8.5. Functions in discourse

Detachment constructions have been first described at sentence level, with examples from written texts or created by the authors. These types of approaches are still widespread in different domains of investigation: for instance, syntactic analyses most often use examples of sentences, considering the grammaticality or non-grammaticality of the constructions under invest-igation. This is, of course, not true of all syntactic works, as there are also more and more corpuses of oral language available. The generative framework also mostly makes use of these types of examples.

When analysing real examples from conversations, sentence is not an appropriate category. The detached constructions are by definition linked to a main clause, which is the Rheme of the utterance, but they are not limited to only one clause: the referent of the detached element can be present in previous discourse or can be developed after its occurrence in a detached constituent.

Therefore, we always have the level of discourse that is present.

Fernandez-Vest (2004a) makes a demonstration of the textual functions of final detachments by arguing that they are not a mere repetition of the Theme, but serve as elements which assure the circular cohesion of the sequence and have a special role to play in recalling its main object. This constituent has also a phatic function, as it contributes to the confirmation of a common ground for


the discourse participants (Fernandez-Vest 1994: 202). The examples of K. Aijmer about English show that final detachments tend rather to have a role in the social plane: the speakers use them in order to create intimacy between the participants in situations where there is already some common ground;

often, speakers use evaluating or expressive terms (Aijmer 1989: 153, 150).

Depending on the type of corpus, on the person who is speaking, the occurrence of these elements can be quite variable. They have been studied mostly in the framework of one sentence, but as much as possible, in the present analysis their textual dimension will also be considered. These constituents seem to be a cross-linguistic phenomenon in oral language, however, the languages do not use them with strictly identical functions and quite naturally, differences lie also in the frequence of their use and in the prevalence of one or another type of such constructions. H. Sahkai (2003) has argued that, in Estonian, final detachments do not have the same functions as those in some Indo-European languages, like French or English, even if syntactically there is no difference between them. According to her argumentation (in a syntactic framework), as in Estonian, ‘topicalization’ in general is not usually performed by ‘dislocations’, final detachments do not either carry this function, but could be rather considered as an ‘occasional repair strategy’ (Sahkai 2003: 84, 86).

However, in our corpus, there seem to be at least two groups of examples which may form a continuum: one group that can be associated to the repair function in discourse and another group where the repair function does not play a role at all and where it is clear that the construction has already been planned as such.

Consequently, my approach here is somewhat different, as we observe on the textual level two different functions of these detached constructions and do not try to link one principle defined in syntactic framework (topicalization) to these constructions in Estonian: final detachment has its proper functions in relation to the thematic continuity and the cohesion of the discourse that cannot be assimilated to the thematization (or ‘topicalization’). Moreover, final detach-ment is not just a ‘repetition’ of the Theme when considered in the framework of the discourse, as we will show further, and this can be demonstrated especially at the textual level: at the sentence level, examples like ‘she is not here, your wife?’, do not show the functioning of this constituent in discourse, where it carries different functions and participates in more complex articulations – this utterance can not occur in any situation, but there are always certain conditions under which it is possible (common knowledge, appropriate situation, a certain relation between the participants etc.).

The next excerpt illustrates the fact that the final detachment can have multiple roles in discourse: it gives more cohesion to the discourse, helps to specify the reference while the whole sequence is ‘under construction’ (the correction of the pronoun neid >seda (plural demonstrative need in partitive

>see ‘it’in partitive) and it also marks the end of the sequence by assuring circular cohesion (the use of the term struktuur at the beginning and at the end of the sequence). At the beginning it is said about the restructuring of the forest


management system that some kind of structure already exists; then the speaker tries to introduce a plural pronoun neid, probably wanting to develop the plural referent metskonnad, but abandons it, probably because it is too specific to be described by a verbal element and after a short pause resumes with an utterance that contains the singular pronoun seda which will be referred to by the final lexical element that carries the same casual marking (partitive) and is clearly associated to the more general referent metskondade struktuur, introduced at the beginning of the sequence.


