On Designedly Incomplete Utterances: What Counts as Learning for Teachers and Students in Primary Classroom Interaction

The article analyzes a variety of Koshik's (2002) Designedly Incomplete Utterances (DIU) as they are produced in whole-class, teacher-led instruction sequences held in 2 third-year groups in an Italian primary school. This device, one of whose basic pedagogic functions is to solicit displays of knowledge from students in the shape of utterance completion, is a recurrent feature of teacher–student interaction in this setting. The study focuses on one specific and locally managed use of the device, whereby the teacher's orientation to the pedagogic goals of the organization of interaction surfaces in features of talk. I found systematic features in the construction of what I call main-clause DIUs, which teachers recurrently use to cast students as learners, by treating their verbal behavior as providing evidence that some type of learning has occurred in prior talk. The findings provide grounds for a critique of the Initiation-Response-Evaluation model and for a characterization of questioning in instruction sequences, both of which account for the specific institutional relevancies of interaction in this setting.

of what teachers and students actually achieve in terms of the specific institutional relevancies of their interaction (Drew & Heritage, 1992, p. 15). For instance, one major issue for both partiesand for those who take part in any instruction activity-is the identification of and achievement of consensus as to what exactly counts as learning: that is, how teachers and students make their behavior accountable by showing it to be oriented to a particular learning process that has been implemented through the organization of their interaction. Another closely related issue concerns a more specific and cogent characterization of the actions that are performed through questioning; teachers ask questions frequently, and so describing their function as eliciting displays of knowledge is far too generic.
This article analyzes one specific way of questioning students in instruction sequences. This form of questioning first came to attention as a form of word-search in Lerner (1995) and was crystallized as Designedly Incomplete Utterances (or DIU as it will be henceforth referred to in the article) by Koshik (2002). It consists of incomplete utterances produced by teachers as a means to elicit missing information in the shape of utterance completion. Koshik has described one particular use of the practice, as it occurs in one-on-one second-language writing conferences to elicit the student's self-correction. In this article I will investigate another use of DIU, as it takes place in whole-class instruction sequences in an Italian primary school. I am concerned here with a type of instruction activity known as the talk and chalk method, whereby teachers involve the whole class in a special type of "language game" (Levinson, 1992, p. 66;Wittgenstein, 1958, aphorism 7), specifically designed to help students acquire new forms of knowledge through a series of linked question/answer pairs, starting from students' implicit knowledge (Mercer, 1995).
I describe these sessions of talk as having an argumentative line of reasoning because, as the analysis will show, in these instruction sequences the teacher designs questions that are closely related to one another, so as to build an argument. 2 Rather than constructed as one single unit, with its own individual scope and somehow disentangled from the surrounding talk, each question/answer pair is designed to lead to the next, so as to build a line of reasoning that will gradually guide students toward new forms of knowledge. This way of approaching classroom instruction talk contrasts with the characterization of question/answer pairs in instruction sequences as self-contained units, each reaching its conclusion with the teacher's third evaluative turn, as is the case, for example, with examination questions.
Using Conversation Analysis, I will investigate the specific pedagogic relevancies of DIU as one specific questioning practice very commonly used by teachers in this setting. In order to give the reader an idea of what DIUs look like in my data, I have chosen the following example as one explicit case of the phenomenon. The excerpt takes place during one instruction sequence in a Geometry Hour. 2 The type of pedagogical activity I am concerned with here is very similar to the "language game" described by Levinson (1992). In his analysis of classroom data, Levinson shows that series of linked questions can be used by teachers to accomplish two distinct pedagogical aims: (a) to make a "request to follow a procedure" or (b) to "make explicit a selected part of the implicit knowledge that the teacher assumes pupils to have" (pp. 86-92). Instructions sequences that are the focus of this study are of the second type. According to Levinson, this management of the questioning activity by teachers is similar to the way in which questions are used in courtrooms. He argues that in both contexts questions "extract answers that would amount to a specific argument" (p. 91). In this regard, see also Drew (1992, p. 472) for a characterization of the "line of questioning management" in courtrooms.

317
The excerpt isolates one of the questions of the series because, for the moment, I am interested only in pointing out to the reader some very basic features of the phenomenon that make it recognizable as such and that are shared by all similar occurrences in my data. The characterization of the phenomenon as far as its sequential deployment and its pedagogic function are concerned will be developed in later sections of the article.
Prior to Excerpt 1, the teacher has invited one girl, Caterina, to walk around the desk in order to show how many times one has to change direction before completing the full round. This is done for the purpose of demonstrating that the number of angles on the surface of a shape (here the desk) is correlated to the number of changes of direction (here Caterina's changes of direction when walking around the teacher's desk). Throughout the whole duration of Caterina's demonstration, the teacher engages the class in commenting on the performance by addressing the students with a number of linked questions, of which the following is one instance: As the transcript makes clear, the teacher's turn that begins in line 1 presents a cluster of prosodic features (intonation, pauses, sound-stretching), which culminates in the premature halting of the turn before its grammatical completion (line 3). This practice is evidently taken by students as an invitation to supply the missing word. By providing the word "angles," students display (a) that they are paying attention to the teacher's talk, (b) that they know what the teacher is talking about, (c) that they know the technical name of that specific part of the surface of the teacher's desk, 3 and (d) that they are willing to take part in the activity underway. Thus, by withholding in this way the last word of the utterance, the teacher elicits a display of knowledge in the shape of utterance completion. Questioning students in order to elicit displays of knowledge, which teachers confirm or correct, is a type of activity that occurs frequently in instruction sequences. Both characterizations of the activity-questioning and eliciting displays of knowledge-are far too generic and do not provide an account of the specific pedagogic relevancies of classroom interaction as related to its instructional goals. DIUs like the one occurring in line 3 share with other classroom questioning devices the known nature of the missing information (Mehan, 1979a). But it is exactly because teacher and students are oriented to this quality of the action (as evidenced by the teacher's repetition of the students' response in line 7), that classroom questions simultaneously make other actions relevant, besides eliciting information. These actions include checking students' knowledge about a specific topic; setting topics as more or less pedagogically relevant; highlighting key notions; inviting preferred responses both in terms of content and shape; conveying assumptions about the students' state of knowledge; doubting or accepting prior answers; initiating repair; and also achieving, maintaining, and sustaining students' participation in the activity underway. This is an issue that has been somehow neglected in studies on classroom interaction, 4 probably owing to the success of the IRE model (Mehan, 1979a;Sinclair & Coulthard, 1975), which has favored a certain atomization of classroom interaction into isolated exchanges.
Through the analysis of (a) the specific design of one variety of DIUs and of (b) the way in which occurrences of this type are linked to other questioning practices in a series of questions, I will identify the distinctive functions of the device, demonstrating the way in which DIUs reflect pedagogically relevant aims in instruction. Thus, starting with the analysis of one of the questioning devices that is more frequently used by the teachers in my data, the article aims at dissecting the IRE model, which has long been considered the molecule of instruction talk. The way in which the practice is constructed and deployed within a series of questions will disclose patterns of larger sequences of instructional talk, and the way in which teacher and students are oriented to them in relation to the specific pedagogic goal of their interaction.
