The multiple constraints of addressed questions in whole-class interaction: Responses from unaddressed pupils

This article explores pupils’ responses to addressed questions in two third-year primary school classes, organized as plenary interaction and based on the next-speaker selection. In this context, unaddressed pupils often produce responses of various kinds spontaneously, showing that the next-speaker selection per se does not exclude unaddressed pupils from participating. Analysis of the design and position of these responses show their orderly nature as mainly depending on the following dimensions: the position of the address term in the question and who has primary access to answers. Pupils’ responses display a high level of awareness of the next-speaker selection rule operating in this setting, and more globally, of the turn-taking system. This competence enables pupils to understand and navigate the other-selection rule, often gaining their right to speakership. In line with prior studies on multiparty interactions, the article shows that teachers’ questions pose multiple constraints on responses.


Introduction
This article presents an interactional account of a well-known but not extensively studied phenomenon occurring in school early grades, during activities organized as whole-class interaction, with the teacher addressing questions in round to one pupil at a time.
Following an addressed question, non-selected pupils often engage in requests for permission to answer and other responsive actions, as well as direct answers, substituting or accompanying the selected speaker (Erickson, 1982;Shultz et al., 1982). In this article, all these reactions (bids for selection, claims and displays of knowledge, complaints, and also direct answers) are considered responsive actions for all practical purposes.
In the context here examined, the production of series of questions is based on the other-selection rule (rule 1a, in Sacks et al., 1974) with the aim of ruling an orderly access to the turn, whereby the preference for the answer is subordinate to the selection of the recipient. However, the responsive behavior from other-than-the-selected speaker that so frequently occur in this context displays the orientation of the class to at least two constraints posed by teachers' addressed questions in this context. First, that the otherselection rule operates  and second, the conditional relevance of the answer upon the question (Heritage, 1984;Schegloff, 1968). The following extracts, from two third-year classes in an Italian primary school, illustrate this phenomenon. Questions are indicated with arrowed lines; responses from non-selected pupils with the 'greater than' (>) sign. Ex This sequence, from a geometry lesson, follows a demonstration by Giuseppe. According to the teacher's instructions, he paced the room, changing direction several times and drawing right-angle trajectories on the floor. The demonstration was for Caterina, who was absent the previous day when all other students attended it the first time. Here we see that, despite the teacher naming Caterina in line 1 before the question, non-selected pupils spontaneously self-select in lines 3-5, producing bids for selection, without waiting for Caterina to answer. 1 Extract 2 is from a class on Italian grammar that took place in a different group with pupils in the same age brackets as the first extract. In line 1, the teacher asks Fabrizio to offer the subsequent verb form from a reading passage, during the collective correction of their homework. In line 2, some pupils, acting in unison (Lerner, 1993(Lerner, , 1995, answer before the selected speaker. The two extracts illustrate two types of responses from non-addressed pupils: in extract 1 pupils bid for selection, in extract 2 they directly answer to the question, anticipating the selected pupil. Both conducts are responsive actions, although each displaying the pupils' orientation to a different type of constraints. 3 When bidding for selection pupils show to be orienting to the selected pupil as the principal recipient of the questioning, while they simultaneously cast themselves as knowledgeable and eligible next speakers. In the context of classroom interaction, bids for selection represent a very common type of responses to teachers' questions. This type of conduct can be included among non-answer responses (Heritage, 1984;Stivers and Robinson, 2006). More specifically, in the classroom context, bids for selection address the relevance of the answer while simultaneously recognizing that rule 1a (Sacks et al., 1974: 704) 4 is operating. However, as shown in extract 2, non-selected pupils may also answer directly instead of (and, sometimes, in addition to) the addressed one. This behavior displays the pupils' orientation to the relevance of the answer and of the progressivity of the interaction (Stivers and Robinson, 2006) over the other-selection technique.
These two reactions contrast with that in extract 3, where non-addressed pupils refrain from any type of responsive behavior, leaving the floor to the selected next speaker. Ex. 3 [PM:FZ:12b:geography] In this extract the interaction unfolds according to the traditional three-part model or IRE (Initiate-Response-Evaluate, as in Mehan, 1979;Sinclair and Coulthard, 1975). These examples significantly show that three different reactions can be produced in response to the teacher's selection of the next speaker; evidencing the pupils' different understanding of the contributions each pupil is allowed to in each case. This article aims at investigating what drives unaddressed pupils to produce such responsive actions in settings where the turn-taking system is explicitly organized according to the otherselection allocation technique.
The graph below represents the distribution of the pupils's reactions to addressed questions. In my data, in about two-third of all the addressed questions, the non-selected pupils engage in responsive actions of various kinds (vocal and visual bids for selections, claims and displays of knowledge, complaints and direct answer to the question). These responses often accompany the answer of the selected pupil (extracts 1 and 2). Moreover, in one-fifth of all the addressed questions in the corpus, answers are produced directly by one non-addressed pupil (either instead of or in addition to the selected next speaker). By contrast, in about one-third of the collection, the sequence develops 'orderly' according to the next-selection rule and the IRE model (extract 3), and without any responsive conducts from the class.
These proportions portray a turn-taking system where the other-selection accomplished by the teacher is differently interpreted and actively negotiated by the class. The article investigates these responsive conducts from unaddressed pupils, providing an interactional account and an institutional understanding of these forms of behavior. Indeed, from a lay person's perspective, these conducts can be considered typical forms of child-like behavior and evidence of their lack of social competence in interaction. This article takes a different view. In contrast with the opinion that children of this age do not have sufficient interactional competence to comply with the explicit and implicit constraints of the teacher's questioning and, more generally, with the social norms that underly classroom interaction, these instances are treated as places where children display not only their knowledge and awareness of the social norms operating in this setting, but also 'reflexivity' in their interactional use (Stivers et al., 2018). As this article will show, pupils perform actions that are appropriate and sequentially relevant to the specific context they are acting in.

