impacted by the communicative work in these inter-ethnic settings.
The questions asked in this study can be summarized as follows: First, are there differences in the attributions parents and teachers make for a child's school performance during a teacher-parent conference depending on the ethnicity of the participants? Second, how does the attributional process differ for parent-teacher couples with the same ethnic background as compared to teacher couples that differ in ethnic background? In particular, how does (the lack of) 'common ground' impact upon how these conversations develop? Asking both the first and the second question reflects the notion that attributions can be seen as double-natured, that is, as both relatively instantiated beliefs that 'belong to' the cognitions of individuals and as processes that are discursively constructed. Lastly, we are interested in what can be said about the quality of these attributions when studied as interactive processes and how this can shed light on understanding the explanation of school success as something that is constructed across home and school.

Corpus and participants
The corpus consisted of 54 recordings of parent-teacher conferences in the final year of primary school (when the child is 12-13 years old). In these conversations, the primary school teacher presents his or her advice for the child's level of secondary education to the parents. Children and their parents are not free to choose these levels but have to apply for admission. In deciding about admission, a secondary school considers two pieces of information: the child's score on the National Test (CITO test) and the advice of the teacher of the primary school the child is leaving. For parents, a lot is at stake in these conversations as the school level their child is assigned will be decisive for their child's (school) career. Teachers experience these kinds of conference as difficult as they have to steer between the parents' wish to assign the highest level possible and their concern to give a realistic advice to the secondary schools. The complete conversations lasted on average 16.5 min (M majority 016.47, SD05.28; M minority 016.46, SD07.70).
In most cases, one of the parents was present in the conversation, either the father (18 times) or the mother (22 times). In one case, the older sister of the child was present, and in ten cases, both mother and father were present. For the readability of the manuscript, we chose the term 'parent' to refer to the parent/caretakers who participated in the study. If both parents attended the parent-teacher conference, the responses of the parent who made the largest contribution to the conversation (i.e. indicated by the highest number of utterances) were used in the analyses.
The 54 children (26 boys and 28 girls) who were subject of the conversations were mixed in ethnic background (i.e. 15 ethnic majority Dutch and 39 ethnic minority; of whom 22 Moroccan-Dutch, 3 Turkish-Dutch and 14 of another ethnic background). A child was considered to have an ethnic minority background, if one or both of the parents were born outside the Netherlands. This criterion refers to how ethnicity is defined in the Dutch social context (Keij 2000). 1 Both the fathers, χ 2 (4, N 048) 012.88, p 00.01, and the mothers, χ 2 (4, N053)024.11, p00.00, of the ethnic majority children had higher educational backgrounds than the fathers and mothers of the ethnic minority children. This difference in parental educational background matched the distribution in the general Dutch population (GGD 2008). Overall, the populations of the participating schools can be considered as representative for the population of inner city schools in the Netherlands.
Five different teachers (four females and one male) were included in the study; all had a high educational background and spoke the Dutch language fluently. All conferences were conducted in Dutch, although parents did differ in the fluency of their Dutch. The schools (four in total) were located in multi-ethnic neighbourhoods in Utrecht, one of the four largest cities in the Netherlands. The percentage of ethnic minority children in the classes in these four schools ranged from 39% to 91%, which is fairly common for the urban areas where the immigrant population in the Netherlands is concentrated.

Procedure
Before the parent-teacher conferences took place in these schools, all concerning parents were informed about the research. It was assured that if the parents would decide to cooperate, the data would be handled with strict care and for purely scientific purposes. When the parents entered the room where the parent-teacher talk took place, the teacher introduced the researcher, asked whether the parents had read the letter and, next, whether they had any objection against the recording of the conversation. If the parents agreed to cooperate, an audio recorder was switched on and started recording the conversation. The researcher took a non-intrusive seat in the back of the room. Most parents (54 out of 64, 83%) gave permission for the recording of the conversation for scientific purposes.