The study is based on a specific subset of DIU, which I have called main-clause DIUs and of which Excerpt 1 is a clear example. This characterization relies on the fact that these special types of DIUs are produced to elicit completion of self-standing, full-formed main sentences. Their construction differs from that of other subsets of DIUs with different functions, as is the case, for example, of DIUs embedded in secondary clauses or of those that follow prior unanswered questions, neither of which I have space to illustrate here.
I have called this variety of the device main-clause DIUs, owing to the type of syntactic units on which it operates. I will show that the device is used to highlight some key notions in the argumentative line of the activity. DIUs occurring in main clauses are also built to be successfully completed by means of a special mechanism systematically used in this subset. Due to their sequential deployment, completion is designed to be a repetition of a prior occurrence of the 4 A few decades ago, with the rise of the progressive trend in teaching, this practice, as well as other ways of questioning students, has been prey to strong criticism. Questioning students with incomplete utterances or with other close-ended questions was negatively judged because they were considered a way of constraining students' response to predetermined answers already contained in the questions (for a short description of the controversy, see also Mercer, 1995, pp. 28-32). In this regard, it is worth highlighting also what Atkinson and Drew wrote in 1979 about similarities between the organization of talk in courts and classrooms concerning some possible constraints deriving from their multiparty settings (pp. 220-228). See in particular footnote 8, about the way in which some organizational aspects of talk in classrooms have been considered "unpleasant" features by some narrow and one-sided critiques. same item that is to be completed through the DIU. Both features-the repetition format and the request to complete an unfinished main clause-give salience to the notion on which the DIU operates, as compared to the information elicited through other questioning devices within the instruction sequence. I argue that DIUs of this type construct these notions as the teaching goal of the activity and demonstrate that both parties (teacher and students) show in this way that they have reached consensus as to its importance within the whole instruction sequence. Furthermore, because the completion task is a repetition of a prior occurrence of the teacher's own talk or of another student's prior correct answer, with these DIUs teachers construct students' responses as evidence that their knowledge-displays occur during and by means of the talk underway, and not as the result of some prior implicit knowledge. Through the way in which the DIU is constructed, students' responses are treated as evidence that some kind of learning has taken place then and there and as the product of talk.

DATA
The research draws on video-recorded data from lessons that took place in 1999-2000 in two third-year groups (ages 7-8) in an Italian primary school where I was working as a teacher (nevertheless, none of the recordings were taken in the group I used to teach). The recordings lasted 1 week (8 hours a day) for each of the two groups, and involved around 50 pupils and their four teachers. Teachers adopted teacher-led methods of instruction and forms of classroom management. For most of their lesson time, pupils were seated in parallel rows, with the teacher facing the class. Two cameras were used in each classroom in order to capture the interaction from both points of view. The corpus consists of approximately 80 hours and about 40 lessons on different topics, which were not shaped for the purpose of the study. Data collection was approved by the participants and their families, who provided informed consent recorded prior to the study. All the names used in the transcriptions have been anonymized.

DIU: BASIC FEATURES AND PEDAGOGICAL FUNCTIONS
Since Schegloff's article on the ambiguities of the notion of "question" in conversation (1984), the term has to be used with particular care (but see Steensig & Drew, 2008, for an outline of the principal research traditions on questions and questioning). In its vernacular meaning, with reference to classroom interaction, the term question is normally used to indicate a set of practices that teachers adopt to elicit displays of knowledge from students in the shape of verbal responses. 5 By describing the teachers' questioning activity as a way of eliciting knowledge displays from students (rather than asking questions), two main aspects of the questioning activity are highlighted: First, teachers' initiating actions are separated and distinguished from the implicit and erroneously exclusive identification between teachers' questions and the interrogative format. Second, these actions are characterized as a particular type of questioning. 6 In their study on the English spoken by teachers and pupils in the classroom, Sinclair and Coulthard (1975) have provided the most basic characterization of teachers' action of questioning students: "A teacher rarely asks a question because he wants to know the answer; he asks a question because he wants to know whether a pupil knows the answer" (pp. 36-37). This fundamental difference between real and instruction questions has been addressed by Mehan with his oft-cited definition of the latter as "known information questions" (Mehan, 1979b, p. 285), which explicitly points to the knowledge-issue in the action of questioning.
This specific work of drawing out from students the right answer to their question has been variously described in literature as "cued elicitation" (Mercer, 1995, p. 26) and"cluing" (McHoul, 1990, p. 355). Through this practice of "wording a question in a certain way" (Mercer, 1995, p. 26) teachers "lead students to correct answers by small steps" (McHoul, 1990, p. 355). DIU is precisely one such distinct way of wording a question, whereby the questioner both exhibits that he knows the answer and provides clues as to which answer is required, in terms of turn design, content, and sequential deployment.
Before addressing main-clause DIUs, in what follows I will describe the basic constructional features of the device and some of the pedagogical functions it accomplishes in instruction sequences. I will also illustrate how DIUs differ from interrogative utterances and what teachers can do with them that cannot be done with syntactically formatted questions.

The Phenomenon's Basic Features
In the type of instructional activities with which I am concerned here, teachers elicit utterance completion very frequently. The request is implemented through prosodic patterns and specific features of turn construction. Excerpt 2 illustrates a further clear occurrence of the phenomenon (see arrowed lines): The excerpt captures the interaction at the beginning of a History Hour. In introducing the topic of the activity, the teacher refers to the content of prior lessons. However, the utterance is not produced in a complete manner by the teacher; in line 2 she withholds completion after the compound preposition nel and before the last word, which would bring the utterance to completion. 7 It is worth knowing that the basic Italian word order of sentences does not allow closing the sentence with a preposition. Thus, when the teacher says nel, any native speaker of Italian would expect a noun or a noun phrase (masculine and singular) to follow, which would bring the sentence to its grammatical completion. 8 In this case, the proposition nel links the verb ("to enter") to a period in prehistory (the Neolithic period), here metaphorically treated as a physical place.
In this context, the pause in line 3 functions as an implicit request for completion. Indeed, at its most basic, the pause in line 3 indicates failure on the students' part to provide the word that would bring the utterance to completion. However, the fact that the teacher allows a 1-s pause indicates that she designedly withholds the progression of the turn, expecting the answer. This meaning appears more evident if we compare the example in Excerpt 2 with the series of "failed DIUs" in Excerpt 3. In Excerpt 3, which is from a geometry lesson on angles, the teacher is commenting with the class on a demonstration performed by a student. She uses DIUs frequently in this commentary. After the open question, the teacher begins to give the answer, produced with a rush-through. But she produces a cutoff soon after the first item of the answer, thereby allowing the first pause. 7 Italian prepositions can be linked with articles in one single word (in Italian they are called preposizioni articolate), and this is one such case. The simple preposition in (in English "in") is here linked with the determinative article il (masculine and singular) to form the preposizione articolata (compound preposition with article) nel. 8 In Italian some adjectives can be used also as nouns, as is the case, for instance, of adjectives of nationality. For example, the word italiano can be used both as an adjective and as a noun. In the latter case, the noun can be the only component of a noun phrase. The word neolitico here has the same properties: It is used as a noun and stands for the noun phrase il periodo neolitico ("the Neolithic period").