A: a kas mingi kava või mingi programm on juba olemas ka või but Q some plan or some program be.3sg already existing also or on see (.) aint mingi tuleviku (.) [suunat]

be.3sg it only something future.ILL direct.PPP E: [täpset] programmi ei ole veel=hh E: exact.PART program.PART NEG be3.sg yet

K: ((ebakindla häälega)) präägu on nagu see alu: jah seline nagu:

((uncertain voice)) right_now be.3.sg like DEM base yes such like metskondade struktuur noh on midagi on forest_managegement_units.pl.GEN structure PRTCL be.3sg something be.3sg nendel juba olemass, aga täit noh täpselt veel maha pandud they.ADE already existing but full.PART PRTCL exactly yet down put.PPP nagu ei ole, (1.0) aga e midagi on nendel ikka olemass, like NEG be.NEG.3sg but something be.3sg they.ADE PRTCL existing ja:= ja neid hakatakse nüüd iga:=iga (1.0) pidevalt and and DEM.pl.PART begin.IMPS now every every constantly

hakatakse seda siis välja töötama seda struktuuri. (1.0) begin.IMPS DEM.PART then out work.INF2 DEM.PART structure.PART K: ((hesitating voice)) ‘Right now there is like this: yes like a structure of forest management units something like this they do already have, but the whole system is not determined yet (1.0), but something they have already and and these will be now every every (1.0) it will be progressively built up, this structure.’


In different frameworks, several researchers have advanced the idea that final detachments do not only serve to clarify the referent: for example, Horlacher and Müller (2005), in the Conversation Analysis framework discuss their


disambiguating function in the interactional perspective and tend to think that this is not their primary function. Based on the examples of their corpus, they propose that the ‘right detachments’ serve to manage a disagreement in the discourse interaction. On the pragmatic level, they found that these constituents are used mostly in the context of (positive or negative) assessment. It is true that this claim can be supported by some evidence from my corpus: there are indeed examples that contain different evaluation devices, such as adjectives, adverbials of informal register, but it is difficult to analyse their frequency, especially because the detached constituents appear naturally in copular con-structions, and oral conversations are anyway often about assessment, charac-terization, personal impressions etc. Horlacher 2012, in her PhD dedicated to final detachments, lays emphasis on the so-called incremental function of final detachments in the framework of Conversation Analysis (a more detailed account is given in the next section 8.6.).

The recent Finnish Grammar describes this construction as a stabilized and grammaticalized device whose function is to specify and characterize the constituent referred to by the pronoun. From the formal point of view, the Grammar states that often the particle ni is placed between the detached lexical element and the main clause (Hakulinen et al.: 1013). As an example that invalidates the claim according to which the main function of this structure is to add a clarification, the grammar proposes a dialogue where the listener does not wait for the detachment construction to be added, but rather begins his/her own turn.


V: Onks nää tuonne Kuopioon tehty [nää matkat.

be-INTERR DEM.pl there Kuopio-ILL do-PPP DEM.pl travels

A: [Nii o.

So be-pl.

V: ‘Are they made to Kuopio, these trips?

A: Yes, they are.’

8.6. Distinction between final detachments and

‘afterthoughts’ or repairs

This controversial distinction has been discussed for several languages and from different approaches: is it necessary and relevant to make a distinction between afterthoughts or self-initiated repairs (cf. Schegloff et al. 1977 ) and final detachments, the first constituent being understood as an expression which serves only to eliminate potential referential ambiguity and can thus also contain other features that refer to a repair mechanism (reformulations, hesitations etc.) and the second as primarily an information structuring device, which has other motivations of use. Ashby (1994) has shown the existence of a phonological


distinction between final detachments and afterthoughts in French; T. Fretheim (1995) has discussed this distinction for spoken Norwegian, arguing that these two types of constituents cannot be assimilated and that there is no continuum between these categories: afterthoughts are by their structural characteristics more dissociated from the main clause (‘Norwegian RDs (right detachments) belong to grammar, afterthought phrases do not’ (ibidem: 53)). There are more examples of analyses that show the macro-syntactic independence of afterthoughts (Ziv 1994, Ziv & Grosz 1994, Lacheret-Dujour 2003).

Ziv and Grosz (1998: 295–298) draw this distinction using examples of English and referring to previous works (supported by Hebrew) and argue that there are mainly four differences between these two phenomena:

1. Position: final detachments (‘right dislocations’ by the authors) can only occur in clause final position, whereas in afterthoughts it can occur in other positions as well; the following example (94) shows an NP in afterthought:


I met him, your brother, I mean, two weeks ago.

2. Possibility of extracting this element outside of its clause: this is not possible in case of a final detachment, as in the next example:


*The story that he told us was interesting, Bill. (where NP is non-vocative) This is however possible in the case of afterthought:


Remember the two of them were telling us all sorts of stories?

