The turn-constructional unit and transition relevance place

How talk is organized into units has long been a topic that has interested many linguists and some sociologists. How do speakers construct units of talk and how do their recipients recognize them? Early studies of this phenomenon contrasted spoken language with the long dominating written language and analyzed narrative monologues and speakers’ resources in marking unit boundaries in them (for example, Chafe 1979). These studies introduced structures in talk such as the idea unit and the intonation unit.

However, these earlier studies neither adopted an orientation towards interaction nor the concept of the interactive organization of talk. The paper by Sacks, Schegloff and Jefferson (1974) was seminal in, among other things, that it highlighted the accomplishment of talk and units of talk being a joint enterprise between the speaker and his/her recipient(s). This starting point enabled the appreciation of many features of talk that now are thought of as central, including talking simultaneously. This conversation analytic study by Sacks and his colleagues assumed that the basic unit of talk is the turn-at-talk or a turn-constructional unit and focused on how turn taking is managed in everyday interaction.

From this theoretical bedrock, the current section will discuss the following questions: How are turn taking and turn transition organized?

Furthermore, what resources do speakers use to construct their turns-at-talk and to interpret and project the construction and completion of their co-participants’ turns-at-talk? And how do these various resources relate to each

other? Another important question is whether there are language or culture specific matters involved. More specifically, what types of units are there in conversational talk? What is then the importance of these units for turn taking and turn transition? And finally, what do we know about overlapping talk, which is thought to be a departure from the norm?

Spontaneous talking among people in everyday life is accomplished turn-by-turn. As Sacks et al. observed in their seminal paper (1974; on turn-taking organization,¹⁶ see also Hakulinen 1997), in ordinary conversation, turn transition not only occurs but recurs. A conversational turn is therefore usually defined as a chunk of talk that is situated between speaker transitions, in that the speaker change delimits a turn (ibid.). In addition, turns¹⁷ are composed of turn-constructional units,¹⁸ which can be of various sorts and sizes (Sacks et al. 1974). In essence, this is the basic conversation analytic answer to the question of units of spoken language: they are phenomena that occur in interaction, typically among several participants.

The turn-constructional unit (TCU) has been initially characterized as a

“unit-type with which a speaker may set out to construct a turn” (ibid. 702).

This perspective emphasizes the function of units in conversation. In other words, speakers construct turns-at-talk from them. Let us consider a simple example from Sacks et al. (p. 721):

(2.1)

1 Tourist: Has the park cha: nged much, 2 Parky: Oh:: ye:s,

3 (1.0)

4 Old man: Th' Funfair changed it 'n [ahful lot [didn' it.

5 Parky: [Th- [That-6 Parky: That changed it,

This example demonstrates, among other things, the diverse composition of turn-constructional units. Here every line (except for the fifth, perhaps) represents one such unit, and as we see from above, they take various forms.

In this connection, the term utterance¹⁹ is also used; it is defined as “a functional whole that occurs in speech” (ISK § 1003). It follows then that an utterance can be a clause, but it can also be a particle or a phrase (ibid.).

Furthermore, utterances make up a turn or a part of it (ISK § 1004); they can thus be equated with TCUs. An important feature of the turn-constructional unit is that it always occurs and exists only in context; its production and recognition are context-dependent and context-sensitive processes. The TCU and its boundaries are therefore determined only in its context of occurrence

16 Finn.vuorottelujäsennys, Est.vooruvahetusmehhanism

17 Finn. vuoro, Est.voor

18 Finn.vuoron rakenneyksikkö, Est.vooruehitusüksus

19 Finn.lausuma, Est.lausung

and by the conversational parties jointly and, as Ford et al. (1996: 428) observe, it is “contingent and interactionally achieved.”

When participants interact, they need to know when the turns by others will end and when it will be their turn to start talking. Conversation analysts have called this point in time the transition relevance place²⁰ and it is by reference to these places that participants organize turn transitions.

According to Sacks et al. (1974), a transition relevance place (TRP) emerges at the end of each TCU, at its first possible completion point (for illustrations of these points, see the example above, especially the places where Parky attempts to take a turn in line 5). The TRP is therefore a place where a legitimate turn transfer occurs from one speaker to another and, according to Sacks et al. (ibid.), all turn transfer is organized around TRPs. (For later discussion on TRPs and TCUs in different positions, for example, see Houtkoop & Mazeland 1985, Selting 2000, Ford 2004 and below.) However, TRP is not merely the narrow slot or the short gap between TCUs, but can be thought of as extending somewhat in both directions, backward and forward (in time). This means that a TRP begins (or can begin) a few sounds, syllables or even words before the end of the current TCU, and it can extend to the very first elements of the subsequent TCU (for example, see Jefferson 1983, 1986).