In the 0.8-s silence that follows, students do not provide any continuation of the turn. Therefore, it is the teacher who resumes her own talk. But in the remaining part of the turn she stops again twice with a distinct rising intonation, after "fa::re? (.)" and after "un altro? (.)", thus inviting students to step into the answer space again to provide completion. This time, pauses are shorter (micropauses) and, again, they are unanswered. All three occurrences can be described as failed DIUs; by not allowing more time for the silence before continuing her talk, the teacher seems to display that she considers students unable to provide the completion of the answer. It is precisely on this basis that the characterization of the pause in DIUs can be rather controversial. In fact, it is designed as an intra-TCU silence (that is, deployed in nontransition space) precisely in order to make recipients hear that the utterance is not finished and, thereby, elicit completion. The sense of incompleteness provides for turn transition by way of TCU completion, which, subsequently, makes the pause into an interturn pause (however, for clarity they will be referred to as intra-TCU pauses in the article).
Yet, we already know that DIUs are also, at their minimum, requests for displays of knowledge in the shape of TCU completion. Silences after a request for a knowledge-display in instruction sequences have an additional meaning: They "can display recipients' understanding that it is expected that they should know the answer, not merely that they might know the answer" (Drew, 1981, p. 260). In other words, by remaining silent when not knowing the answer (instead of, for instance, claiming their lack of knowledge) students "treat their problem as being, not that they just happen not to know the answer, but that they ought to know (are being expected to know) and don't" (Drew, 1981, p. 260). Having said so, let us now return to Excerpt 2. That also the teacher has such an assumption about what the students ought to know when she interrupts her turn in line 2 is indeed confirmed by the fact that the request for completion is reissued given the students' failure to complete the utterance in the time allotted in the pause. This is done with a simple repetition of the preposition and with a more distinctive rising intonation (line 4). By doing so, the teacher implies that students do not need to be addressed with a more explicit question in order to supply the expected information, nor do they need any additional help. Consequently, she casts them as "knowing recipients" (Goodwin, 1987, p. 121).
Thus, the first salient aspect of the deliberate nature of the practice is the presence of intra-TCU pauses in positions where turn transition is not relevant (Sacks, Schegloff, & Jefferson, 1974). Although students do not provide the expected completion, the pause in line 3 is indeed deployed as a device to elicit completion. This is also so because it separates two elements of a prepositional phrase, which in the Italian language is grammatically very tight. Thus, the deployment of a pause in a position where transition is not relevant poses the question: "why that now?" (Schegloff, 1998, p. 415).
The second relevant feature of the phenomenon concerns a distinct prosody pattern in approaching the point where the teacher withholds part of the TCU. In all instances examined here and in general in all DIUs in my data, the final items uttered by the teacher before the pause are delivered with distinct upward intonation (sometimes it occurs only in the last one, but often in the final two words or syllables). One basic function of this intonation pattern is to highlight that the TCU has not reached completion. For instance, in Excerpt 2 the teacher uses a combination of emphasis and slightly rising intonation in the last vowel of the verb ("entrati,")a prosodic pattern then repeated in the delivery of the preposition ("↓nel,") before the pause. This use of prosody differentiates the last two words from previous talk and seems specifically designed to alert students that some kind of request is forthcoming. A third element, which confirms this characterization in the excerpt, is the way in which the teacher pursues completion in line 4, once students fail to provide the expected response.
In sum, with regards to turn design, the phenomenon is characterized as a request for knowledge-displays by means of two main basic features: (a) the deployment of intra-TCU pauses and (b) the repeated use of upward intonation in approaching that pause. The same features are shared by the DIU we have examined in Excerpt 1, here reproduced as Excerpt 4: As in Excerpt 2, here the teacher uses rising intonation in delivering the last two words before the pause in line 4, thus indicating that a completion is expected. Whereas in Excerpt 2 the teacher uses a combination of emphasis and slightly upward intonation, here the teacher employs a more distinct rising intonation contour in delivering both the last and the penultimate word of the incomplete utterance, preceding the pause in line 4. The phenomenon can also obtain the coproduction of one word, as in Excerpt 5. The request for completion here concerns the last two syllables of the Italian word for "work": lavoro (syllable spelling: 'la/vo/ro'): Again, the same prosodic pattern is produced: The simple preposition di ("of") preceding the word is delivered with upward intonation, along with the first syllable of the following word, whose full delivery is suspended for 0.2 s. In addition, here the teacher produces a micropause that separates the two components of the prepositional clause: the preposition (di) and the subsequent noun lavoro ("work"). The micropause is here interpretable in two ways: It might be used as a means to separate the preposition from the linked subsequent name, as a form of emphatic delivery, and/or it can indicate that the teacher has designed a DIU, which she then judges students will not be able to complete and, thus, resumes the turn. 9 In line 3, after the pause, one student produces the expected completion, with other students soon joining in the activity. Instances of DIUs similar to Excerpt 5, in which teacher and students coordinate the coproduction of a word, are rather frequent in the data of my corpus. In this case, the teacher uses sound-stretching to separate the word into its syllables: the vowel sound of the first two syllables of the name (here indicated in bold characters: ve/ner/dì) of the day is stretched. The syllable spelling stops for 0.2 s after the second component and is followed by the students' completion in 3. 10 Examples like Excerpts 5 and 6 clearly show that DIUs of this type rely on the expectation that the request for completion will be fulfillled also on the basis of a third feature, which is shared in all instances of my corpus. Such DIUs are specifically designed to provide students with all resources to successfully complete the unfinished utterances, with very little cognitive effort on their part. By halting the syllable-by-syllable spelling of the name of the day after the second syllable, the teacher has reached the last possible point where students can actually provide completion of the name of the day. Similarly, the DIU in Excerpt 5 works on the recognition of the expected word, of which the teacher provides the first syllable. In this case, the teacher is commenting on a previous answer given by a student that mentions "work" as an important aspect of social organization. Therefore the teacher seems to rely on the students' short-term memory to elicit completion as repetition of prior talk. Another explicit case is Excerpt 7: Here the DIU precedes rather than follows the question that is produced in line 4, and therefore it cannot be a case in which the DIU is used to prompt the answer to a prior question. The pause is deployed in a position where the word-similarly to the case illustrated in Excerpt 6-is highly recognizable. By the same token, the pause does not occur in a competitive or talkative context. Thus, by leaving the last two syllables of the word unsaid in such position and context (in Italian very few words have a different continuation after such a beginning), completion is likely to be successfully produced by the majority of the students, as indeed occurs with the choral response in line 3. Through this specific way of delivering words, the componential aspect of words is highlighted, thus enabling students to project and produce the remaining part that would complete the word. The device exploits a mechanism that has been described by Lerner (1991), whereby one participant can complete the utterance-in-progress of another participant via the inspection of the utterance as a "multiple component turn-constructional unit" (p. 441). As Lerner describes, the features of talk that can be described as "compound turn-constructional units" (p. 442) and that can enable the joint production of an utterance in ordinary conversation are various and mainly involve syntax: the if X-then Y format, quotations, parenthetical inserts, list structures, prefaced disagreements, and the instead of X-Y format. However, also prosody and certain features of turn delivery can work as additional devices to encourage joint completion. Lerner discusses a case where the syllable spelling of a name "can provide a 'projectable compound format' as based on its phonological feature" (p. 450).