Background
Whole-class interaction, also known as 'plenary,' interaction, is one of the most common settings in which formal teaching takes place (Lyle, 2008;Mercer et al., 2009). In this setting, teachers produce initiating actions (Mchoul, 1978;Mehan, 1979;Sinclair and Coulthard, 1975), mostly information requests, through various forms of questioning that are alternatively designed to address pupils individually, the class as an ensemble of individuals responding in unison or as a community of individual speakers, each responding discretely (Ko, 2013;Lerner, 1993Lerner, , 1995Margutti, 2006;Waring, 2013).
According to Sacks et al. (1974), the primary constraint that questions pose concerns what should be done in the next turn: 'e.g., a 'question' making especially relevant 'answer' in the next turn' (p. 717). 5 Like other types of current turns in interaction (greetings, offers, request, assessments, invitations, etc.), of which questions are the prototypical example, questions project 'a relevant next activity, or range of activities, to be accomplished by another speaker in the next turn' (Heritage, 1984: 245). This 'sequential implicativeness' (Schegloff and Sacks, 1973: 296) characterizes the linkage between the two paired actions of which an 'adjacency pair' is composed (Sacks, 1987;Sacks et al., 1974;Schegloff and Sacks, 1973).
The normative character and conditional relevance (Schegloff, 1968) of questionanswer sequences have been demonstrated in cases where questions are not met by answers from the intended speakers (Heritage, 1984;Stivers and Robinson, 2006). In these so-called 'deviant' cases, the speakers' conduct provides evidence of their expectation that the question would be answered, despite delays, failures or the 'interactional difficulties' of the addressed answerers (Stivers, 2001). This evidence includes the pursuit of the answer by the questioner, accounts, or other non-answer responses by the answerer (Heritage, 1984;Zemel and Koschmann, 2011).
The notion of responsive actions that do not answer in accordance with the terms of the question, but nonetheless responding to its sequential requirements or constraints, is clearly described in Heritage (1984: 248-251). He illustrates a series of two-party interaction cases where 'second speakers respond to first pair part questions, but their utterances are not hearable as answering the question as put ' (p. 249). In multiparty settings, such as, for instance, classroom interactions, this notion can intersect with that of 'intervening actions' where 'other-than-the-addressed recipient speaks next' (Lerner, 2019). In other words, the relevance of the answer as the second pair part in the question-answer sequence can be evidenced by the fact that, under specific circumstances, unaddressed speakers can provide it. Prior studies on multiparty settings in ordinary and medical interactions demonstrated that, although in most cases addressed questions are responded to by the addressed recipient, another participant can provide the answer (Stivers and Robinson, 2006) when the addressed recipient fails or delays in answering. In this case, the conditional relevance of the answer overcomes that of the preference for the addressed respondent to answer, displaying the participants' preference for the progressivity in interaction. Consequently, these findings suggest that a first action might pose multiple constraints upon the next relevant action, depending on the contextual interactional circumstances.
The notion of multiple constraints in interaction was first addressed in Pomerantz (1978), with reference to compliments. Pomerantz shows that, following compliments, two systems of constraints operate at the same time: the preference for (i) avoiding self-praise and (ii) accepting and agreeing with the compliment are at odds with one another. If a compliment were to be agreed with, a recipient would be, at the same time, praising himself. And, since agreements and acceptances are co-implicated, the satisfaction of the acceptance preference is in potential conflict with the satisfaction of the selfpraise avoidance (p. 92).
Similarly, the questions teachers address to a single participant in the multiparty context here examined establish different constraints on the participants, in part owing to the special institutional inferences that characterize classroom interaction in general (Drew and Heritage, 1992) and, in part, to the multiparty setting of the specific whole-classroom organization. So, for instance, on one hand, by requiring the answer from the addressed speaker, these questions are designed to prevent the access to talk from other participants than the addressed next speaker. On the other hand, however, in this precise setting, they also project the answer as the relevant next activity (Sacks, 1987;Sacks et al., 1974;Schegloff and Sacks, 1973); thus, in favor of the progressivity of the ongoing interaction. Therefore, it might happen that these multiple constraints and preferences are at odds with one another. Regarding classroom activities organized as whole-class interaction, the presence of about 25 participants involves that all pupils become prospective possible next speakers. This may create the conditions whereby a question can be recognized and responded to before the addressed recipient is explicitly selected or when other circumstances provide the other-than-the-selected pupils with the right to gain speakership. The article investigates such circumstances, and the relative principles or constraints pupils display to orient to when answering or responding to questions that are not addressed to them. It shows that pupils engage in responsive actions that anticipate, accompany, or substitute that of the addressed pupil.
By focusing on a collection of addressed questions from two third-year classes in an Italian primary school, this article claims that these actions are the outcome of two main dimensions: (i) question turn design and (ii) the structural features in the organization of the activities that facilitate pupils' access to the answer. First, whether the address term precedes or follows the question unit in the teacher's turn is crucial to determining the options pupils have in anticipating the answer and in positioning and constructing their responses. As posited in Sacks et al. (1974), the rules that govern turn construction and allocation are ordered, each constraining the application of the other (p. 704). Therefore, to have the conditions for self-selection to apply (rule 1b), the current-speaker selection (rule 1a) needs to be accomplished before the transition-relevance place of the questioning unit. Accordingly, until the selection is not explicitly made, participants can legitimately treat the question as potentially addressed to any participant in the audience. In the specific context of primary school classroom interaction, it is argued that this circumstance encourages pupils to self-select and produce responsive actions. Second, this feature's potential for soliciting responses from individual unaddressed pupils is strengthened by some contextual features in the architecture of the pedagogic activity, which are related to how early and easily pupils can project the completion of the question, have access to the expected answer, or to whether the question requires only one or multiple legitimate answers.