The specifics of turn transition in the sense of who speaks next are not explored in depth within the scope of this study. However, as these questions are significant for grasping the full picture of turn taking, I will briefly summarize the most relevant literature on these phenomena.

There are certain ways in which the transition of turns from one speaker to another is accomplished so that a simple rule-set applies. According to Sacks et al. (1974: 704), the turn-taking rules for ordinary conversation are the following – in this order:

1. At the end of a TCU, at a first TRP, one of the following will occur:

a) If the speaker in their turn has selected the next speaker, the latter (and no one else) is entitled and obliged to take the next turn, b) if the next speaker has not been selected, anyone (a first starter)

may take a turn,

c) if no one else takes a turn, the prior speaker may continue.

2. If at the first TRP rule 1c has operated and the same speaker has continued, the same rule-set a-c will apply again at the next TRP (and the next) until the turn has been transferred to another speaker.

(Sacks et al. 1974: 704.)

The particulars of the next speaker selection have later been specified by scholars such as Lerner (2003), who examines the role of gaze direction, address terms and other, more subtle ways to select a next speaker (see rule 1. a), and by Mondada (2007), who focuses on pointing as a practice to self-select oneself as a next speaker (see rule 1. b).

20 Finn.siirtymän/vuoronvaihdon mahdollistava kohta, Est.voorusiirdekoht, vooruvahetuskoht

We notice that in the system described above, turn transition is organized exclusively around transition relevance places (TRPs) (see Sacks et al. 1974:

708). Interactants may, however, also start talking elsewhere than at the TRP, which regularly results in overlapping talk. This is one of the observations that underlie the research interest of the current study.

A participant who wishes to talk next not only waits and determines when the ongoing turn by another has come to a completion, but instead uses a range of resources to predict a possible end point to the turn. Indeed, a highly relevant and determining feature of turns, TCUs and TRPs is that their possible ends are projectable. This was first noted by Sacks et al. (1974: 720), who state that turn-constructional units “have points of possible [- -]

completion, [- -] which are projectable before their occurrence.” This is also the cornerstone of a speaker’s ability to anticipate the turn transition places and to position his/her next turns accordingly. (On projection, see Auer 2005 and chapter 3 below.) In the following, I will provide an overview of the literature that analyzes these phenomena in different languages. The relevant questions are therefore the following: How do speakers know when the turn (-unit) is going to be complete and turn transition can occur? How are turns and TCUs organized, and which resources do speakers use to construct and recognize them?

Within the past decades, researchers in conversation analysis have been engaged in an extensive discussion on the construction and nature of units of talk. A number of scholars have analyzed the build-up and recognition of units and especially the resources that speakers use when accomplishing these tasks. The earliest paper to address this question – and actually the paper that first introduces the whole topic – emphasizes the syntactic structure of units and its importance. Sacks et al. define the possible ending points of TCUs with reference to syntax only (1974: 720–721; for an illustration, see also the example provided above). Syntax has been the focus of investigation in later studies as well, especially in the comparisons between English, the language that the first (and still most) studies investigate, and structurally very different languages such as Japanese (for example, Lerner & Takagi 1999, Tanaka 2000). However, several contributions, Ford and Thompson (1996) being among the first, have demonstrated that prosody – including intonation, rhythm, and phonetic features of talk – is at least as important in constructing and interpreting the trajectory and completion of turns. This was briefly mentioned already in Sacks et al. (1974: 721–722)²¹; later studies have demonstrated how speakers of various languages (mostly English but also languages such as Japanese and Finnish), in constructing turns and organizing turn transitions, orient to aspects of language such as speech rhythm, prominent pitch peaks (accents),

21 Sacks et al. only mention that “discriminations betweenwhatas a one-word question and as the start of a sentential (or clausal or phrasal) construction are made not syntactically, but intonationally”

(ibid.).

intonational patterns, loudness changes, sound lengthening, as well as changes in tempo and voice quality (Auer et al. 1999, Chafe 1994, Couper-Kuhlen 1993, Cutler & Pearson 1986, Fox 2001, Local & Kelly 1986, Local et al. 1985, Ogden 2001, 2004, Schegloff 1998, Szczepek Reed 2004, Tanaka 2004, Wells & Corrin 2004, Wells & Macfarlane 1998, Wells & Peppé 1996).²² Furthermore, the relative significance of the features is discussed in the literature.