As the analysis of the previous three excerpts shows, the phonological feature of the item on which the DIU operates can be used as a variable feature: Depending on how many syllables the teacher produces before the pause, the final component is more or less likely to be projected by recipients. Thus, recognizing the full words venerdì and temperatura in the teacher's unfinished production of vener-and tempera-, is indeed a much easier task than projecting the word lavoro from the first syllable only. As the analysis of main-clause DIUs will show in the later sections of the article, when teachers cannot rely on the phonological composition of the word to enable students to project the expected completion, they employ other methods to ensure that the DIU will be successfully completed. One of these methods is to make the DIU operate on items that have occurred in prior talk and elicit a repetition based on the students' short-term memory of prior talk.
Silence as a variable feature in DIUs' intra-TCU pauses. The duration of silence in the intra-TCU pauses is a variable feature of DIUs, and it is linked to the students' ability to recognize the missing part that would provide for the completion. The length of pauses ranges from that of a micropause, to 0.2 s, to 1 s. Moreover, as shown in Excerpt 8, completion can be produced even without any silence intervening after the implicit request for completion. In line 1 the teacher produces the typical prosodic pattern, followed by the students' response in lines 2 and 3, without any intervening pause. It is worth highlighting that, in this case, students do not produce a proper completion of the word, but instead the whole word. This behavior seems related to the specific pedagogic function of the DIU that is here accomplished: mainly, that of prompting an expected answer after a prior unanswered question. Notice that the device follows a rear-loaded question, which is often produced in this type of classroom instruction activities. With this term I am referring to instances in which teachers change their statements into interrogatives simply by substituting part of the sentence with the corresponding wh-element (what, why, where, when, etc.). One reconstructed example would be "You want what?" instead of "What do you want?" Furthermore, as shown in Excerpt 9, an extended version of Excerpt 8, the DIU is produced to elicit the answer to a previous question, which students failed to answer as expected. The vertical arrows indicate the right-dislocated wh-element in the questioning turns preceding the DIU. The response provided by the student in line 2 is not the expected answer. First, it isn't grammatically conforming to the constraint of the question. 11 The open interrogative, and precisely the "what," makes a noun phrase relevant as the direct object of the verb "to administer," whereas the student offers a verb phrase (vivere in pace ["to live in peace"]), hearable as the argument of bisognava quindi ("one needed therefore"). Probably owing to this nonconformity of the student's turn as the expected answer to the question, the response passes unnoticed. Instead, the teacher reissues the question, followed by the beginning of the correct answer in the shape of a DIU. Thus, it seems that here students produce the entire noun phrase, rather than just the missing syllables that would complete the noun because here the DIU has the function of prompting the answer to a previously unanswered question. This can also explain the absence of a pause before the students' responses, as I will explain in the next section.
To conclude, in this section I have outlined the basic features of the phenomenon as it is produced in argumentative sessions of whole-class talk and chalk instruction sequences. The analysis, so far, has focused only on aspects of turn design and speech delivery that characterize the device, independently from the functions it accomplishes. In the following section I will address issues concerning the uses of DIUs in this setting as related to the local pedagogical agenda.

The Phenomenon and Its Pedagogic Functions in Talk and Chalk Instruction Sequences
DIU as a phenomenon is not new to conversation analytical studies on classroom interaction (Jones & Thornborrow, 2004, Koshik, 2002Lerner, 1995). Actually, the acronym that identifies this practice stems from Koshik's work, in which the author describes its use in oneon-one second-language writing conferences as a means to elicit the student's self-correction. The same device has been analyzed by Lerner (1995) as it takes place in whole-class instruction activities to prompt an expected answer after a prior unanswered question, and to furnish a procedure for definition writing. In his discussion on the possible uses of the device in classroom interaction, Lerner also suggests that "this seems like an apt procedure to employ with a talkative class, since it musters, coordinates and limits the participation of all those students who are prepared to reply" (Lerner, 1995, p. 117). These pedagogic functions cover only some of the functions of DIUs in the various settings and multiple activities that take place in classrooms. 12 For instance, a large number of the DIUs occurring in my corpus are not sequentially linked to a prior unanswered question, nor are they produced in a competitive and talkative setting, nor used to elicit self-correction or definition writing. Rather, they seem to be produced to enhance the coconstruction of a line of reasoning. Furthermore, as shown earlier, one feature that DIUs in my corpus seem to have in common is that they are specifically designed to be successfully completed with very little cognitive effort. This aspect-together with the observation that the device can operate on different turn-constructional units such as words, phrases, clauses, and sentences-suggest that DIUs can be extremely flexible in terms of the pedagogic functions they can serve. One way to address this issue is to explore what can be done with DIUs that interrogatives cannot achieve.
What DIUs can accomplish that interrogative-formatted questioning cannot. In what follows, I will compare the way in which questioning is accomplished with a DIU and with an open interrogative (see arrowed lines in Excerpt 10); both instances are from the same instruction on angles reported below with which the reader is already familiar (Excerpt 1). The talk is produced to comment on Caterina's performance. The teacher has invited her to make a full turn around the teacher's desk in order to count how many times she has to change direction. The demonstration is designed to suggest the correlation between changes of direction and angles (see lines 23-28). The excerpt starts when Caterina is about to complete her turn around the desk and to change direction for the fourth and last time. The two arrows indicate the open question and the DIU that I will compare. In both cases, the teacher solicits specific information and confirms the correctness of the response through the repetition of the answer in the third position (lines 18 and 43). It is worth highlighting also that both instances address the same observational datum (the presence of four angles on the surface of the teacher's desk). Yet each format pursues a different aspect as relevant in the local pedagogical agenda. The open question elicits the quantity of angles on the desk, whereas the DIU pursues the word "angles." Other similarities concern the collective production of the answer: In each case, two cohorts of students answer in unison.
It is to be noted, however, that after the open question one student leads the answering session by producing the answer as it overlaps with the teacher's turn, whereas the DIU has the pause before the response. The different timing of the answer in the two cases seems related to the linguistic resources that each format uses to accomplish the questioning. 13 In order to indicate that a questioning is being accomplished, the open interrogative exploits syntax and the position of words in the sentence-the interrogative pronoun quanti ("how many") is deployed in the TCU initial position-whereas the DIU uses prosody-mainly rising intonation and a type of delivery that can be described as not-forthcoming talk. These features of the DIU alert recipients that a request is made only when talk reaches the position where the item is due or is in the close vicinity of the missing item. Owing to this mechanism, this questioning device does not permit recipients to anticipate on which item the DIU will operate and, thus, to respond in advance toand not concurrent to, as overlapping-the questioning. This occurs even if there might be some variation in the timing of the completion elicited through DIUs. Therefore, the resources used to accomplish questioning in the two formats involve different types of responses (full answer and syntactic completion) and have different sequential consequences for the structuring and the timing of the responses.