Number of teacher and parental attributions per category
Trained research assistants transcribed all conversations. A coding instrument was developed for the coding of the attributions using transcriptions of the audio recordings, based on Graham's (1991) categories. When teachers or parents made a statement about how they explained the level of school performance of the child (now or in the future), this was considered an attribution. To suit the context and the corpus of the current study, in addition to the original categories of Graham (1991), we differentiated between help from parents and from the school and added a category 'educational system'. Consequently, the attribution categories that were used in the current study were ability, effort, task difficulty, psychological factors, personality factors, help at home, help at school, educational system and 'other'. In Table 1, descriptions and examples of the categories can be found. For both the teacher and the parent, each attribution for the child's school performance was scored according to one of these categories. Additionally, for each attribution, we determined whether the attribution concerned a so-called lack attribution (for instance, when the parent attributed the child's level of school performance to a lack of effort) or a so-called presence attribution (for instance, when the parent attributed the school performance to the presence of ability). The interrater reliability of the coding instrument was established to be good, Cohen's kappa00.71 (Zwaal 2007).

Attributions as interactive
Given the dominant role of the teacher in the conversation, for each teacher attribution, we determined if the parent confirmed the attribution of the teacher (by giving a similar attribution after the teacher had given his or her attribution) or if the parent contradicted or questioned the Table 1 Definitions and examples of the attribution categories

Attribution category Definition Example
Ability When performance is ascribed to aptitude or acquired skills by the child.
'Because Mike is able to do many things.' 'She is a very intelligent girl'.

Effort
When performance is ascribed to both temporary and sustained effort (endeavour) on the side of the child.
'He is working really hard.' Task difficulty When performance is ascribed to difficulty of the subject or of the task.
'Math was difficult this year.'

Psychological factors
When performance is ascribed to factors that refer to the psyche of the child (motivation, concentration, fear of failure).
'She seems not really motivated.'

Personality factors
When performance is ascribed to a personality characteristic of the child (seriousness, interested, structural personality).
'She is just interested in many things.' 'Miriam has always been eager to learn.' Help at home When performance is ascribed to the help the child receives at home from family members.
'Fatima gets a lot of support from you and her sister.' Help at school When performance is ascribed to help from school personnel.
'We are giving him extra attention.'

System
When performance is ascribed to the working of the educational system.
'It's a problem that the CITO-score is so determining.' Other All other factors. 'The divorce had a big impact on her school work.' teacher's attributions (by expressing that they did not agree with the teacher of by posing a question that questioned the attribution made by the teacher). Furthermore, we assessed the number of times the parents put forward an 'independent' attribution by themselves (i.e. an attribution that differed from the foregoing teacher attribution). This analysis enables to establish the inter-dependency of the attributive process, in particular as related to the teachers attributions.
Attributions in their discursive context In order to shed light on how the attributions were constructed discursively, a second qualitative analysis was done focussing on the following questions. What was the main thematic content of the conversations? What particular explanations for school success were introduced and elaborated? How did the respective partners react on, confirm, dispute, contradict the explanations of the other party? And to what extent are the participants able to establish common ground, or do they stress their diverse views while they are constructing accounts for the child's school success? As we depart from the idea that parents and teachers do not just present their views on the child and on upbringing 'as such' but introduce and present their ideas strategically and use them purposively to reach particular aims, attention was paid to these strategic aspects, especially paying attention to how the attributions made are a reaction to what happened earlier in the talk. In carrying out this analysis, while taking the transcript with the marked attributions as a point of departure, we looked at the interactive construction of these attributions in the context of the conversation as a whole. For each transcribed conversation, text fragments were labelled using MAXqda2 software focussed on the above-mentioned questions. Then, as a second step, for each transcript, summaries were made paying attention to these focus points and the particular course of action in each conversation. As a third step, conclusions were drawn on general patterns that were characteristic for the sample as a whole using the summaries paying particular attention to possible differences between the two groups (minority versus majority parents). For this analysis, we randomly selected 34 parent-teacher conferences from the first sample. In order to have an equal representation of Dutch and migrant parents, the parents had to have either a Moroccan or a Dutch background. In this second sample, 18 parents had migrated from Morocco, 14 parents are native Dutch. In one family, the mother is Dutch and the father is a migrant from Ecuador; in another family, the father is from Morocco, whereas the mother is Dutch.