Acknowledging the importance of both syntax and prosody, many researchers have focused on their interplay in unit construction (for example, Bockgård 2007, Helasvuo 2001a, J. Lindström 2008, Selting 1996, 2005, Steensig 2001a, 2001b, Tanaka 1999, Tao 1996). In different languages, the precise roles that syntax and prosody play in interaction may and do vary, as the scholars mentioned above have established for Swedish, Finnish, German, Danish, Turkish, Japanese and Mandarin. Concerning the languages that this study examines, the study of Finnish, Helasvuo (ibid.) is the only one that deals with these questions.

Not only syntax and prosody, but the actions and sequential position of the turns are also generally widely acknowledged in the CA literature to be significant and to contribute to turn-unit construction and projection (for example, see Houtkoop & Mazeland 1985). This aspect is rather often called

“pragmatics” – for instance, Ford, Fox and Thompson (1996: 429) use that term to cover “the sequential location and the interactional import of an utterance.” Sometimes the term “semantics” is also used for this feature (for example Oreström 1982, 1983). Schegloff (1996a: 59) speaks of a pragmatically completed turn when the turn “recognizably implements an action.” However, not all studies that analyze turn construction and projection take into consideration the import of actions specifically. Some exceptions include Huiskes (2010) on Dutch and Steensig (2001a, b) on Danish and Turkish, in addition to the scholars mentioned above (see also Ford & Thompson 1996, Schegloff 1988). One issue that unifies many of these scholars is that they see “pragmatics” as a challenging feature to take into account. Ford and Thompson admit that their analysis of this remains

“intuitive and provisional” (1996: 150), and they even include intonational completion in the parameters for pragmatic completion, which has later been criticized. Ford and Thompson make a distinction between two types of pragmatic completion. These types are, first, local pragmatic completion that is a place where the speaker projects more to come but a co-participant can take a minimal turn, while the second type is global pragmatic completion that refers to a point where no more talk by the current speaker is projected, i.e., the whole story or other agenda is complete. Nonetheless, these two types both count equally as “pragmatic completions” for their study. (Ibid.

150–151.)

22 For a different viewpoint concerning the importance of prosody, see De Ruiter et al. 2006.

An additional cue that has been found to provide essential information for turn transition was first introduced by Charles Goodwin (1979, 1981, 1995), who acknowledged the bodily-visual makeup of turns and units and the bodily-visual cues for turn transition. The research conducted by Schegloff (for example, 1984b, 1988) is also important in this respect. These bodily-visual cues include gaze, gestures, body position and movements, facial expressions, material objects in the scene and the surroundings in general. In later contributions on turn construction and turn taking, these facets of interaction have received more attention in studies of talk-in-interaction that have been conducted on a variety of languages including Lao (Enfield 2009), Finnish (Kaukomaa et al. 2013, 2014, forthcoming, Laury & Ono 2014, Seppänen 1998), English (Ford et al. 2012, Kendon 2004), Italian (Kendon 2004), Chinese (Li 2013), French (Mondada 2006, 2007), Ilokano (Streeck 1995, Streeck & Hartge 1992), and Thai, German, and Japanese (all three in Streeck 1995). These scholars have demonstrated the rich ways in which speakers deploy various resources for the tasks related to turn taking in combination with syntax, prosody and pragmatics.

Studying and describing units of talk by relying a priori on linguistic units – as in some of the works mentioned above – has recently drawn criticism.

For example, Ford, Fox and Thompson (2013) suggest that this type of analysis is “neither adequate nor appropriate” when investigating naturally occurring interaction from the participants’ perspective. Instead, they argue that analysts need to adopt “a descriptive meta-language” that is based on the participants’ categories and on actions as they emerge online. These authors further argue that traditional linguistic categories can be used when empirical evidence warrants that they are relevant for the participants.

In addition to the resources for turn construction discussed above, conversation analytic literature has established that various repair practices (Fox et al. 1996, Jefferson 1974, Kurhila 2006, Schegloff 1979a, Schegloff et al. 1977) and recipient monitoring (for example, Goodwin 1979, Houtkoop &

Mazeland 1985) are also relevant for turn construction. However, these facets will not be further discussed in the present analysis.

A crucial feature in constructing turns is that they are always flexible and negotiable between the participants at the very moment of their interaction (Schegloff 1987a). The turns, their elements and factors such as their length (for example, Local 1992) are manipulatable and exploitable by the speakers according to interactional contingencies and exigencies. Thus, turns emerge in real time. Some examples of this specific line of research include Selting’s study (2001) on the fragments of units and their use in interaction in German data (see also Kim 1999 for Korean), and Koivisto (2011), who investigates the occurrence and the use of turns that end in conjunctions in Finnish.