These observations also raise issues related to the directness/indirectness of the question. First, right from the beginning and unequivocally, the interrogative pronoun in the open question indicates the object of the question (or the questionable) as relating to a definite and clear piece of information. The questioning turn that is thus formatted also projects the content of the answer, to be produced when the question reaches its intonational, syntactic, and pragmatic completion. I am referring to these notions as they are defined by Ford and Thompson (1996). Therefore, returning to Excerpt 10, the turn beginning in line 10 reaches its syntactic completion point when the TCU introduced by quanti ("how many"), as well as the whole turn, has reached its terminal boundary. From the point of view of intonation, the interrogative clause in line 14 seems lacking in the typical prosodic features that are usually associated with finality: a marked fall/rise in pitch (Ford & Thompson, 1996, p. 147). The clause is produced here with plain intonation, thus contrasting with the various rising intonation contours the teacher has produced in the course of her turn at the previous possible completion points. However, although no conventional final intonation (rising or falling) is used, the contrast with the prosody employed in previous talk (as well as syntax) enables recipients to hear the interrogative as accomplishing turn completion. Finally, as for the pragmatic completion, the interrogative clause carries a sense of pragmatic completion insofar as, being a first pair-part of an adjacency pair, it conveys a strong expectation for an answer. All these features characterize the open interrogative as accomplishing questioning unequivocally and directly.
By contrast, the resources that DIUs exploit to convey the expectation of completion are not as explicit. The ambiguity of the device plays at two levels, at least. The first level concerns the fact that DIUs seem to mobilize mainly intonational and/or prosodic resources to mark that a request for completion is accomplished. Indeed, also syntax is at play here, but in the reverse way to how it works in the open interrogative. DIUs base their pragmatic force precisely on recipients' understanding of the nonfinality of the talk-so-far (or its noncompletion) exactly in those points where marked prosody is used. It is also worth considering that prosody-in contrast to syntaxincludes gradient and not segment-based phenomena rather than discrete ones (Couper-Kuhlen & Selting, 1996, p. 11). This means that, despite the effort of some linguists to identify the components of intonation contours (for a related and detailed discussion see Couper-Kuhlen and Selting, 1996, p. 17-21), intonation patterns are not unambiguously associated with their functional meanings per se.
For example, in the turn beginning in line 37, there is neither an interrogative pronoun nor any other syntactic marker to tell the student that a question is being posed. Instead, the mechanism here functions according to the contrast between the syntactic noncompleteness of the utterance precisely where the rise in pitch and the pause are produced (the pitch is emphasized also by means of other copresent features of talk delivery, like sound stretching and harder pronunciation). These features convey a sense of transient and relative finality, as related to the item on which the DIU operates, alerting students that the focus is on that precise item. This way of constructing the questioning has a number of consequences. For instance, owing to this ambiguity and indirectness, the device maintains a certain ambiguity about whether the teacher is actually posing a question or if s/he is just pausing in talk. This ambiguity conveys what I call a certain resistance to qualify the action as an explicit questioning. The indirectness and therefore weakness of the action (as compared to the way in which questioning is done with open interrogatives) are also grounded in the occurrence of the so-called failed DIUs whereby the device favors a reversible type of questioning. In other words, when students do not provide the requested completion, the teacher can continue and finish the utterance her/himself, thereby avoiding any apparent disruption of the ongoing talk and hence nullifying the unanswered question. This is visible in lines 1-2, 25, and 27 in the previous excerpt. In these occurrences, the device imposes a weaker constraint on recipients than interrogatives do. Evidence for this is provided by the fact that the teacher does not pursue further the fulfillment of the request when students fail to provide the expected completion. This weaker pragmatic force exhibited by DIUs might account for the device's high distribution in instruction interaction in whole-class and talk and chalk classroom management. The device permits teachers to address the group-class or individual students at any moment in the course of the teachers' ongoing talk, without much disruption in the case that recipients fail to fulfill the request. If students fail to answer, the teacher can continue talking, transforming a request for completion into a pause, or retrospectively, into a hesitation or even an emphatic delivery of the talk. On the other hand, owing to the fact that the device is mainly based on prosody, it can be repeatedly applied to any syntactical unit that composes the utterance: word, phrase, clause, and sentence. This permits the teacher to invite a large group of students to participate in many ways and often.
On the other hand, and almost paradoxically, the indirectness of the device attests to the questioner's expectation that the response is highly accessible to recipients. By posing the question so indirectly, the teacher seems to consider recipients as fully knowledgeable on the matter and as able to finish the utterance almost spontaneously or with minimal indication that a request has been put forward. DIUs, like other ways of managing utterance completions in ordinary talk-ininteraction (Goodwin, 1979;Lerner, 1991;Sacks, 1995), such as word searches (Goodwin, 1987;Schegloff, Jefferson, & Sacks, 1977) and list constructions (Jefferson, 1990), exhibit the speakers' understanding of the recipient as knowledgeable about the matter at hand. In the classroom context, by eliciting completion of the ongoing utterance, rather than requiring an answer to a question, the teacher holds students to be on board about the matter at hand so as to be capable to complete the syntactic unit that is left unfinished, as if they were reading the teacher's mind (Jones & Thornborrow, 2004, p. 419). Each of the collaborative practices mentioned above is worth considering in its own right, but I would like to point out the quality the practices have in common and which DIUs also share. This quality is described by Sacks in the automobile discussion excerpt, in which three teenage participants in a group therapy session collaborate in producing one sentence: Joe: (cough) we were in an automobile discussion Henry: discussing the psychological motives for Mel: drag racing in the streets Sacks describes the phenomenon as follows: "In the way that these fellows fit their talk together, which would involve a hearer in seeing that they 'know what's on each other's mind', he could presumably as well take it that what it is they are talking about, they also know" (Sacks, 1995, Vol. 1, p. 147). In conclusion, it can be argued that what the device conveys is the teacher's assumptions that students would be able to complete the utterance because they know (or they are guided to know) what the talk is about.

MAIN-CLAUSE DIUs: ELICITING A DISPLAY OF LEARNING
So far, the phenomenon has been described in terms of its basic constructional features and some of its pedagogic functions. I will now focus on the specific subset of DIUs which I call main-clause DIUs. The analysis will show that, although the device generally demands little of students in terms of the cognitive effort necessary to accomplish the task, this type of DIUs has a rather crucial function in the instructional sequence where it occurs, owing to the syntactic unit on which it operates and its linkage with prior talk. I will show that occurrences of this variety of DIU are produced to highlight specific pieces of information that have been mentioned in previous talk and reproposed in the context of a main clause to be thus jointly produced by the teacher and the students. This specific construction of the DIU provides students with the opportunity to produce a display of knowledge shaped as a completion and as a repetition of prior talk. The two features-the fact that (a) the DIU occurs in main clauses and that (b) the item on which the device operates is a repetition of the same information earlier mentioned-permit the teacher to accomplish the specific pedagogical function of main-clause DIUs: to highlight the information on which it operates as a key notion and as newly acquired knowledge, which in turn casts students as learners.