Comparing individual attributions
We compared the individual attributions made by both teachers and parents in the two groups. First, we conducted a repeated measures analysis with the total number of attributions during the complete conversation of both the teacher and parent (as repeated measures) to find out whether teachers or parents made more attributions. The results indicated that, first of all, teachers made significantly more attributions than parents, F(1, 50)044.64, p00.00, and second, that this difference was even more pronounced in the minority group, F(1, 50)05.88, p00.02. Besides, the total number of attributions per category (for teacher and parental attribution separately) served as the continuous dependent attribution variables in parametric quantitative analyses with ethnic background of the parents as a factor, in order to examine if certain attributions were more frequent in the majority or minority group. Additionally, we analysed if there were differences in the mere presence of each attribution category. The analyses were performed both with and without control for the level of advice the teacher presented to the parents (as an indicator of the child's level of school success) and the total numbers of attributions parents made, see Table 2.
These results indicated that the total number of attributions the teacher made differed significantly depending on the ethnic background of the child, t(49,31)0−2.38, p00.02, with a mean of 13.2 attributions when the teacher was speaking to the parents of a minority child compared with a mean of 8.4 attributions in the case of a conversation with parents of a majority child. The total number of attributions the parent made did not differ significantly for the ethnic groups, t(52)00.98, p00.33. On average, the majority parents made 3.3 attributions, while the minority parents made 2.3 attributions during the parent-teacher conference (Table 3).
Furthermore, univariate analysis of variance indicated that teachers significantly made more effort attributions in the conversations with minority parents (F (1, 50)08.13, p<0.01, partial η 2 00.14). In addition, independent of the frequency, thus while only looking at whether or not an attribution occurred (once or several times versus never), chi-square tests revealed that teachers' attributions to a 'lack in' psychological factors (such as a lack in motivation or concentration in the child) significantly occurred more in the conversations with majority parents than in conversations with minority parents. 'Lack in' psychological factors occurred 33% in conversations with majority parents versus only in 8% with minority parents (χ 2 (1)05.22, p<0.05). For the parents, no significant differences were found with respect to the frequency of attributions the teacher presented to the parents and the amount of attributions. This is not surprising given the overall low frequencies of attributions of parents in these conversations. However, independent of the frequency, thus while only looking at whether or not an attribution occurred, chi-square tests revealed that Other factors (presence) 0.00 0.00 0.14 0.59 majority parents made more attributions to the personality of the child (for instance, eager to learn, confident personality). Majority parents did this in 33% of the cases against only 10% of the minority parents (χ 2 (1)04.15, p<0.05).