Owing to these features, despite displaying their alignment with the formality of the speech-exchange system operating (McHoul, 1978), and with the teacher's allocating technique, non-selected pupils master local-allocational techniques of various kinds to acquire the right to speakership and perform different responsive actions; for instance by bidding for selection through vocal and visual resources (i.e. hand-rising, in Sahlström, 2002), claiming their knowledge of the answer or even by directly answering to the question.
These findings introduce a new perspective on classroom interaction in whole-class activities, describing a more flexible speech-exchange system for classrooms than that described in previous literature. It also provides evidence that pupils aged 8-9 master a refined interactional competence that enables them to gain the right to respond and contribute in many ways, negotiating their right to talk, even when others have been selected. More generally, this article contributes to the characterization of question-answer sequences and their different organizational orders.
This article builds on an earlier work (Margutti, 2021) that focused on pupils' conduct in Initiation-Replay-Evaluation (IRE) sequences (Mehan, 1979;Sinclair and Coulthard, 1975) in whole-class interaction, documenting pupils' interactional work in the sequence second position, and showing the complexities and competitiveness of their behavior. In that study it was argued that the pupils' responding conduct results from two conflicting identities: that of the 'knowledgeable recipient' (Goodwin, 1987) and that of the compliant recipient with the other-selection technique. By focusing on teachers' addressed questions and their responses, this article elaborates on these results with the aim of pinpointing the relationship that the features in the questions' design and in the structural organization of the pedagogic activity have with the pupils' responding conduct, and robustly characterizing the interaction in this setting.

Data and method
This study uses conversation analysis (Schegloff, 2007;Sidnell and Stivers, 2013). Recordings were transcribed according to the conventions adopted in the field (Hepburn and Bolden, 2017;. Each line is translated in English. When the idiomatic translation is not transparent, an intermediate line is added with a word-by-word translation accompanied with grammar glosses. Participants' names and other references to their private life were anonymized. The data consist of 156 questioning turns designed to select one pupil as the next speaker using addressed terms and other visual resources in the first 20-30 minutes of seven different classes, held by four teachers for two thirdyear groups (pupils aged 7-8 years). I selected activities in which all participants have the same access to the relevant information needed to provide the answer. This dimension is particularly relevant to the purpose of this study because it excludes techniques in which the next speaker is selected on the assumption of individual knowledge of the matter at hand or based on some types of shared and attributed social identities (Sacks et al., 1974: 718;Goodwin, 1987), which would have made naming the selected pupil or other forms of visual address dispensable and redundant.
Concerning the type of addressed questions investigated here, this article focused on a restricted collection of 104 instances. The main addressed tool is the pupil's proper name, either occurring before or after the questionable 6 in the information-seeking turn or as a self-standing unit. In the latter case, the proper name can be used either as a unique addressed form or in combination with/substituted by pro-term deictic references (second person singular pronouns tu/'you'), the directive dimmi/'tell me,' and non-verbal allocations (pointing, gazing and head tossing). I excluded repair techniques and addressed turns that can be used to accomplish actions other than information seeking, such as summoning one pupil to perform a demonstration, assist in setting an activity, perform requests for attention, reprimands, and invitations to comply with other noninformative requests.

Analysis
This section investigates the two main formats of teachers' questions separately, according to whether the addressed term precedes or follows the questioning unit. All the interactions here examined involve a request on pupils to process a list of items that are technically known and available to pupils. So, for instance, pupils are invited to classify a list of verb forms, name the components of the soil, and identify artificial and natural elements in pictures. This organization provides for questions in a series and, therefore, for an equal distribution of the answers among the participants ('equalization' of the answers, Sacks et al., 1974). 7 Furthermore, in all these activities, the information needed to arrive at the expected answer is directly and equally accessible to all pupils in the 'here and now' of the interaction, either because the activity aims to revise some tasks that pupils did before, or because the information is directly available on the materials used to accomplish the requested task. 8 In this way, all pupils have access to the expected answer.
However, as will be shown, according to the design and the terms of the question and other structural features of the activity, access to the answer can vary slightly either among pupils or within the activities, causing different interactional outcomes. This article shows that (1) the position of the address terms and (2) the accessibility of the answer are conditions for the non-selected pupils' responsive conducts to be brought forward or to prevail over the addressed pupil's answer.

The relevance of the position of the addressed term for pupils' responsive conducts
In this section two main forms of addressed questions will be presented, according to whether the address term is delivered before the question or not, together with their sequential consequences for the addressed and unaddressed pupils' conducts.