I will start the description of this variety of DIUs by focusing again on the "angle" excerpt with which we are already familiar, here reproduced as Excerpt 11. One of the features that distinguishes this particular type of DIUs from other occurrences of the device is its embedding in a simple, independent main clause, rather than in secondary clauses in a compound sentence. I argue that this specific deployment in the turn seems to be responsible for the emphasis that is thus given to the item being elicited. As a result of this emphasis, it qualifies as a key notion in the local pedagogical agenda enacted through the ongoing instructional activity.
In what follows I will provide more robust grounds for this observation. The instance can be seen again, but this time in the larger context. The reader will recognize in Excerpt 12 part of Excerpt 10. Our target case starts in line 28 (arrow). The turn is shaped as a full, well-formed simple main clause. The structure of the sentence has the regular format of Italian simple sentences (SVO: Subject Verb Object), composed of subject ("the surface of the teacher's desk"), verb ("has"), and object ("four angles"). The turn occurs at the end of Caterina's performance (her full turn around the desk) and seems to be produced precisely to mark the end of it, as indicated by the teacher's invitation to Caterina to return to her seat, produced immediately before the DIU (line 25). The sequence closure is also indicated by the disjunctive marker the teacher produced in line 23 ("Alo:::ra"). The disjunctive function of the item is emphasized by the increased volume of the teacher's voice used in the initial vowel sound and by the stretched middle-syllable vowel. The fact that talk has reached a crucial point is also indicated by the 1-s pause that precedes the disjunctive marker (line 22), and by the routine understanding-check question that the teacher addresses to Caterina in line 21. All these features bring the ensuing talk to the fore, and as a result the turn with the embedded DIU qualifies as a key point in that instruction activity.
Let us now focus on the item on which the teacher chooses to make the DIU operate, and its relation with prior talk. I have already pointed out that generally all DIUs have the property of being specifically constructed to be successfully accomplished. I have described one mechanism that facilitates students' identification of the missing part: In delivering the missing part of a word, the teacher uses particular features of talk delivery that emphasize the phonological pattern of the word on which the DIU operates and that highlight the compound pattern of that item (syllable composition). The item, however, on which the DIU in line 30 is built (like all the other main-clause DIUs in my corpus) is a full word ("angles"), not a part of one. The task is therefore more open and, thus, less likely to be successfully accomplished by students. We will see that, under these circumstances, the teacher uses a different strategy to ensure that students accomplish the task. This strategy consists of asking students to complete the utterance with a word that has occurred before; thus, students are asked to recall that word and repeat it, albeit in a different utterance context.
It is important, however, to see the mechanism actually in action. As we see in Excerpt 12, the closest occurrence of the word "angles" is indeed in the vicinity of the DIU. The teacher produces it in line 19, soon before the question-check she performs on Caterina. It is worth recalling that the whole sequence is about angles or the fact that a rectangular-shaped object like the teacher's desk has four angles (lines 5 and 9) and that angles are generated by changes in direction (lines 14-19). In order to guide students to become aware of this datum, the teacher organizes the instruction activity as a demonstration in which Caterina walks around the desk, and the class counts the times she has to change direction.
I will now focus on the different requests that the teacher makes of the class and their sequential deployment within the larger sequence of questions in a series; in this way, I will reconstruct how the DIU in line 30 refers back to prior occurrences of the item thus elicited. The sequence opens in line 1 with an open interrogative prefaced by a secondary if-sentence that contains a failed DIU. Apparently, the DIU in line 1 is designed to make students focus on the object (the teacher's desk) about which the teacher will formulate the question in line 5. This is also indicated by the teacher's pointing to the desk that she wants the students to mention. The DIU fails to elicit the requested completion that is subsequently produced by the teacher, as is often the case with this subset of secondary-clause DIUs. The question is promptly answered by students in lines 6-8 and line 10 (see discussion above on the comparison between open-interrogatives and DIUs), but the teacher decides to elaborate further on this datum by counting again the number of angles (line 9). A second DIU is then produced; it is embedded in the counting and insists on the number of angles on the teacher's desk (line 9). Once more, the DIU fails to elicit the final item of the counting, which is provided by the teacher in line 12. It is noticeable that here the teacher seems to pursue the completion of the counting despite the students' failure: In line 12, before articulating "four," she pauses, and postpones the place where "four" is due, by inserting the calculation "and one." The reason for the insistence on the number of angles (actually, on number "four") becomes clear in the talk that develops from line 14 to 19; here the teacher sets the equation between the number of angles and that of the changes of direction that Caterina made in walking around the desk and which seems to be the teacher's goal for the activity.
The two DIUs (lines 9 and 17) that follow the open interrogative in line 5 are thus constructed as linked to and elaborating on the notion of four angles that the open interrogative has elicited. This emphasis on the specific datum (the number of the desk's angles) emerges also in the syntactic formats of the turns that contain these DIUs and in the type of action that they are designed to carry out. In line 9 the teacher counts the angles, to confirm the correctness of the students' answer and to highlight the datum "four angles," which then will constitute one of the terms of the equation. Syntactically speaking, the turn is an expansion of the answer and does not have the status of a main clause.
By contrast, the turn beginning in line 14 with the equation has a rather different status. First, grammatically speaking, the utterance is a nominalized statement; 14 the absence of the verb indicates that the speaker designs the utterance to refer back to an implicit, shared knowledge and contributes to make the structure of the sentence more compact and more similar to a single, independent main sentence. Syntactically speaking, it can be observed that the same referential content of the equation could be expressed more broadly with a compound correlated sentence (of the type: "when there are four angles on a surface, then there are also four changes of direction"), or more concisely with a single main sentence (of the type: "four angles mean four changes of direction"). In this case, by omitting the verb, the teacher avoids making the choice, but obtains an even more compact resulting effect than with the explicit main sentence. It seems, therefore, that the turn beginning in line 14 is produced to be heard as one of these main clauses that are designed to mark a salient step in the argumentative line of the instruction sequence. The ritual understanding check in line 12 ("è chia::ro?" ["is it clear?"]) and the pause in line 13 typically preface this type of utterance. In short, the structure of the sequence can be described as follows: It initiates with an open interrogative, whose answer, after the teacher has acknowledged it (line 9), is expanded by the counting. This action, in turns, confirms and emphasizes the datum that, in turn, is used to lead students to become aware of the correlation stated in the equation (line 14-17) that functions as the core of the activity so far.
However, it is worth considering that often, as is the case here, the teacher's pedagogic agenda can remain unfulfillled by students. In Excerpt 12 there is evidence of the disalignment between the teacher's pedagogical goals and the students' contributions to the interaction. The first indication of this disalignment is constituted by the fact that the DIU in line 17 fails. The second is represented by the two instances where one student produces twice an unsolicited observation (line 15) about the edge of the desk-a proposal that passes completely unnoticed by the teacher, probably because it conflicts with her pedagogic plan.