Well, the story that he told us was very interesting, Bill, I mean.

Here, the only difference is the addition of another sentence, for setting the context and making a reference to two possible items, which allows then a looser referential link in a contextualized sentence, typical to oral language.

3. Referentiality: in final detachments, the pronominal must be coreferential with the NP, whereas in afterthoughts the reference can be corrected by the final constituent as in the next example.


I met John yesterday, Bill, I mean.

It is not clear, however, if this distinction should also be maintained in cases where the contrast between two referents is not so evident: there can be other


types of clarifications such as grammatical cases or other relations. Moreover, in this sentence there is no pronominal used, but the repair of the referent.

4. Intonation: according to the authors, final detachments seem to constitute a single contour with no pause between the two elements, whereas in the case of afterthoughts there is a distinct pause and the two elements form two different intonational units.

This point can be corpus-/language-specific, but if the authors consider specific repair constructions as afterthoughts, it is plausible that there are pauses or other markers between the two parts of the utterance.

These distinctions seem thus somewhat artificial: firstly, the examples have been invented, the authors use a written-type sentence without context and for demonstration purposes add systematically a marker of repair (I mean).

Inversely, M. Avanzi suggests that final detachments and afterthoughts share a certain prosodical feature such as using a mechanism of prosodical copy of a preceding segment in the detached constituent (Avanzi 2009: 65–66).

K. Lambrecht (2001: 1076) uses similar arguments in favour of a clear distinction between these constituents, adding another point: the frequent and conventionalized use of dislocated pronouns which shows that these cases cannot be considered as repairs (afterthoughts).

As this distinction has been evoked in numerous works and by authors of different backgrounds, it seems to be based on a phenomenon which may be difficult to describe, but needs still to be treated in more detail in order to take into account the different cases of elements detached at the end of the clause. In this sense, the present analysis is a first attempt to approach these problematics in spoken Estonian.

Geluykens (1987a), who analyses a corpus of English examples, considers most final detachment occurrences (Tails in his terminology) as repairs, but at the same time gives some examples of cases which cannot be interpreted as repairs and are assumed to be functionally different from the latter ones, because of the fact that the referent seems to be clear without any repair and the use of the detached element seems to be therefore planned by the speaker. It should also be noted that he underlines the fact that there is no pause between the two constituents in some instances, as in the following example:


A: (…) she’s been talking about leaving and going a train . teacher’s training college –

B: but she’s brilliant that girl

The next example from Geluykens (ibidem: 124) illustrates another case of detached construction with no possible interpretation as a repair, since this construction is clearly planned as such and is formulated within a special evaluative frame. These types of occurrences probably form a separate pattern.

158 (90)

He did it to me again, the bastard.

Apothéloz and Grobet 2005 propose several arguments in order to support the distinction between final detachments (appendixes in their terminology) and repairs: they consider that appendixes are part of the grammar of the speaker, thus planned elements in the discourse, and use two distinct terms for these phenomena – construction and configuration (Apothéloz & Grobet 2005: 108).

They recognize however that it is not always easy to assign a clear status to this type of utterance – for example, pauses that have been in some other works mentioned as a distinctive criterion are not reliable, because the occurrence and the perception of pauses can receive very different interpretations. According to these authors, final detachments are programmed in advance (thus are not repairs) and they do not form an intonationally independent unit.

Horlacher 2012 suggests a more cautious approach, by showing that rather these constructions form a continuum and there are no reliable criteria permitting a clear distinction to be drawn between them. She analyses a corpus of French examples, concluding that these constructions are to be examined as flexible, potential patterns in a constantly changing syntactic environment. She uses the term ‘increment’ for all these types of structures. According to her, and based on the analysed examples, there is no single criterion such as prosody or syntax which could help to distinguish between final detachments (right detachments in her study) and afterthoughts, and that it is not necessary. I adopted a similar approach in the present work, based on the observation that the examples seem to form a continuum as it is difficult to determine clear-cut criteria for distinguishing between the two; however, in the final part of the next section, where the examples from corpus will be analysed, those cases where repair mechanisms are clearly identifiable (abandoning of an initiated structure, clear self-repairs etc) will be discussed within a separate group.

Im Dokument MARRI AMON Initial and final detachments in spoken Estonian: a study in the framework of Information Structuring (Seite 152-158)