Format 1: When the selection precedes the questionable. In the first type of addressed questions the next speaker's selection precedes the question unit. One environment in which this practice frequently occurs is series of same-format questions (Levinson, 1992;Margutti and Drew 2014). In my collection, 44 of the 104 questions of the collection are in this format, of which extract 4 is one clear example. The teacher asks pupils to classify elements in pictures representing different inhabited environments (towns, countryside villages, seaside resorts, etc.) and to qualify them as 'natural' or 'artificial.' Since each request has the same format, with the teacher following the pictures' order, pupils have access to the general terms of the questions and to their succession order. By contrast, the precise terms of the question -whether the request is for an artificial or for a natural element, will be known at the completion of the teacher's turn. Owing to word limitations, I restricted my analysis to those aspects in the sequence relevant to the focus of this study. First, the question is a multi-unit turn, in which the selection (l.2) precedes the actual question (l.3); second, the turn design lets participants know whether the answer would be to name a natural or an artificial item only at the turn completion in line 3, with the production of the last word of the turn. The potential of this turn design to exert control on the audience is evidenced by the two gaps 11 at the teacher's turn transitions (ll. 4 and 6), 12 where neither Marco answers, nor any of the nonselected pupils bid for selection or produce any other type of response. Therefore, it can be observed that the sequence develops according to the IRE model: it is initiated by the question, which Marco finally responds to in line 7, followed by the teacher positively evaluating the answer in line 8. Having the teacher verbally selected Marco before asking the question, and though Marco does not immediately answer, no other student selfselects, either to make bids for selection or to answer directly, leaving the floor to the selected student. The next extract from a lesson on history is another such example. In this example, in line 3 the teacher selects Raffaele before formulating the question, occurring in line 7. Like in prior extract 4, though Raffaele does not answer promptly (see gaps in lines 9 and 11) and the teacher further increments the question in line 10, other pupils refrain from bidding for selection or other responses. In my corpus, considering only the questions with the address term before the actual questioning, those that are responded by the selected speaker only, like in the two extracts above, amount to a percentage that varies from 66% to 71%. This variance is due to the difficulties in coding the questions subsequent to a first addressed question as being addressed to the same recipient, especially when the teacher does not repeat the pupil's name or other addressing techniques. 13 In any case, these numbers provide evidence that the position of an explicit addressed term, accompanied with a delayed completion of the question unit, plays a relevant role in the management of a multiparty audience, especially if compared with cases where non-selected pupils produce intervening actions.
Format 2: When the selection does not precede the question. The extracts of this second section are from two lessons: an Italian grammar and one on natural sciences. In the Italian class, the teacher checks pupils' homework collectively; the task consists in classifying a list of verb forms from a reading passage. In the natural sciences class, the request is to remember one component of the soil, which pupils observed and identified in a lump of earth the previous day. Like the examples in the prior section, activities are shaped as series of questions/requests delivered in a round, through which the teacher asks for information that pupils are expected to have. In the format analyzed here, the difference is that the selection does not precede the questionable.
It is worth highlighting that the organization of the series of addressed questions slightly vary in the two lessons. In the Italian lesson the class works on a pre-established list of verb forms from a reading passage. The teacher asks pupils to apply the same procedure to each verb, addressing one pupil at a time in a round. Once the procedure is established, the mentioning of the verb form is understood as enacting the question each time. In the lesson on natural sciences, by contrast, there is no pre-established list of items; pupils are requested to produce it themselves, by remembering the components they identified the day before. In this context, the question is fully delivered only once at the beginning of the session; then pupils are selected through their personal name, or through pointing, gazing or directives (dimmi/ 'tell me'). With the productions of these addressed terms, the teacher refers to the original question.
As will be shown, the production of series of questions in the class does not always develops so neatly but can vary in relation to the circumstances that may arise. For instance, as we be shown, it is not rare the case that the design of the question changes both across formats and resources employed by teachers to meet the selected pupil's hesitation in answering or that of the teacher in making the selection.
In the Italian lesson, once the teacher assigns one verb form to a pupil, the task consists in identifying the present infinitive of the verb and finding out the conjugation it belongs to. 14 Three main structural features of this activity enhance pupils to arrive at the answer before the selection is made. First, all verbs are permanently available to pupils in the reading passage on their book; second, the teacher follows the order in which these verbs appear in the reading passage; and finally, one pupil (Fabrizio) is tasked with keeping the list by announcing the subsequent verb, before each new selection of the next speaker. In this way, at any question-answer pair, pupils are reminded of which verb will be the next and, potentially, assigned to them. However, vis-à-vis the established order of the verb forms, the selection of the next speaker is not programed in advance. There are 17 such cases of 104 in my collection, constituting the second group of the addressed questions, in which the selection is not in the first position.
Extract 6 below illustrates one clear instance. The example is the extended version of prior extract 2. The interaction starts with the teacher asking Fabrizio (the list keeper) to indicate the next verb (l. 1), which is then assigned to Riccardo (l. 4). Here the questioning is achieved by asking Fabrizio to tell the class which is the verb form (l. 3). For the moment, we won't consider the first question in line 1 and the response provided by unaddressed pupils (Ex. 2), on which I will return later. It suffices to say that by simply nominating the subsequent verb in the passage, each time the teacher refers to the terms of the question and to the procedure that is to be applied in the classification of the verb forms. 15 Crucial to our purpose, the verb form is explicitly offered to all pupils, before the selection, which is accomplished in l. 4. Therefore, it can be argued that the position of the selection after the questioning unit gives all pupils the opportunity to legitimately bid to answer until the next speaker is explicitly indicated. Other features in the teacher's turn design further delay the selection, offering pupils other options to produce unrequested and self-selected responsive actions. First, in line 4, the teacher constructs the selection as the result of an ongoing decision. So, for instance, the turn begins with the delayed token ah. The selection is accomplished through verbal and visual resources. The teacher produces a sentential unit ('lo fa (.) Riccardo / it will do it Riccardo'). The sentence has a marked format, with the subject ('Riccardo') dislocated to the right end of the turn, after a micropause; these features further postponing the selection. This provides Jasmine with the opportunity to deploy a bid for selection before the teacher completes the selection: she raises her hand in line 5, in the last-item-onset position (i.e. after the first syllable of the name 'Riccardo') Drew, 2009). In contrast with the examples of the first format, here all pupils are given access to the type of task before the selection is made and, as consequence, they can legitimately cast themselves as eligible next speakers, like Jasmine does here.
As we have seen, the teacher accomplishes the selection also by pointing in the direction of the selected pupil. In this regard, it is noticeable that Jasmine seats next to Riccardo; therefore, it can be argued that she might have interpreted the teacher's pointing (l. 4) as addressed to her. Also, note that the pointing precedes the naming. These aspects are relevant in determining the ensuing talk. In the inserted sequence (ll. 6-7), the teacher does not let pass unnoticed Jasmine's bid, in contrast to what teachers often do in response to unaddressed pupils' bids: he rejects it (l.6). Perhaps, this time, he did not let Jasmine's bid pass unnoticed because of the potential misunderstanding, due to the pointing. In this way, the teacher acknowledges Jasmine's bid, and indirectly treats his right to accomplish the selection as negotiable.