Comparing attributions in as interactive
With respect to what extent parents went along with the attributions of teachers, we checked if there were differences in the presence of each of the reactions that were scored (affirming, contradiction or questioning, and bringing in new attributions). The results demonstrated that parents made more attributions that confirmed the attributions of the teacher than they made 'independent' or new attributions (mean number04.62 for confirming versus 2.02 for independent attributions, F(1, 51)015.99, p00.00). This implies that the attributions made by teachers are the ones that dominate the conversation as a whole. The ethnic groups of parents differed significantly in whether they made 'independent' attributions, or attributions that differed from the foregoing teacher's attribution, χ 2 (1)05.71, p<0.05. The percentages per ethnic group showed that of the ethnic majority parents a high percentage (87%) made one or more independent attributions (versus 13% who did not make such attributions), while in the minority group, this was more evenly distributed. That is, 51% of the minority parents made one or more independent attributions and 49% did not make any independent attributions during the conference.
Comparing assigning attributions as a discursive process The qualitative analyses revealed first of all that in all conversations, there was some form of tension between what had been reached and what both parties wished for the child. Moreover, as a rule, parents seemed to have an interest in arguing for a higher advice, while the teachers pleaded for 'realism'. In both groups, the conversations between the teachers and parents basically centred around two issues. One issue was 'What went wrong?', 'Why did the child not do better?' or 'Why does the child better now than before?' and the other issue was 'How can the teacher and the parent make sure that the child is going to do better or continues to do well in the future? In these conversations, one recurring aspect was how much weight should be placed on ability. Negative attributions on ability (the child is not smart enough) were in all cases addressed with a certain level of cautiousness and seemed overall to be face threatening. Ability was often weighted against effort (the child should work harder) combined with 'external' factors (how the social environment is able to support greater effort) in all cases. However, the issue of how effort should be established differed considerably between the conversations with minority and majority parents. For majority parents, a dominant focus of the conversation was the establishment of the right psychological and motivational structure so that the child will take initiative. In contrast, the discussions with minority parents focussed on whether or not or to what extent parents should force the child to work hard, or prohibit non-school related activities. More specifically, while in the conversations with minority parents, the issue was how external factors should be acted upon, such as to what extent the child should be monitored, the discussions with majority parents focussed on how to act upon internal, psychological factors such as motivation. Besides these differences in content, the process through which these accounts were constructed was different. While the diagnosis on what went wrong was more commonly constructed in case of the conversations with majority parents, they were more characterised by opposition or a passive position by the parent in case of the conversations with minority parents.
A typical example of a conversation with a minority parent is presented below. The effort explanations that were dominant in the conversations with minority parents were not build as a common construct between the teacher and the parents but had a more oppositional character. Typically, teachers worked around or argued against the effort accounts of migrant parents as is the case in the example of Karim's father below (Excerpt 1). In this conversation, the teacher introduces his explanation for the disappointing school result of Karim in turn 1: Karim finds school difficult. And this claim is supported by referring to 'objective' school results in the following turns (up until turn 13). The father does only passively support these accounts with hmm and yes in turn 2, 4, 6, 8, 10, 12 and 14. These accounts of the teacher all seem to support a lack of ability explanation. Then, in turn 15, the teacher combines the lack of ability explanation with arguing against a lack of effort explanation: 'He does his best, but he just finds it difficult'. The father does not directly oppose this view, but he brings in his own account when he suggests that the school should perhaps help his child. The teacher then extensively makes clear that the child is getting a lot of extra help but still finds learning difficult. As the conversation develops, the teacher tries to further sustain the image of an unable child through referring to dyslexia, while the father holds on his plea for more help at school. Towards the end of the conversation, the teacher again anticipates the father's explanation that it is the child's fault because he has not worked hard (turn 20). While the father seems to oppose this view in turn 21 with a 'but…', the teacher continues arguing that the child cannot do better in turn 22. While the father is trying to reformulate in turn 23, the teacher interrupts him in line 24, stating that the child is effortful (he does a lot), but he just finds learning difficult.  In contrast, reference to psychological causes to explain a school result was more dominant in the conversations with majority parents. Teachers referred, for instance, to (a lack of) motivation, concentration or attention but also to particular 'disorders' such as performance anxiety. Parents and teachers would discuss the measures that needed to be taken, including professional help such as therapy or organized leisure activities such as sport clubs. In these discussions, often both parents and teachers co-constructed explanations or solutions to be taken, searching for how the right balance of challenge can be reached through a common effort of parents and teachers, such as is the case for the conversation between the teacher and the mother of Annelies, a Dutch girl, shown in excerpt 2.
1) M: Yes, but because you are also saying that she asks you for help very easily 2) T: Ja ze is heel onzeker.
3) M: (laughing). And I heard from Petra that you had told that she sometimes stands at your desk every 5 min with her arithmetic work 4) T: Ja. In this excerpt, the teacher and the mother are discussing why the results on math of Annelies are still not satisfying and how her attitude and self-confidence is impacting on that. Both the parent and the teacher bring in different explanations from different perspectives to support the same thought: Annelies lacks self-confidence as well as concentration and this hinders her school work. In the conversations with minority parents, these psychological explanations were sometimes also brought in, mostly by the teacher, but they were hardly elaborated in the interaction. Parents would in most cases confirm such explanations, but not expand upon them.

