Request for confirmation sequences in Egyptian Arabic

4.1 Syntactic design

An overwhelming majority of 158 (79%) examples present a clausal structure. A smaller portion of 42 (21%) RfCs present different kinds of phrasal structures including nominal, adjectival, and adverbial phrases. Formally, these phrasal structures are closely tied to the prior speaker’s turn since they present a syntactic expansion or a retraction to a syntactic slot that was already opened (Auer 2009, 2014). Also in terms of their function, phrasal RfCs are closely linked to the prior turn as they mostly implement understanding checks or clarifications of local scope, mainly targeting the specific reference of a previously mentioned object, person, time, or place (cf. Deppermann et al., 2024 (this issue), for a similar observation in German). The following excerpt (reproducing (1)) illustrates a case of retraction and substitution. Fatḥi asks Sāmiḥ when is he coming to Egypt for a visit:

Extract (4): CALLHOME_5253 00:31-00:33

Following Sāmiḥ’s answer that he comes ‘at the end of the month’ (l. 3), Fatḥi issues a RfC to check his understanding that the referred-to month is October (l. 4). The RfC presents a construct state chain (‘end-of month-of October’), which retracts to the immediately preceding prepositional-phrase and proposes a specific nominal reference for ‘the month’ mentioned by the prior speaker.

4.2 Polarity

The majority of the RfCs in the collection are formatted with positive polarity (n = 168, 84%). A small number of 32 negatively formatted RfCs (16%) present a variety of negative polarity markers, as shown in Table 1.

An interesting observation concerns the correlation between the polarity of the RfC and its dis/confirmatory response: while positive polarity RfCs are confirmed in 70% (117/168) of the cases, negative polarity RfCs are confirmed in only 53% (17/32) of the cases. Heritage and Raymond (2021) have shown that negatively polarized questions are used to take an epistemic stance towards the probability of the question’s underlying proposition and construct it as unlikely. A close inspection of negatively formatted RfCs will show if such a stance is also proposed in this subset of questions and what local tasks are implemented by these instances, which may explain the lower number of confirmations they attract (e.g. (18)).

4.3 Modulation

Epistemic modulation of RfCs is not common in the studied data: only 16 (8%) examples include an overt marker of modulation. Epistemic modulation is achieved by a variety of lexical and grammatical resources including particles, adverbs, prefabs, and fully inflected verb-phrases, as shown in Table 2. These devices foreground different aspects of the epistemic stance of the requesters, for instance, their prior expectations, source of knowledge, level of certainty, or the degree of accuracy they attribute to the confirmable.

Markers of modulation occupy different syntactic positions relative to the RfC: they can precede it, be embedded in it, or follow it. The next set of extracts illustrates these possibilities. In (5) the marker ʔē da ‘what’s that’ pre-frames the RfC. Amīna provides an account of why her voice sounds hoarse:

Extract (5): CALLHOME_4486 01:45-01:47

Amīna’s mention of Ḥanān and Hiba’s wedding prompts ʕAlā to issue an RfC concerned with the earlier-than-expected occurrence of the event (l. 6-7). The RfC presents a marked V-S order which lays focus on the event and more specifically, modified by xalāṣ ‘that’s it’, on its completion. Preceded by ʔē da, it conveys ʕAlā’s astonished stance towards the confirmable information (cf. Selting 1996). Noticeably, this RfC is answered with an expanded confirmation (l. 8) which might be occasioned by the display of an epistemically divergent stance.

In the next example the modulation is not making reference to the prior expectations of the requester, but to his level of certainty. The epistemic marker used in this case is the first-person verb-phrase ‘I think’:

Extract (6): CALLHOME_5595 08:38-08:47

Following Amgad’s confirmation that Amāni is staying with him in the US (l. 2-3), Yeḥya issues another RfC concerned with the location of her son. The verbal form ʔaẓunn ‘I think’ is embedded in the utterance, occupying a middle position between the prepositional predicate maʕāha ‘with her’ and the nominal subject ibnaha ‘her son’. By using a marker of epistemic mitigation such as ‘I think’ (Kärkkäinen 2003), the requester downgrades the level of certainty that he claims in proposing the confirmable.