In terms of question constraints and of how they are associated to aspects of turn design, the way in which talk develops (from l.8 onward) provides further evidence of the potential of the first-format questions; that is, with the addressed term in first position. As mentioned, in whole-class instruction sessions such as these, it is not rare to observe changes in the design of questions, owing to contingent matters, as is the case here. In line 8, following the inserted sequence with the rejection of Jasmine's bid, the teacher needs re-addressing the question to Riccardo. However, being a second instance of same request addressed to same recipient, the teacher modifies the design of the question, resorting to the first format with the addressed term in first position. Quite significantly, although Riccardo does not answer, none of the other pupils offers any response. In line 14, the teacher himself provides the answer, after 1-second gap.
This example provides further evidence that the position of the address term has consequences for the pupils' responding conduct, posing specific constraints on what is understood as locally allowed to them in each circumstance. This, in turn, adds to the claim that pupils show an impressive level of understanding of the turn-taking rules operating in the class on a moment-by-moment basis and, also, of their ability to navigate the other-selection rule, to gain speakership through local allocational techniques.
In the next extract, I illustrate another example in which other-than-the-selected pupils self select to perform bids during the teacher's turn, before the next speaker is addressed. In line 1, as is typical in this format of questioning, the teacher first offers the verb form to be analyzed ('riflettevano'), then he selects the next speaker. Again, the selection is accomplished as a decision being taken on the spot, as indicated by the self-repair (Schegloff et al., 1977) of the name of the next speaker. In response to the teacher's turn, the first self-selected response by Jasmine occurs at the turn transition point, after the delivery of the verb form and before the selection. The second occurs in line 3, where St1 bids for selection after the teacher has named the first pupil ('Josef'), followed by an immediate self-repair instance ('no'). In the meantime, the teacher selects Jasmine, by naming her twice. This casts the subsequent St2′s turn in line 4 (io/'me') as a postselection action. Owing to its position (i.e. after the selection is made), and to its prosodic markedness, featuring high pitch and stretched delivery, 16 St2's 'me' can be heard as accomplishing a different action than bidding. The teacher's response to St2 confirms that he treats St2′s turn as a complaint for not having been selected. In line 5, the teacher softly re-asserts the choice of Jasmine as next speaker, gazing at St2. In doing so, he acknowledges St2's conduct and provides an implicit account for the selection. The sequence then develops with Jasmine answering (l. 6) followed by a further IRE sequence, with Jasmine as the main and undisturbed recipient (ll. 8-10).
These two examples show that addressed questions, in which the selection does not precedes the question, make relevant and legitimate earlier bids for answering, as compared with those in which the addressed term precedes it; thus, promoting the pupils' negotiation of the next speaker selection and potentially challenging the teacher's exclusive right to make the selection. By designing the question in this way, the teacher gives access to the question in a context where the selection is not operating yet. Against any prevision that the teacher will continue with the next-speaker selection, pupils promptly exploit this opportunity to self-select on a local basis, proposing themselves as knowledgeable next speakers. The position in which their bids are deployed show an impressive level of understanding of the other-selection allocation technique in the classroom setting and their ability to access speakership.
The next extracts are from the class on natural sciences. In this activity, pupils were asked to remember one component of the soil they observed the day before. As mentioned before, the teacher delivers the question first, soliciting bids for selection, and then selects each pupil in turn. Again, this technically provides pupils with equal access to the terms of the answer before the selection is made. The fully formed question is delivered only at the beginning of the activity, as shown below (ll. [1][2][3][4][5] It is worth observing that this specific question makes relevant and legitimate multiple candidate answers, which pupils provide through self selection. The teacher then solicits pupils to raise hands (data not seen). Subsequently, pupils are selected through various self-standing addressing terms such as their name, second person singular pronoun (tu/'you,' second person pronoun singular), and invitation to speak (dimmi/'tell me'), accompanied by pointing, gazing, head tossing, or through a combination of these resources.
The teacher's selection follows the order in which pupils' seats are arranged. When selected, pupils need to provide their own answer, based on their recollection of what they did the day before. As compared to the examples in the Italian class, it can be argued that here pupils have more unconditioned access to the answer, whose knowledge depends on whether they remember it. The fact that the teacher follows the seating arrangement in the class to make the selection gives to the organization of the turn-taking system a higher degree of pre-allocation. Arguably, pupils can project that the otherselection rule is operating at any question-answer pair and that sooner or later each one will be given the opportunity to answer. This form of questioning occurs 43 times of the 104 occurrences I collected. However, in contrast to the projection of the equalization of turns associated to the other-selection technique adopted, in almost all instances, otherthan-the-selected pupils take the turn to bid for selection, claim their knowledge of the answer and, sometime, directly answer to the question without waiting for the teacher's selection. Despite this behavior can be described in lay term as 'out of order,' in what follows it will be shown that, by contrast, it responds to a precise interactional order.
The example below is a clear case of the way in which pupils typically self-select in questions of this format to bid for the turn and to answer (see < lines), without waiting for the teacher's selection. The question-answer sequence begins with the standing-alone address term in line 3. The teacher's gaze and pointing gesture accompany the naming of the selected next speaker.