Whereas these turns are traditionally thought of as incomplete, Koivisto demonstrates how participants can treat them and orient to them as complete, and Koivisto therefore re-analyzes the “conjunctions” as final particles.

That turns and (grammatical) units are “(semi-)permeable” by nature has been convincingly demonstrated by Lerner (1991, 1996; also 1992, 1993, 2002, 2004a; however, see earlier studies by Sacks 1992a: 647ff, 1992b:

437ff). Lerner has also provided evidence that systematic places in turns occur where a co-participant can (and indeed frequently does) come in and join in their construction.²³ Co-constructing a “compound” turn/TCU is also a systematic place for the occurrence of overlapping talk, and this will be discussed in more detail in chapter 6 below.²⁴ This line of investigation calls into question the widely accepted principle proposed by Sacks et al. (1974) that turn transfer is organized solely around TRPs. This theme will also be addressed later in this dissertation, especially in chapter 9.

Various scholars have illustrated how turns may be extended or expanded even after their projected completion (for example, Ford et al. 2002, ISK § 1052–1078 and references therein, and Schegloff 2001). The construction of turns is therefore incremental by nature. The studies dealing with this phenomenon have focused on the formatting of continuations/increments (Auer 1996) in a cross-linguistic perspective (Couper-Kuhlen & Ono 2007, Luke et al. 2012, Vorreiter 2003), on the specifics of same-speaker versus other-speaker increments (Lerner 2004b, Sidnell 2012a), and on the intersection between increments and clause combinations (Couper-Kuhlen 2012a, Seppänen & Laury 2007). These studies are yet another demonstration of the contingent and flexible nature of both turns and TCUs, a claim which will be corroborated by the current study as well.

Early studies in the CA literature addressed the question of the turns that consist of multiple units and how a TRP is projected in them. For instance, the Sacks et al. (1974) model for turn construction and turn taking includes an apparent weakness: it states that at the possible completion point of each TCU, turn transfer becomes timely. Yet speakers do frequently produce turns with more than one TCU without involving turn transition attempts from the co-participants. Some scholars have analyzed the mechanisms by which speakers manage to get and produce these long turns (for example, the earliest studies in Sacks 1992b: 137ff., 1992b: 175ff, 1974, 1978, Jefferson 1978), but others have concentrated more on the definition of a unit or a TCU with reference to the following situations: Are there different types of TCUs, such as final and non-final?; How are longer stretches of talk organized and

23 Some other scholars have also studied similar phenomena; they only have not always referred to them as co-constructed turns, as various terminology has been used (for example, see Falk 1980, Helasvuo 2004, Goodwin 1984, 1987, Mandelbaum 1987, Tainio 2000, Szczepek Reed 2006); for a discussion, see section 6.1.3.

24 Lerner (1996) has examined what are referred to as co-constructed turns and the recipient’s possibility to produce talk starting up at a point where the current turn is not yet completed. Some, but not all of my cases resemble those of Lerner’s in terms of how the previous turn is structured, where the next turn begins, and what the turn is doing. The starting points and the questions in my work and his are different, and nevertheless there are several interfaces between them.

how should they be analyzed? The earliest solution was proposed by Houtkoop and Mazeland (1985), who suggest a distinction between TCUs and open and closed discourse units. However, their idea and terminology has not received wide acceptance.

A more influential work in this respect has been conducted by Ford, Fox and Thompson (1996). They first aimed at defining a TCU and at counting the number of TCUs in long turns but end up emphasizing the multitude of practices (for example, instead of focusing on syntax only) in constructing participation and in projecting turn trajectories. They highlight the manipulatability and negotiability of turn end points and convincingly demonstrate that by concentrating on “units” only, the true nature of everyday talk-in-interaction can be easily missed. Along the same lines, Ford (2004) heavily emphasizes the role of contingency for unit construction and

A more influential work in this respect has been conducted by Ford, Fox and Thompson (1996). They first aimed at defining a TCU and at counting the number of TCUs in long turns but end up emphasizing the multitude of practices (for example, instead of focusing on syntax only) in constructing participation and in projecting turn trajectories. They highlight the manipulatability and negotiability of turn end points and convincingly demonstrate that by concentrating on “units” only, the true nature of everyday talk-in-interaction can be easily missed. Along the same lines, Ford (2004) heavily emphasizes the role of contingency for unit construction and