In the next example, the modulation marker comes after the RfC and qualifies the accuracy of its content. Sāmiḥ enquires Fatḥi about his plans to go on a honeymoon right after his wedding:

Extract (7): CALLHOME_5253 04:23-04:29

Fatḥi confirms that he and his wife are going on a honeymoon right after the wedding (l. 5-6). Sāmiḥ pursues this with another RfC that suggests an exact date for their leave, November 22. This formulation is post-framed by the marker kalām min da ‘that sort of thing’ which downgrades the precision attempted in the RfC. Here as well, the requester orients to the asymmetry of knowledge between him and the responder by hedging his contribution and thus overtly indexing his inferior epistemic position.

4.4 Inference marking

RfCs introduced by an inference marker comprise 47 (23.5%) cases in the collection. The discourse marker yaʕni (lit. ‘it means’) is by far the most common device for marking inference (43/47, 91%). Yaʕni generally functions as a marker that initiates or invites the completion, repair, or elaboration of prior talk (Marmorstein 2016, 2021). It can either pre-frame or post-frame an RfC, and it is more commonly used with phrasal RfCs (14/42, 33%) than with clausal RfCs (29/158, 18%). Other markers, for which only one or two instances are recorded, are baʔa ‘then’, ʔaṣl ‘the fact is that’ and ṭab ‘well then’, as shown in Table 3.

The following extract (reproducing (2)) illustrates the usual case in which inference is indicated by yaʕni. In response to Yeḥya’s question about whether Rafīʔ or Rifʕat have come to him, Šawki reports that it was Rafīʔ whom he met:

Extract (8): CALLHOME_5595 02:00

Following the production of a news receipt at line 6, Yeḥya issues a yaʕni-prefaced RfC. By using yaʕni Yeḥya makes explicit that his upcoming proposition follows from – indeed, it is the meaning of – what has been said so far, namely, that in meeting Rafīʔ Šawki did not meet with Rifʕat since the option was initially presented as exclusive to either one of them (l. 2).

Unlike the ‘neutral’ or unmarked inference indicated by yaʕni, the marker ṭab ‘well then’ (a short variant of the form ṭayyib ‘good’, see (10)) is stance-laden and indicates the disagreement of the requester with some aspect(s) of the contents of the previous turn. Consider, for instance, the following excerpt. Nafīsa, who is invited to visit her sister Ṣafā, expresses concern that this will be complicated by the fact that she is babysitting for Ǧina, another sister of theirs:

Extract (9): CALLHOME_5827 13:40-13:45

In response to Nafīsa’s claim that she cannot come without the baby while Ǧina, his mother, is at work, Ṣafā issues a RfC that takes issue with this claim, constructing the inference that Ǧina can be separated from her baby as unlikely. The marker ṭab which prefaces this RfC is both retrospectively and prospectively oriented: it indexes Ṣafā’s disagreement with the prior turn’s contents while projecting the delivery of a proposition that contrasts it in some way.

4.5 Connectives

Connectives linking the RfC to prior discourse are attested in 30/200 (15%) of the cases. The additive particle wi ‘and’ is the most commonly used connective; the particles bass ‘but’ and ṭayyib ‘okay’ are also attested in a few cases, as summarized in Table 4.

The next example presents the use of the connective ṭayyib ‘okay’ (lit. ‘good’). ʕAbla and her son Yusūf are discussing the prices of flight tickets from Egypt to the United States:

Extract (10): CALLHOME_4644 01:43-01:46

ʕAbla provides an estimation of the ticket’s price in Egypt, and in response, Yusūf seeks confirmation that the ticket will be cheaper if purchased in the United States. The connective ṭayyib prefaces the RfC at line 9. Unlike its derivative ṭab which indexes disagreement (see above (9)), ṭayyib marks understanding and acceptance of what has been said so far and allows thus for transition to the subsequent action (cf. Betz and Deppermann 2021, Mondada and Sorjonen 2021). Noticeably, ṭayyib is compatible with the inference marker yaʕni, and while both link current talk to previous talk, their relative positioning presents an order of interactional tasks whereby displaying understanding and acceptance precedes the more progress-advancing proposal of an inference.