Discussion
Differences in attributions explained from the perspective that groups 'have' certain explanations for school success Both the quantitative and the qualitative analyses revealed that there were thematic differences between the two groups of parent-teacher conversations. In the conversations with minority parents, the issue of effort was more prominent, while in the conversations with majority parents, attitudes and personality as conditional factors were more prominent. It is important to notice, however, that given the overall dominance of the teachers in assigning the attributions, these differences were mainly due to how the teacher led the conversations and 'set in' the attributional process. The differences can be explained referring to the different explanations for school success the teachers have for minority versus majority children. It is known that teachers generally have lower expectations for minority children than for majority children (Van Ewijk 2009;Weinstein et al. 2002), although not much is known about differences in how they explain school success of minority students as compared to majority students. The differences could also be explained by referring to different theories parents might have of how they or their children can reach success. Teachers might have anticipated on these theories, given their knowledge of the accounts both minority and majority parents have of the school success of their children. Minority parents and in this case the mostly Moroccan parents might see effort of the child as the main vehicle to a higher social status and have less eye for how parents and teachers can invest in creating the optimal circumstances for 'success', in line with what was found in earlier studies on these immigrant groups (Douma and de Haan 2008). Furthermore, it is known that minority parents in the Netherlands, in comparison with majority parents with the same SES background, have higher educational aspirations for their children (Coenen 2001;Müller and Kerbow 1993;Nijsten and Pels 2000;Phalet and Schönpflug 2001). These high expectations could go together with different kinds of attributions, for instance, attributions to controllable factors such as effort. Moreover, research has shown that sociopsychological distributed among majority parents as compared to minority parents. Parents in Western cultures have a long history in helping their children with school and are more experienced with seeing their children, in general, as psychological beings and, more specifically, as young learners (Epstein 1987).
However, rather than putting forward an explanation that sees these thematic differences as referring to knowledges or positions that belong to particular ethnic groups, we want to highlight the dialogical nature of how these differences come into being. This does not rule out the explanations just mentioned, but it does put them a perspective in which these more 'instantiated' forms of expectations are not independent from how they are enacted on a day-to-day basis in settings with multiple others that confront, confirm, stabilize or reshape these explanations.
Differences in attributions explained from its situated and co-constructed nature The fact that attributions do not only 'belong to' certain parties in the conversation but are at the same time 're-active' in the sense that they are (a) shaped by the attributions of the other conversational partner and the course of action and (b) based on inferences that are made of the mindset of the other can be illustrated from the data in the following way. First, the attributions as they were enacted can be seen as a chain of reactions to former attributions and are therefore not independent of what the other person is stating or of the course of the conversation as a whole. For example, as is the case with Karim's father in Excerpt 1, the explanation of a parent that lack of school success might be induced by a lack of help from school, occurs as a reaction to accounts from teacher that the child lacked ability. Second, partners seem to give their attributions based on what they assume the explanations of the other partner are. For instance, the claim of the teacher at the beginning of the conversation in Excerpt 1 that effort is not the problem is a reaction to a presupposed explanation of the lack of school success by the father that his child should work harder (compare Grossen and Apothéloz 1998 for a similar finding on how participants in a therapeutic setting anticipate on the presumed explanations of the other conversational partner). Thus, the attributions expressed by the participants cannot be seen as 'direct' representations of their explanations of school success but are formed by the need to react on the (presupposed or expressed) explanations of the other.