In terms of question design, the teacher's turns are very brief and occupied by one single unit, often consisting of only one word (the name of the selected pupil, the pronoun tu/'you' or the directive dimmi/'tell me'). These features make quite hard for the not-yetselected pupils to position their turns before the selection is made. As shown in the lines marked with the greater-then sign (>), pupils willing to be selected as next speaker apply a new self-selection technique: they anticipate the position of their responsive actions (bids and answers) by climbing back in the IRE sequence. For instance, in line 7, St1 begins a series of vocal emissions (working as bids for selection) in overlap with the teacher's evaluation of the prior answer; this position being well in advance with respect to the moment where the next selection is due. However, it is to be considered that the position right after the endorsement of the given answer is indeed the first possible place where recipients can understand that the interaction will move to the next item. This behavior evidences the pupils' knowledge of the principles underlying the turn-taking system here operating and their ability to navigate them, showing that children of this age not only know the social norms on which interaction is based, but also reflexively use their competence. Significantly, this understanding and ability is shared by other pupils who produce their bids in lines 8, 9, and 11, and by St4, who directly gives a candidate answer in line 10. The extract shows that, although the organization of the activity projects and possibly provides for the equal distribution of the answers, rather than complying with the teachers' programmed turn allocation and waiting for the selection, when the terms of the questions are accessible from the beginning, pupils adopt local allocation techniques and apply the 'starting first' rule. This example also suggests that the orientation to answering rather than complying with the teacher's selection might be stronger when pupils have a facilitated and direct access to the information that is needed to answer. In this specific case, the facilitated access to the answer is granted by the fact that the question has multiple legitimate candidate answers and, furthermore, that answers need only being recollected from the activity they did the day before.
The analysis so far has shown that the position of the addressed term in the design of teachers' turn with respect to the questioning unit is a relevant feature for the organization of the unaddressed pupils' responsive actions, displaying their ability to understand and navigate the turn-taking system operating in whole-class interactions. However, this feature does not provide the entire story. As shown in extracts 1 and 2, despite the selection occupies the turn-initial position in the question, other-than-selected students bid for selection (Extract 1) and even answer to the question (Extract 2). In the following section, I will provide an account for these cases, expanding upon the other dimension that is determinant in the production of pupils' responses: the accessibility of the answer.

The relevance of answers' accessibility for the unaddressed pupils' responsive actions
In section 1, I have shown that questions with the addressed term in first position (format 1), combined with a question design that delays the understanding of the terms of the question to the end of the turn (extracts 3 and 4), can lock down the sequence from other-than-the-selected recipients' intervening actions. The relevance of the position of the address term with respect to the question unit in teachers' turn is further confirmed by the evidence that in the format where the question is accessible to pupils before the selection is made (format 2), pupils produce responsive conducts such as bids for selection and candidate answers more frequently and earlier than in response to questions shaped as format 1.
Evidence provided by this conduct is twofold. On one hand, through the design of their turns as bids for selection, whereby they hold off answer-type responses, pupils show their understanding of the other-selection rule as fully operating. On the other, if we observe the position of pupils' self-selected incoming in the teacher's question turn, and further up in the prior IRE sequence, it can be also argued that their compliance to the other-selection technique does not prevent them from exploiting all possible opportunities that the turn-taking system offers to gain speakership, also when the organization of prior talk provides robust basis to expect a distributed speakership among all participants (Extract 9). Furthermore, as observed in prior extracts, with slightly variations in the design and position of these bids, pupils manage to convey actions (such as complaints) that can open a negotiation of the teacher-select-next technique (Extract 6).
In what follows, the issue of the accessibility of the answer and, more precisely, the epistemic gap between selected and non-selected pupils will be discussed. In this regard, extracts 1 and 2, which are considered as 'deviant' cases of format-1 questions, will be explained. Despite the address term precedes the question unit, other-than-the-addressed pupils respond. It will be shown that the epistemic gap between addressed and unaddressed in accessing the answer is responsible for the preference for the answer and the progression of the activity, over the other-selection rule.
In what follows I will show that format 1, which so efficiently controlled the incomings of unaddressed pupils in extracts 3-5, has different sequential consequences. Extract 10 is from the lesson on geography where extracts 3 and 4 are from, where pupils are requested to name artificial or natural elements in pictures. The extract illustrates the third question in the series from the same lesson. Following the teacher question (l.1-3), Elio, the selected pupil, displays hesitation. The questioning unit (l.3) has the same format of previous extracts 3 and 4: [Pupil's NAME + can you tell me X]. In the following analysis I will focus only on what occurs after line 4; that is, after the first possible completion of the teacher's turn, where the answer from the selected speaker is expected.
The 5-second gap in line 4 indicates that Elio has some problems answering. In line 5, St1 raises her hand, tacitly asking permission to take the turn. In lines 6 and 8, ignoring St1, the teacher pursues the answer from Elio, reformulating the meaning of the last word in line 3. After the first inaudible answer from Elio (l. 7), in line 8, the teacher re-proposes the request with a two-unit turn. This obtains a response from Elio in line 9: he starts with a delay token (e::/'uh::'), followed by the recognizable beginning of a sentence unit that he self-repairs with a cut-off before completion. 17 After Elio's first delay token, in overlap with the subsequent incomplete turn unit, in line 11 St2 raises her hand too.
The timing and the format of St1's and St2's conduct shows that they treat Elio as the primary recipient: by allowing a substantial amount of time before bidding for selection, they treat Elio's right to answer first as still operating, showing a preference for the selected speaker to answer (rule 1a, in Sacks et al., 1974). In this, the example confirms the potential of the design of the questioning with the address term in first position to curtail the access to speakership of non-selected pupils and maintain the other-selection technique in the context of series of same-format addressed questions delivered in a round. However, it also shows that pupils attentively monitor the interaction on a moment-by-moment basis, ready to exploit any option that would allow them to gain the right to speak.