4.6 Tags

Tags are not a prominent design feature of RfCs in Egyptian Arabic. They are found in only a fifth of the cases (n = 39, 19.5%). Formally, tags range from vocalizations to particles, adjectives, adverbs, and small clauses, as detailed in Table 5.

The two main types of question tags in the data are (i) walla-based tags, which consist of the alternative particle walla ‘or’ and an interrogative pronoun (ʔē ‘what’, kām ‘how much’), an indefinite pronoun (ḥāga ‘something’), or an interrogative clause (ʔē il-mawḍūʕ ‘what is the matter’); and (ii) miš kida, which consists of the negation particle miš ‘not’ and the manner demonstrative kida ‘like this’. The choice between the two types of tags seems to be related to the epistemic stance they index: miš kida is used when requesters claim greater epistemic access or authority while walla-based tags indicate a less knowledgeable position. This difference is evidenced in the treatment of tagged-RfCs in the subsequent response: RfCs accompanied by miš kida nearly always invite a confirmation while walla-tagged RfCs are responded with confirmation only in about half of the cases.^[2]

Consider the next two extracts in which walla ʔē and miš kida are used. In (11), Farūʔ enquires about the car which a mutual acquaintance is currently driving after selling her own car. He proposes for confirmation that she drives her brother’s car, but this assumption is disconfirmed:

Extract (11): CALLHOME_4778 10:00-10:02

Farūʔ’s less confident stance is indexed by the adverb bāyin ‘apparently’ and by the appended tag walla ʔēh (l. 2-3). In using this tag, Farūʔ lifts the constraints set by the RfC, by making relevant a response that goes beyond a two-way, yes/no answer. Indeed, unlike miš kida, which reiterates the polarity-based format of the RfC, the alternative-based walla ʔē acknowledges a greater number of response options. It is therefore more congruent with situations in which the requester presents less solid information which as such is more vulnerable to correction.^[3]

In the next example, miš kida is used. Ṣafwat seeks confirmation about the death of a mutual acquaintance:

Extract (12): CALLHOME_5241 08:45-08:50

The tag miš kida is appended to the RfC (l. 12), and while attributing epistemic primacy to the responder by making the appeal for confirmation explicit, it also serves to indicate the requester’s high level of confidence. This confidence may be grounded in the reliable source of the information – a death notice published in the newspaper. The responder’s minimal response makes apparent that he treats the confirmable information as unproblematic, that is, as accurate and comprehensive enough.

4.7 Prosodic design

The prosodic feature that was examined in the study was the final pitch movement of the RfC. In the Egyptian Arabic data, nearly half of the RfCs end with a rising pitch. When a tag question is not appended to the RfC, the numbers are even higher and slightly over half of the cases present a final rising pitch. When a tag question does follow, a larger group of cases presents the final level pitch (Table 6). This tendency may evidence the fact that, even when not clearly integrated into the same unit, the prosodic boundary between the tag and the preceding RfC is for the most part fairly weak or ‘insufficiently marked’ (Auer 2010, 15).

In the majority of cases, tag questions are not integrated into the preceding RfC unit, as shown in Table 7. As for their pitch movement, non-integrated tags mostly present a final rising pitch (Table 8). All in all then, the predominant prosodic pattern of RfCs in the studied data – whether tagged or not – is finally rising. These findings align with previous observations about the rising intonation of polar interrogatives in Egyptian Arabic (Woidich 2006, 358). However, the fact that a considerable number of RfCs in the data present a final falling pitch suggests that: (i) final rising pitch is a typical but not an essential design feature of (what is treated as) a question in Egyptian Arabic; (ii) the falling pitch is a formal feature that distinguishes (a subset of) RfCs from other actions accomplished through polar questions, such as information-seeking questions, requests for action, and offers.