Interestingly, however, the teacher seems to modify the argumentative trajectory of the instruction sequence. Once the two subsequent failed completions on the students' part (lines 11 and 18) show that students are not aligned with the teacher's pedagogic project, the teacher seems to return to the earlier step in the discourse: the discussion of the four angles on the desk (line 28). In this context, the DIU produced in line 30 is clearly designed to summarize what the students' responsive behavior indicates they have understood so far and, specifically, that the surface of the desk has four angles (lines 6-8). With the main-clause DIU here produced, the teacher provides a context where the noun phrase "four angles"-which remained incomplete in line 17-now finds a new slot where it can be newly and appropriately assembled with the students' joint production of the final item.
It would be therefore erroneous to infer that since the same word ("angles") is elicited in order to be collocated in the same noun phrase ("four angles"), the DIU in line 30 accomplishes the same function or elicits the same type of information as the DIU in line 17, or that the teacher is pursuing the same response all along. As shown in the analysis, I claim that the pedagogic functions of DIUs are inextricably linked to their sequential deployment within a series of questioning actions. Knowledge-displays that are thus elicited from students are, in turns, responsible for the teacher's subsequent question, as well as of the question's format, content, and deployment in the talk. In other words, as the analysis has shown, teachers' questioning is highly reflexive of the students' responsive behavior and of their understanding of the talk so far.
With regard to the main-clause DIU in line 30, parallel and partially contrasting with the cognitive paucity of the task is the emphasis that is given to the task as related to prior talk and to the activity underway. It is worth recalling once more that the turn in which the DIU occurs is produced at a closure boundary of Caterina's performance, as indicated in the following organizational points of the talk: the teacher's ritualistic way of checking Caterina's understanding in line 21 15 (it should be recalled the teacher attempts to move into this stage earlier in line 12 before the failed main-clause DIU in lines 14 and 17 ); the disjunctive token allora, which is used to mark the shift from one phase of the lesson to the other (line 23) and in particular the marked delivery of that term; the invitation to Caterina to return to her seat and the subsequent thanking, whereby the teacher further underlines that the activity has reached its closure (line 25); and finally the pause in line 26. All these features provide a framework that gives salience to the forthcoming talk.
Owing to the fact that the turn in which the device is embedded is brought to the fore, the DIU in line 30 sets the topic on which it works as particularly relevant in the local agenda of the activity underway. Furthermore, as a repetition of earlier produced occurrences of the same word, the DIU tends to be successfully completed. The repetition format of the completion and the sequential deployment of the whole turn where the DIU is embedded present an opportunity for students to master the information that the whole instruction sequence has made explicit and visible. On this basis it can be argued that the device is designed to convey the information that the desk has four angles as newly acquired information or, in other words, information that is being learned as the result of the teaching activity now and there, which is the specific goal of that instruction activity.
These features are recurrently produced in my corpus, of which the next Excerpt 13 is a further case of main-clause DIUs. Excerpt 13 is from the same Geometry Hour. Prior to the exchanges reported in the transcript, another student (Giovanni) performed a demonstration on the angles for the benefit of Caterina, who had been absent the day before. The performance consists of showing that angles are generated out of a rotation movement. In order to show this, the teacher directed Giovanni to walk a few steps and then to turn right and left a couple of times, while keeping his arm raised straight in front of himself and parallel to the ground. In this way, each time he turned and changed direction, the rotation (and the angle thus generated) was made more visible by the turning left or right of the outstretched arm. When the excerpt starts, Giovanni has been sent back to his seat. The teacher is commenting on the student's performance while she herself repeats the rotation, assuming the posture with her arm raised. She addresses Caterina directly with questions about what she has just seen. The upward symbol (ˆ) that precedes the description of the teacher's nonverbal behavior in the glosses indicates where gestures start in relation to talk. The target line, in which the DIU occurs, is line 10, and it is indicated by an arrow. The question in line 1 is addressed to Caterina (note that the selection is accomplished by the teacher with the gesturing of her head slightly forward and upward, as indicated in the gloss to line 1). Caterina labels the movement of Giovanni's arm in line 3, along with another student who gives a different response in line 2. Each student provides a different plausible definition, but both use lay terms. Through the repetition in line 4, the teacher seems to display that she endorses the first answer. However, since the teacher does not release the arm posture (as indicated in the description of the teacher's nonverbal behavior in line 5), she manages to keep the sequence open. That the matter is not settled is also confirmed by the way in which the talk progresses. In line 6 the teacher replaces the lay terms produced earlier by the students with two technical terms (the verb ruota [third person singular present indicative "he rotates"] and the name rotazione ["rotation"]) to define the action and the circular movement of the arm upon the turning body. We can also observe that the teacher uses gestures to convey and underline the specific meaning of the action she wants students to name (glosses to line 6). Hence the DIU, which is later produced in line 10, seems to be specifically designed to insist on this specialized word. The particular importance of this information (i.e., "my arm does a rotation") is further underlined by the successive turn designed to solicit completion exactly at the word rotazione ("rotation").
Like the DIU in Excerpt 12, here too the teacher gives salience to the turn by detaching it from prior talk in various ways. The production of the stand-alone confirmation token in line 8 as well as the teacher's body behavior provides a framework whereby attention is drawn to the turn in line 10. The turn is thus constructed as a standing-alone sentence, with the DIU located in the main clause. The teacher, therefore, designs her turn in order to make Caterina contribute the final syllable of the technical word earlier introduced. Owing to this sequential position of the DIUs, here again the completion is designed as a repetition of a prior occurrence (line 6), whereby the prior unknowing student is now treated as if s/he is now able to master the newly acquired information as the result of the instruction activity just enacted.
More importantly, it is to be noticed that, independently from whether Caterina already knew or ignored the meaning of the word rotation, by designing the request in this way (as the completion of a second occurrence of the word), the teacher manages to construct Caterina's production of the word as a repetition of the teacher's prior talk. As a result, here, like in the other occurrences previously described of main-clause DIUs, the teacher manages to cast the completion provided by the student as the result of the activity just enacted. In other words, the request for completing the utterance with a repetition of a prior occurrence of the term embodies the teacher's stance that recipients are not familiar with the item, independently from whether this is actually so. The device considers recipients' verbal behavior and their ability to provide the expected completion as the result of the instruction activity, and consequently, as evidence that some kind of learning process has taken place, through interaction.
Besides their deployment in main-clauses and the repetition format of the item requested for completion, there is another systematic feature of DIUs that contributes to making the information elicited particularly salient: the utterance in which they occur conveys a blatantly true meaning as their content is conceived either as universal truth (as in Excerpts 13 and 14), or as a plainly accessible state of affairs (as in Excerpt 11). The fact that rotation is necessarily correlated to change of direction (Excerpt 13) is an assertion whose meaning is a general truth. Similarly, the utterance in the next example (Excerpt 14, lines 11-14) stating that services must be increased when the population grows is also an assertion that conveys some general truth.