Furthermore, the qualitative analyses make it plausible that people also act strategically when they account for the school success of the student. They might use face saving strategies, seek opposition or may be concerned to keep peace with each other. An example of this is the fact that teachers might avoid to express lack of ability explanations and instead use 'not effort' explanations as a disguised form of lack of ability explanations. This has been found in other work in which intelligence was a possible explanation for a phenomenon The attribution of school success in multi-ethnic schools 309 factors play a large role in the explanation of school success (Portes and Rumbaut 2006), and it might be that this kind of knowledge on how to stimulate school success is more widely at hand in an institutional setting. In the same study by Grossen, intelligence was avoided both by parents and the professional to explain the child's problem. This is in line with previous studies on sensitive topics in which indirectness is a common conversational strategy (Bergmann 1992;Linell and Bredmar 1996). Moreover, there were significant differences between the two groups in the nature of the interactive process which we suppose have an impact on how these attributions function in understanding school success across school and home. While in case of the majority parents, the attributional process was more characterised by common construction and an active participation of both parties, in the case of minority parents there was, next to a passive attitude of the parent, more 'strive' and opposition around what attribution could explain the child's school performance. It seemed that while the conversational frame in the case of the minority parents was defined by the assumption that relatively large differences existed between the teacher and the parents in terms of identity, knowledge of the system, language, pedagogical views and the view on the professional role of the teacher, this was much less the case for the majority parents. For instance, the effort explanations that were dominating the conversations with minority parents were shaped in dynamics of opposition in which denying effort and lack of ability were put against (presumed) effort explanations. In contrast, in the case of majority parents, the personality and psychological explanations were often but not exclusively introduced by the teacher but then were picked up and further developed by the majority parents together with the teacher while apparently building upon 'a priori' similar views. In line with intercultural communication theories, it could be claimed that successful coordination depends on the establishment of common ground, a frame of reference based on shared knowledge, beliefs and assumptions (Clark 1996;Jacobs 2002). However, from a dialogical perspective on difference, we see these differences not only as the result of primordial differences that are 'present' in the individual participants, or in the groups they represent, but rather as both the consequences of the a priori differences and the differences constructed in these dialogues. For instance, the effort explanations in the dialogues with minority parents have gained a particular meaning through how they were employed in these conversations. Instead of becoming a common ground on which parents and teacher can take action to improve the efforts of the child, they have become a means for the teacher to blame the parent not to acknowledge the 'actual' reason for the child's failure. Moreover, from the parents side, the (strive around the) effort explanations might result in an impasse, a feeling of not being supported by the school in their efforts to stimulate their children to do better. Thus, these effort explanations are characterised by disagreement and the tension of being weighted against other explanations in which the school and the parents take different positions. In contrast, the psychological explanations as expressed in the dialogues with majority parents have become strengthened in these conversations by a process of confirmation from both sides and therefore more ready to take up in a common strategy in which the teacher and the parent can collaborate to find common means to act upon the child's problems.
The rational of the paper was to contribute to the understanding of minority student underachievement from the perspective of how expectations of their school success are formed in the daily settings of parent-teacher conversations. We hope to have shown that instead of a simple mismatch between expectations of the home and the school, these expectations are interactionally co-constructed and are formed over time, as both parents and teachers necessarily 're'-act on each other's claims and understanding of school success. In addition, our study pleas for a situated view on attributions rather than one that focusses only on 'a priori' cognitions. In line with Potter and Edwards' (1990) critique on the classic attribution theory, we think that to see attributions just as the outcome of an internal state of mind would deny the fact that cognitions are both social, external and psychological, internal phenomena. Expectations of school success are not just fixed representations that people have in their minds but also are the result of multiple interactive enactments of these representations. In terms of the practical implications of this study, we would like to stress the notion that the partnership between parents and teachers is formative for how explanations for school success develop and become enacted. The less developed partnerships with the minority parents then are not just a concern related to the lesser likelihood of achieving common goals as is already documented in the literature (e.g. Epstein 1987). They are also relevant for how school success is interactively understood across school and home and the possibilities parents and teachers have based on this understanding to coordinate their actions to steer towards greater competence of the child. However, the insight generated by this study that these conversations can be key sites of differential production of the explanations for school success also creates the possibility for schools to develop initiatives to act more strategically in avoiding divides in the explanation of school success based on ethnic or other membership categories.