5.1 Responsive actions

The main responsive action made relevant by RfCs is dis/confirmation. Responses in which dis/confirmation was not explicitly stated were classified accordingly as ‘neither’ (cf. König et al. forthcoming). In the Egyptian Arabic data, most of the RfCs receive a confirming response (n = 134, 67%); however, the confirmation may be modified later in the turn ( (27)). Disconfirming responses are less common (n = 36, 18%), and even less so are responses that neither confirm nor disconfirm (n = 30, 15%). The latter is a heterogeneous group of cases including other initiations of repair, other corrections, transformative answers (Stivers and Hayashi 2010), no-knowledge answers (Keevallik 2011), religious invocations expressing thankfulness, wish or hope (e.g. il-ḥamdu li-llāh ‘praise be to God’, bi-zni llāh ‘with God’s permission’), and noncommittal responses.

Responses that decline to commit to a straightforward confirmation or disconfirmation of the prior speaker’s proposal pertain to two types. The first type consists of the response token yaʕni ‘sort-of’. Formally, this token is identical to the discourse marker yaʕni that indicates inference (Section 4.4).^[4] However, yaʕni has an additional use as a noncommittal response ‘with implied reservation’ (Hinds and Badawi 1986, 606). The following extract provides an illustration. After trying to chat in Arabic with her granddaughter in America, Faiza goes back to speak with Ḥātim, her son/father. In line 10, Faiza presents for confirmation the assumption that, despite her apparent weak ability to speak in Arabic, the granddaughter does understand Arabic. Ḥātim’s minimal response via yaʕni conveys the stance that while he does not deny this assumption, he is also not subscribing to its truth.

Extract (13): CALLHOME_4194 13:22-13:24

Responders can also indicate reserved commitment through the turn-initial marker waḷḷāhi lit. ‘by God’. While waḷḷāhi originates in an oath expression, in turn-initial position it serves to project commitment that is relativized to the speaker’s limited access or rights to knowledge and which is therefore more restrained (Marmorstein 2023). Consider, for instance, the next example. Following Saʕīd’s report about a programming course he is about to take, ʕAzmi seeks confirmation about the specific programming languages that the course will cover:

Extract (14): CALLHOME_4455 00:20-00:26

Saʕīd declines to provide a type-conforming, yes-or-no response to ʕAzmi’s enquiry. Instead, he uses waḷḷāhi to assert his commitment to what is evident to him and makes a less constraining proposition, namely, that there exist many programming languages that might be the subject of the course. Thus, a waḷḷāhi-prefaced response circumvents the difficulty of answering the question as formulated by claiming certainty of a more limited scope.

5.2 Response tokens

Responses that include tokens of dis/confirmation constitute the majority of the cases in the studied collection (n = 155, 77.5%). No-token responses which only include repetition comprise a small number of 7 (3.5%) cases. No-token and no-repetition formats amount to 38/200 (19%) of the cases.

Response tokens in Egyptian Arabic range from non-lexical vocalizations (ʔmm) to particles and adverbial forms. While mostly singly produced, particles can be repeated or reinforced by adverbs. Affirmative tokens constitute the largest and most varied group; the negative particle laʔ ∼ la presents a variation of glottalized and non-glottalized forms; the noncommittal yaʕni is attested twice and presents no variation.

The basic affirmative particle ʔāh ‘yes’ is the most common response token in the data (Table 9). It is used in confirmations, either singly or in a token cluster. Typically, ʔāh is singly produced when the confirmation is straightforward and unproblematic, often referring to simple factual information. For instance, in (15), a minimal confirmation via ʔāh is provided in response to a RfC concerned with the time gap between Egypt and the United States:

Extract (15): CALLHOME_5006 01:40-01:43

Responders can go beyond mere confirmation and display a more nuanced epistemic stance or orientation through specific lexical devices. In the next example (reproducing (1) above), Sāmiḥ confirms Fatḥi’s suggestion that his visit to Egypt will take place in October. In using ʔaywa ‘indeed, exactly’, he treats the confirmable not just as true but as accurate, thereby acknowledging the high level of epistemic congruence displayed by Fatḥi:

Extract (16): CALLHOME_5253 00:31-00:33

The adverb ṭabʕan ‘of course’ can cluster with the simple affirmative ʔāh or be used as a standalone confirmatory token. In the following example, Sāmiḥ seeks confirmation that a debtor has paid Fatḥi less than he should have:

Extract (17): CALLHOME_5253 05:45-05:51

Fatḥi uses ṭabʕan to confirm that the assumption made by Sāmiḥ is not simply true but beyond question. Notice that in this affiliative context, ṭabʕan is not used to index a problem with the askability of the question (Stivers 2011), but to upgrade the confirmation by rejecting any alternative, thus making the complaint about the debtor stronger.

The negative particle laʔ ∼ la ‘no’ is used both as a disconfirmatory token and as confirmation of negatively polarized RfCs. For instance, in the excerpt below Aḥmad seeks confirmation that his father in Egypt is not in need of anything that he can send him from the US. This RfC thus also serves to implement an offer. However, the two actions set up different preferences for the upcoming response (cf. Schegloff 2007, 76–8): as a negatively formatted question, the RfC is optimized for a ‘no’ answer, while as an offer it should be preferably met with acceptance. The father’s response, initiated by la, is evidently oriented to the negative design of the RfC and confirms the assumption it puts forth while declining the offer:^[5]

Extract (18): CALLHOME_5203 21:49-21:51

5.3 Clusters of response tokens

Clusters of response tokens, in which two or more tokens of similar or different types occur in the same prosodic unit, are rare in the Egyptian Arabic data (7 out of 155 token-including responses). The next excerpt (reproducing (10)) provides an illustration. Yusūf seeks confirmation that ʕAbla’s flying ticket from Egypt to the US will be cheaper if purchased in the US:

Extract (19): CALLHOME_4644 01:43-01:46

ʕAbla delivers a strong confirmation consisting of a repetition of the affirmative token ʔāh (l. 12). In producing a ‘multiple saying’ (Stivers 2004), ʕAbla seems to convey the stance that further contemplation of this topic is unnecessary as she is completely certain of the fact that the ticket is best purchased in the United States. This stance is made explicit in ʕAbla’s following reassertion that the ticket will be cheaper in the United States (l. 14).

5.4 Position of the first response token

Typically, a confirmation token is placed at the beginning of the responsive turn. Thus, in 52 out of 68 (76%) instances of non-minimal confirmations, the token is turn-initial. This strong tendency can be explained by the fact that, even if more complex work is accomplished by the responsive turn, confirmation is prioritized as the ‘official business’ made relevant by the RfC.

However, in a small subset of cases, the confirmation token is placed at the end of the responsive turn (13/68, 19%) or is repeated twice, at the beginning and the closing of the turn (3/68, 5%). In the next example, the confirmation token occurs in turn-final position. Calling from the United States, Šarīf seeks confirmation that the current time in Egypt is 7 a.m:

Extract (20): CALLHOME_4345 00:24-00:29

Salmān is not immediately confirming: he initiates his response with a repetition of the first segment of the confirmable and only then produces the affirmative ʔāh ‘yes’. It is quite plausible that Salmān is producing this response while simultaneously checking the time (observably, line 3 is delivered with decreasing speed and stretch of the final syllable). Thus, the provision of a straightforward confirmation becomes possible only after the action of verifying the time has been completed.

In the next example, the confirming turn is enclosed by an initial and a final token of the affirmative ʔāh. Yeḥya seeks confirmation about the death of a mutual acquaintance:

Extract (21): CALLHOME_5595 02:13-02:18

The confirming turn unfolds in three intonation units: the first production of the affirmative ʔāh is followed by a repetition of the confirmable ‘he died’ to which another token of ʔāh is appended. In using repetition, the responder in this case underscores his epistemic primacy (Schegloff 1996, Heritage and Raymond 2005, Enfield et al. 2019). Flanked on both sides by ʔāh, this epistemic work is marked off as separate from the plain action of confirmation.