On the other hand, the fact that the teacher's desk has four angles (Excerpt 11) is a sensible datum, plainly accessible to anybody in the classroom and made even more accessible by means of Caterina's performance; as such, it also conveys a general truth.
In this regard, line 10 of Excerpt 13 presents other features that emphasize this meaning: the present tense of the verb and the first person singular pronoun, used impersonally; the hand beating produced by the teacher in delivering the statement, as well as the way in which she takes a central position in front of the students, so as to face the whole class-all these features add to the sense of generality with which the teacher states the principle that "a rotation is generated each time there is a change of direction." To sum up, the features that characterize main-clause DIUs are the following: 1. The completion is constructed as a repetition of prior talk. 2. The DIU is embedded in a main clause. 3. The utterance conveys a general truth.
This list will help the reader in the analysis of the last excerpt proposed for the discussion. Excerpt 14 is from a Geography Hour. The transcription captures the interaction in a moment when the teacher is engaged in eliciting the category-noun servizi ("services")-a term that pupils came across the year before. Prior to the excerpt, students were invited to formulate the reasons why people tend to build towns rather than live apart in isolation. Through a series of questions, the teacher had guided them to realize that living in towns is advantageous because people have access to hospitals, schools, transports, and other services. The collective word has not been mentioned in prior talk, but it will be solicited through multiple prompt-to-answer DIUs (indicated in the transcription with bulleted points). The expected response is finally produced in line 6 by the student Maria and soon endorsed by the teacher in 7 and 9. Our target case occurs in the turn starting in line 11, and it is indicated with the arrow in line 13. Like in the two prior examples, also in this excerpt the teacher designs a DIU that operates on prior occurrences of the word (lines 6, 7, 9) and is thus solicited as a repetition. It should be noted that the word servizi occurs in line 6, produced by a student as completion to the prompt, then in line 7, as the teacher's confirmation of the answer, and finally again in line 9, embedded in a full-grammatical sentence whereby the teacher further assesses the answer as correct. Another similarity with prior examples concerns the syntactic structure of the turn that accommodates the DIU in line 13, which is a main clause, again. Moreover the incomplete utterance has the generic quality of statements of rules, principles, and general truths. The sentence coordinates two main sentences through correlative conjunctions, with the DIU located in the second main sentence. Therefore, as in prior instances, the teacher designs a standing-alone sentence, albeit consisting of two coordinated main-sentences, which solely composes the turn produced at the closure of a prior activity (consider the introductory disjunctive token allora), and sets the information thus solicited as pedagogically relevant.
By leaving the utterance incomplete, the teacher encourages students to fill in the empty slot in her turn to complete the same word that Maria earlier produced as a response. In this way, the completion solicited through the DIU is shaped as occurring in the activity that has brought the topic to the fore. Thus, students are given the opportunity to show that they can master the usage of the earlier produced word and, in a certain sense, that they have learned it.
The analysis of main-clause DIUs and of their sequential deployment has shown that participants orient to this activity as coincident with that of mastering some key notions, here conveyed by technical words such as "angles," "rotation," and "services." This phenomenon uses the mechanism of collaborative completion that speakers regularly use in ordinary conversation to manage the joint production of sentences. In this case, that mechanism is refined and specifically molded to fit the context features of whole-class interaction in order to accomplish one of the institutionally relevant goals of instruction interaction: namely, making the activity of learning accountable as "locally managed, partly-administrated, interactionally controlled and sensitive to recipient design" (Sacks et al., 1974) .

CONCLUSION
The analysis presented in this article has centered on the characterization of one questioning device used in whole-class instruction sequences and demonstrating how parties reach consensus about the teaching/learning goal of the ongoing pedagogical activity as it surfaces in the organization of interaction. The phenomenon under examination is one subset of DIUs (Designedly Incomplete Utterances) deployed in main-clause utterances and designed to elicit completion in the shape of a repetition of words occurring in earlier talk. It has been found that this variety of the device is used to show that some kind of a learning process has taken place as the result of the instruction interaction, whereby students are able to master key notions. In contrast to other strategies used by teachers to facilitate students' identification of the expected completion, the repetition format used in main-clause DIUs casts the information on which the device operates as newly acquired information while casting the students as having just learned the notion at hand.
Main-clause DIUs have been compared to other varieties of DIUs that are used to accomplish different pedagogical functions and with other syntactical formatted questions, namely open interrogatives. It has been found that the device is a valid means to accomplish questioning in a more indirect way than that of interrogative-formatted questions. It offers a major usability in promoting participation in a large group of recipients, as is the case in teacher-led and talk and chalk structured lessons. The indirectness is principally due to the linguistic resources that the device employs-mainly prosody-and to the device's flexibility, which allows it to be successfully applied to any type of syntactic unit that compounds the turn, namely words, phrases, clauses, and sentences. Owing to these features, the device can be used repeatedly on any unit in the progression of the teacher's turn, while also providing for the turn's reversibility when students fail to provide the expected completion. These qualities make DIUs in general a powerful instrument because teachers can modify the trajectory of their talk as well as the preplanned and local agenda of the ongoing instruction activity, depending on the students' responsive behavior, as shown in the excerpt on angles.
These findings raise relevant methodological points for discussion regarding classroom interaction and the way in which pedagogic discourse has been traditionally approached. Much progress has been made in research on classroom interaction owing to the identification of the IRE triplet (Sinclair & Coulthard, 1975) and of the characterization of teachers' questioning as "known information questions" (Mehan, 1979b, p. 285). Since these breakthroughs, research on the topic has continued to grow. The popularity of these analytical categories among scholars, however, seems to have detracted from the importance of fine-grained sequential analysis of teacher-students interaction in instruction sequences. Two areas have been particularly neglected: (a) the characterization, through the analysis of turn design, of the activity of questioning in classrooms as a way to accomplish a range of diverse and specifically pedagogic functions; and (b) the inspection of larger sequences of talk as series of linked question/answer pairs, of their sequential deployment, and of the interreferential nature of the pairs that moves beyond the characterization of the IRE template.
Second, the question of knowledge and how it is differently accessible to different parties is certainly a crucial issue in the classroom, as this is the place where instruction is institutionally and formally imparted to young generations in our society. Regarding this topic, the three-part model with the third evaluative turn and the characterization of the teachers' initiating action as a special known-answer questioning have contributed greatly to the specific relevancies of classroom interaction. However, as the analysis has shown, the question of which party has access to what is not the only relevant element of classroom interaction. Rather, the questions of how they both come to an agreement about what can be recognized as known at that precise moment and on what basis are also important. Equally important is the question of which resources participants use to reach this consensus at any defined stage in their interaction. This analytical approach provides a new way of understanding formal instruction interaction, insofar as it gives an interactional account of the participants' own views of learning, and precisely how it is achieved and how participants perform the activity of instructing and being instructed in talk-in-interaction. Such an approach constitutes a dramatic alternative to both the quantitative and experimental approaches to learning.
Finally, far from an exhaustive account of all that DIUs can accomplish in classroom interaction, the analysis provided here is rather an account of what teachers and students themselves consider as learning in their daily activities as well as of the social organization of classroom interaction beyond the idealized three-part sequence.