As will be shown in the following extract, in questions of this format, non-selected pupils obtain the right to self-select only when the primary recipient shows that s/he has not access to the answer. Under these circumstances, unaddressed pupils exhibit a distinct awareness that the other-selection rule can be negotiated and that they can gain speakership through the management of self-selection. In other words, when a knowledge gap is documented in talk, they display to be aware that the other-selection rule is operating while, at the same time, they show an orientation to a preference for the answer; thus, fostering the progressivity of the activity. Pupils show to be projecting that the teacher's next action will consist in a new selection, for which they propose themselves as eligible next speaker, as illustrated in the next example.
After extract 10, Elio gives another wrong answer (transcript not seen). Extract 11 below starts with the teacher commenting on this answer: she repeats her original question framed with the reported speech ('I told you. . .') and followed by a reformulation of the word umanizzato ('humanized'). Ex. 11 [PM:FZ:12b: geography] Ships  (Fig.2) ((T. withdraws gaze from elio to look at the book for a few tenths of a second, to raise it again, to look at elio again)) 6 St1: > [↑io! m'estra me teacher ((stretches her arm further towards the teacher)) 7 st2: > ((stretches her arm further towards the teacher)) 8 St2: > io >m'estra IO LO SO!< me teacher I know it 9 st3: > i: [i ((not  In line 1, the teacher is still addressing Elio. This is indicated by the personal pronoun (ti/'you'), by the repetition of the original question framed in the reported speech format ('I told you . . .'), and in her sustained gaze on Elio, who sits in the second row, behind St1 (not visible in the camera). During the delivery of the question, the two girls who had silently raised their hands before (extract 10, ll. 5,10) maintain their posture ( Figure 1). Note that both pupils hold the raised arm with the other, displaying the effort of maintaining the position over time. Through this posture they display a preference to be selected next, because of Elio's prior and projected failure. These gestures and postures will significantly change in line 5, when the teacher delivers the subsequent question. In line 4, Elio provides the answer to the question. After having endorsed Elio's answer, in line 5, the teacher makes a new request. This question has a different format than the preceding ones. First, it is constructed as an increment to the prior (Schegloff, 2016): 'in addition to houses.' Second, the teacher does not use any verbal address term explicitly directed at Elio. These aspects can account for the understanding of the question as initiating a new sequence and, with it, a change in the addressing technique, opening the floor to the class. On the other hand, the fact that the teacher designs the questioning as an increment to the evaluation of Elio's answer, and that she did not nominate any other pupil, together with the direction of her gaze in Elio's direction when delivering the request (recall that Elio seats behind St1, not visible in the picture), all these cues might as well indicate that the questions was addressed to Elio. At the same time, however, the turn lacks any explicit term indicating Elio as the next recipient. Also, recall that here, like in extract 6, because of the closeness in the arrangements of pupils' seats, gazing and pointing can be subject to ambiguity and misunderstanding as compared to the name of the recipient. The way in which the interaction unfolds displays that pupils understand the question as projecting a new selection (lines 6-9) or even opening the floor to individual self-selected answers (line 10).
That the teacher's question in line 5 triggers a change in the participation framework is visible in the more active involvement from the other pupils. One immediately visible consequence is how St1 and St2 modify their hand-raising gesture. Recall that St1 and St2 had maintained their hands up since they first raised it (Ex. 5). Here (ll. 6-7), overlapping with the last syllable of the teacher's last word 'houses' (in the last-item onset position, Drew, 2009; and following it (l. 8), they both cease to prop up the raised arm and stretch it further toward the teacher (Figure 2), adding verbal behavior. St1 coordinates her gesture with the verbal bid in line 6, while St2's verbal conduct is slightly delayed (l. 8). 5 T: ". . . in addition to houses" In comparison to the onset of the hand rising in extract 10, when it was first produced, here the arm stretching occurs in a more advanced position in the questioning: St1 and St2 do not wait for any form of hesitation from Elio but strengthen the stretching of their arm in overlap with the teacher's request. Other pupils also display the understanding that a change in the participation framework has occurred. In line 9, St3 makes a verbal bid to answer in the shape of a designedly incomplete response: he pronounces the first vowel of the personal pronoun io/'me,' a very common type of bidding in primary schools. Finally, despite not being selected, St1 produces a candidate answer in line 10, which the teacher silences in line 11.
The analysis of these two extracts supports the evidence that, in response to questionformat 1, pupils tend to orient to the preference for the answer produced by the addressed speaker. However, once the answer is delayed, incomplete or unclear, non-selected pupils display the awareness that this contingency offers them opportunities for speakership, as shown by their responsive actions, including direct answers to the questions. These responses are accomplished on a self-selection basis, according to the 'first starter gets the turn' rule (rule 1b, in Sacks et al., 1974: 718-719).
We return now to the initial puzzle, concerning cases like extracts 1 and 2. In these examples, format-1 questions receive responses from unaddressed pupils, without any visible hesitation from the selected recipient. These conducts can be explained by referring to the knowledge gap in accessing the answer between the selected speaker and the other pupils, in favor of the latter.
Let us now reconsider extract 1, where the teacher addresses the question to Caterina at the end of the repetition of a demonstration on angles. Recall that Caterina was absent the day before, when the demonstration was done for the first time.
Ex. 12 (former ex.1) [PM:LT:5a: geometry] Angles The fact that the demonstration is repeated for her benefit necessarily casts the other pupils as naturally more knowledgeable of the answer than Caterina, prompting them to bid for selection (< lines). Similarly, in extract 2, other pupils respond chorally (l.2) to the question that the teacher has addressed to Fabrizio.  The delivery of the question in line 1 has prosodic features that characterize it as a repetition of a prior occurrence: note the stretched sounds in the final vowels of the name and of the verb. This behavior, and the pupils' choral answers can be explained by looking at the talk immediately preceding it and illustrated in the following example. The teacher asks Fabrizio to announce the next verb form (ll. 1, 3), but Fabrizio does not answer (ll. 2, 7). As glossed in the transcript, Fabrizio does not respond because he does not have access to the answer, being busy away from his desk, where the reading passage is placed. Under these conditions, considering that everyone can check on the reading passage on their own desk the list of verbs, other pupils have access to the answer, while Fabrizio does not. This knowledge gap provides unaddressed pupils to satisfy the requirements of the question and provide for the activity to progress. The analysis of this section strengthens the claim that these responsive conducts show that children of this age display an impressive level of understanding of the turn-taking system operating in whole-class interaction. Furthermore, data evidence the pupils' competence in navigating the other-selection allocation technique, and their ability to monitor when and how local allocational techniques are applicable on a moment-by-moment basis, depending on two main dimensions: the design of the question and whether they have primary access to the answer.