5.5 Minimal and non-minimal responses

Minimal responses to RfCs consist of a freestanding response token. In the studied data, they comprise 58/200 (29%) of the cases. While minimal responses can implement a confirmation and ‘nothing more’ (Enfield et al. 2019; see (15)), they can also serve a more nuanced interactional work. Specifically, in selecting a particular token, the responder is able to communicate a certain stance or orientation towards the confirmable and/or the requester which is interactionally meaningful and consequential ((16) and (17)).

Non-minimal responses, which besides a response token also include repetitions and/or other expansions, are more common than minimal responses and comprise 93/200 (46.5%) of the cases. Other formats of response, which do not include a response token, but can involve a repetition, make the smallest group with 49/200 (24.5%) of the cases. Non-minimal and other formats of confirmation are discussed in the next subsection (5.6). The remainder of this subsection is concerned with minimal and non-minimal formats of disconfirmation.

The great majority of disconfirmations (25/36, 69%) present a non-minimal format initiated by the negative response particle laʔ ∼ la. Typically, non-minimal disconfirmations include a correction, an account, or some other form of elaboration. In the next example, ʕAlā seeks confirmation that Muʕtazz has a day off ‘today’. The disconfirmatory response consists of a long glottalized variant of the negative particle, laʔʔa, and a subsequent correction ‘tomorrow’:

Extract (22): CALLHOME_4486 06:11-06:14

Minimal disconfirmations are rare in the data (4/36, 11%) and present as a rule the glottalized form laʔ. In the next excerpt, Hāni, who is calling from the US, seeks confirmation that a soccer game in Egypt is taking place ‘today’. After Rafʕat disconfirms (l. 5), Hāni issues a negatively formatted reconfirmation request which is plainly responded with laʔ:

Extract (23): CALLHOME_4864 02:17-02:19

Noticeably, in his next turn, Hāni produces a repetition of the confirmable information before acknowledging its receipt via ‘oh’ (l. 8-9). This elaborate way of response is not treated as a closure of the topic, but as an invitation to pursue it further, as indicated by Rafʕat’s subsequent (and perhaps forced) speculation about the time of the match. This form of sequence expansion and the overall low frequency of the minimal format suggests that minimal disconfirmations are generally treated as insufficient responses to RfCs.

5.6 Full and expanded repeats

Full repeats occur in 35 out of 134 (26%) of the confirmatory responses. While a repeat can constitute by itself a confirmation, in the great majority of cases (28/35, 80%), it is combined with a response token, often placed subsequent to it (20/28, 71%). The repeat can present a partial or full reproduction of the confirmable.

Responding by a full repeat is syntactically determined when the confirmable itself is phrasal or consists of a minimal clause (i.e., an inflected verb). In cases where the response reproduces ‘omittable’ material, other motivations may be at play. In the next example, Mariam is a participant in the CALLHOME project calling from the US. Muna, her relative, seeks confirmation that all her family members were similarly granted a free telephone call:

Extract (24): CALLHOME_5047 02:26-02:36

By not simply confirming but also fully repeating the proposition (with necessary indexical shift) ‘everyone at our house’ (l. 9), Mariam can be heard to orient to two things. First, the unconvinced stance displayed by Muna, who keeps interrogating her about the terms of the recording. In articulating the confirmation in full, Mariam can express commitment to each of its parts and thus respond to the doubt implied by Muna’s questioning. Second, since her response comes after a repair sequence that she has initiated (l. 6-7), Mariam can display her understanding by repeating in full the trouble source.

Repetitional responses can also go beyond a verbatim reproduction and expand the confirmable information. As with other types of repeats, an expanded repeat can be combined with a response token or satisfy as a confirmation in itself. While both token-including and token-less formats go beyond the confirmable proposition, they seem to display a different orientation towards it; specifically, when no token is produced, the expansion lends itself to interpretation as corrective rather than additive. Consider the following two excerpts. In (25), Layla is telling Muṣṭafa about the funeral of a mutual acquaintance and the group of people who attended it. Muṣṭafa seeks confirmation that this group included all of the deceased’s friends:

Extract (25): CALLHOME_4213 21:03-21:09

Layla’s response is initiated with the affirmative token ʔāh ‘yes’ and a repetition of the confirmable ‘friends’ which is incrementally expanded to additional categories of attendants. Notice that this list is prosodically constructed as an open one, with each item packaged in its own intonation unit and showing a similar pitch movement pattern (Couper-Kuhlen 1986, 150, Sánchez-Ayala 2003). Moreover, the list is concluded with the manner demonstrative kida ‘like this’ which indicates that the enumerated items are not exhaustive but illustrative of the possible categories that make up the set.

In the next example, the response consists of an expanded repetition with no confirmatory token. Šadia tells Hišām about the entertaining tour she had organized for relatives who came to visit her in the United States. Hišām’s RfC at line 5 is concerned to clarify the specific identity of the participants in this tour:

Extract (26): CALLHOME_4826 11:59-12:01

Šadia’s response goes beyond ‘him and his wife’ to include ‘the daughter’ and ‘everyone’. This expanded repeat is packaged in a single intonation unit and constructed as an exhaustive description of the entire class of participants. This design suggests a more independent version which resists the implications of the original proposal, namely, that the group of participants was selective or small. While not a downright disconfirmation, the lack of an explicit confirmatory token and the formulation of the expansion propose greater epistemic incongruence between the requester and the responder.

Interestingly, departures from the confirmable can also follow after a straightforward confirmation. The next extract (reproducing (3)) is another example in which the participants are concerned with the time gap between them. ʕAzmi seeks confirmation that the time in Egypt is a quarter to six:

Extract (27): CALLHOME_4455 00:34-00:39

Saʕīd’s response is initiated with a strong confirmation: the affirmative ʔāh is reinforced by the adverb biẓẓabt ‘exactly’ which indicates the accuracy of ʕAzmi’s proposal. However, as he goes on to produce a repetition of the confirmable, and apparently while checking the actual time (as the noticeably stretched production of illa ‘except’ in line 5 appears to indicate), Saʕīd replaces ‘quarter’ with ‘ten minutes’. Compared to the ‘conclusive’ confirmation in (20), this example shows the possible ‘premature’ delivery of confirmations in real-time conversation which seems to underscore the widely observed preference for agreement in interaction (Sacks 1987, Pomerantz 1984, Raymond 2003).

Sequential structure

One line

one intonation unit

→

refers to a line of transcript relevant in the argument

[ ]

overlap and simultaneous talk

=

fast, immediate continuation with a new turn

Breaths

°h/h°

in-/outbreaths of appr. 0.2–0.5 s

°hh/hh°

in-/outbreaths of appr. 0.5–0.8 s

°hhh/hhh°

in-/outbreaths of appr. 0.8–1.0 s

Pauses

(.)

micro pause, estimated, up to 0.2 s

(-)

estimated pause of 0.2–0.5 s

(--)

estimated pause of appr. 0.5–0.8 s

(---)

estimated pause of appr. 0.8–1.0 s

(1.5)

measured pause of 1.5 s

Lengthening

:

lengthening, by about 0.2–0.5 s

::

lengthening, by about 0.5–0.8 s

:::

lengthening, by about 0.8–1.0 s

Final pitch movement

?

rising to high

,

rising to mid

–

level

;

falling to mid

.

falling to low

Intra-linear notation of accented syllables

ˆsyll

high rise-fall pitch accent

Laughter

haha hehe

syllabic laughter

<<laughing> text>

laughter accompanying speech with indication of scope

<<:-)> text>

smile voice with indication of scope

Loudness and tempo changes, with indication of scope

<<len>text>

lento, slow

<<f>text>

forte, loud

<<all>text>

allegro, fast

<<cresc>text>

crescendo, becoming louder

<<dim>text>

diminuendo, becoming softer

<<acc>text>

accelerando, becoming faster

<<rall>text>

rallentando, becoming slower

Transcriber comments and unintelligible speech

((text))

inferred but unexpressed elements (in translation line)

(xxx) (xxx xxx)

unintelligible part with indication of syllable number