Discussion
This article investigated a diffused phenomenon in everyday primary school classes: in the context of whole-class interaction, following teachers' addressed questions, other non-selected pupils produce responses of various kinds, such as bids to answer, claims and displays of knowledge, comments, complaints, and direct answers. The article moved from two main observations. First, the other-selection allocation rule technically provides for an equal distribution of answers among the large number of recipients in the class, orienting to the orderly participation from the non-selected pupils. Second, by contrast, autonomously produced bids for selection and other responsive actions from the other-than-the-selected pupils seem to disregard the projection of the equal distribution of turns and be caused by a supposed lack of interactional competence. The article demonstrates that, far from indicating the pupils' low awareness of the underlying social norms, these conducts are very finely and orderly organized. They reveal the pupils' deep understanding of the constraints posed by the teacher's addressed questions in that specific context and of their multiple-preference principles, showing the pupils' impressive ability to navigate them.
In line with the claim that preference organization, which is responsible for the production and understanding of initiating and responsive actions, is the result of multiple principles, often conflicting (Pomerantz and Heritage, 2013), the first finding is the evidence that unaddressed pupils follow at least two implicit principles in their responsive conducts: (i) the preference for the other-selection technique, whereby the addressed pupil is the primary recipient to respond and (ii) the preference for the answer and the progression of the activity, whereby non-selected pupils can proffer themselves as respondents or even answer directly on a self-selection basis. Regarding the former principle, and in line with previous analytic conversation studies (Stivers and Robinson, 2006), the article shows that the pupils' preference for the addressed pupil to answer is weakened in favor of other pupils' self-selection when the answer is not forthcoming, hesitant, uncertain, or wrong. As for the latter principle, the article provides an interactional account for the unaddressed pupils' self-selected responses, produced according to the 'first starter gets the turn' technique (Sacks et al., 1974: 718).
The second finding concerns the identification of the two main features in the design of the question and in the overall organization of the interaction that are responsible for this phenomenon. The article demonstrates that the two responsive outcomes indicated above are associated with (i) whether the addressed term precedes or follows the question unit and with (ii) who has the primary access to the answer. Therefore, in questions where the selection prefaces the questioning, other pupils tend to refrain from a prompt reaction, allowing time for the selected speaker to answer in compliance with the teacher's allocation of the next turn. By contrast, pupils tend to take advantage of the so-far unaddressed question when the selection is postponed, engaging in a series of self-selected actions produced according to more local allocational techniques. Quite significantly, unaddressed pupils' selfselection and their premature production of responsive actions tend to occur in activities where the class has a facilitated access to the answer. As shown, the class's easier access to the answer can be obtained through aspects of turn design of various kinds, such as the position of the address term and other features, entailing how early in the teacher's turn the questioning unit reaches completion. Furthermore, access to the response can also be facilitated through the local and contextual aspects in the organization of the activity, such as the physical access to books, pictures, and other materials. In these cases, at the question completion point, or often in overlap with it (Drew, 2009;, not-(yet)-selected pupils engage in responsive actions of different kinds, including complaints, claims for their right to speakership, and even direct answer to the question. Furthermore, these responses might result in more complex sequences where the selection is accounted for or even negotiated. Based on these findings, this article argues that the knowledge of the answer is associated with and promote the pupils' orientation to obtain a turn.
The claim that the epistemic gap between addressed and unaddressed recipients has relevant sequential consequences for the ensuing responses, to promote a preference for the answer and for the progressivity of the activity, is confirmed by the analysis of two 'deviant' cases. In these extracts, non-selected pupils bid for selection and directly answer the question, despite the teacher has already done the selection. In both examples, the non-selected pupils have a privileged access to the answer, as compared to that of the selected next speaker.
These findings document the pupils' refined interactional competence and awareness of the social norms underlying classroom interaction and its institutional functions. Their behaviour shows a robust orientation and compliance to the teacher-select-next rule; however, at the same time, the fine details of turns' design and sequential deployment suggest that pupils also monitor the interaction moment by moment, to exploit any opportunity to gain speakership. These findings argue for a reflexive awareness of the social norms underlying classroom interaction in children of this age; that is 'an awareness of acting in accordance with, or departing from, a relevant norm' (Stivers et al., 2018: 16).
This account of the mechanisms that link teachers' questions and pupils' responses in whole-class activities radically revises those aspects of the formality of classroom interaction, which have been previously described as substantially uniform (Mchoul, 1978), with pupils having very limited interactional activities, and substantially complying with the teacher's selection of the next speaker. Rather, it offers a radically different picture; one in which pupils have a more active interactional role.
Finally, it contributes to the investigation of question-answer sequences and to the multiple preference principles they are subject to, as well as the specific constraints they pose in the setting here analyzed. The article provides further knowledge on how mundane conversational practices are specialized and specified to enable participants to accomplish their institutional actions in the setting where they are engaged.

Declaration of conflicting interests
The author declared no potential conflicts of interest with respect to the research, authorship, and/ or publication of this article.

Funding
The author received no financial support for the research, authorship, and/or publication of this article.