Syntax
Co-constructed syntax
We consider as co-constructions those syntactic relations (without any restrictions) that hold between utterances produced (i) by different speakers or (ii) by the same speaker but after an interruption by another speaker. In interactive scenarios, indeed, multiple speakers may intertwine their utterances, sometimes even perfectly overlapping (Lerner 1991, Sacks 1992, Ono & Thompson 1996, Helasvuo 2004, Calabria 2023).
This implies the fact that two dimensions need to be considered: a speaker-dependent one, which only accounts for language uttered by the speaker over time, and a time-dependent one, which accounts for the participation of multiple speakers to the same speech unit. This is a relatively new aspect for the UD formalism, as written text is usually only concerned with the former perspective. Eliminating this aspect from the syntactic analysis of spoken language would however introduce a bias in the comparison with planned or written language, as speaker-dependent vision would hide syntactic relations deriving from contributions given by different speakers in the same time frame, and at the same time result in sentence-like units that can only be interpreted (even syntactically) contextually given the contributions of other participants to the interaction. It is worth remarking that the issue is not entirely new, as for instance social media data might show the same features when it comes to sequences of posts in a thread, possibly posted by different users.
Encoding
Coconstruction are encoded with two features in the MISC columns.
-
Coconstruct: A coconstruction is encoded by a feature on the root of the second utterance indicating the syntactic role of the second utterance towards the first utterance, thesent-idof the first utterance, and thetoken-idof the governor of the second utterance:Coconstruct=<deprel>::<sent-id>::<token-id> -
Backchannel: For backchannels (see below), in the speaker-based view, the backchannel has a featureBackchannel=<sent-id>::<token-id>
Additional instruction for different cases of coconstructions will be provided as follows.
Typology of coconstructions
We distinguish three major types of co-constructions: co-construction proper, wh-question constructions, and backchannelling.
1. Coconstruction proper
The key distinction within coconstructions proper concerns whether Speaker 2 realizes a dependency that is already projected by Speaker 1 (case A), or whether Speaker 2 adds a new dependency to the structure initiated by Speaker 1 (case B).
A. Speaker 2 realizes an already projected dependency. When Speaker 2 realizes a dependency projected by Speaker 1, we distinguish three subcases reflecting the status of that dependency at the moment of realization: unrealized, in progress, or already realized.
A.1 Completion: Speaker 1 leaves a syntactic dependency unrealized, and Speaker 2 provides the missing material, completing the structure initiated by Speaker 1, as in Example (1).
(1)
Cuz 31 (Ono & Thompson 1996: 72)
L: his position is pretty uh
A: … % (TSK) (H) stable.
… yeah
A.2. Parallel realisation: the same syntactic dependency is realized by both speakers. Speaker 1 projects and eventually completes the structure, often after hesitation or a false start, while Speaker 2 simultaneously or near-simultaneously fills the same syntactic slot
This case frequently results in overlapping between Speaker A and Speaker B realizing the same slot, and as in (2).
(2)
BOD2018 - KIParla
BO118: c’è architettu:ra::, c’ è:::: [ingegneri~]
there’s architecture there’s engineer-
BO140: [ingegn~ ingegn]eria: sì
enginee- engeenering yes
BO118: si che è [molto buona] anche
yes which is very good too
A.3. Subsequent realisation: Speaker 1 completes their own tree, and Speaker 2 subsequently selects a dependency in Speaker-1’s structure and realizes it again, either by repeating or replacing material in the same syntactic slot, as in (3) and (4).
(3)
Rhapsodie - D2001
$L2 “eh bien” je crois que je ne me suis pas
well I think I haven’t
$L2 conduit d’une façon conforme à ce qu’on
behaved as one would
$L2 attend “euh” { { d’une jeune fille d’abord
expect from a young girl first
$L2 | ˆ et d’une femme ensuite } |} //+
and from a woman afterwards
$L1 {| d’une jeune bourgeoise |} ?//+
from a young bourgeois girl?
$L2 {| “disons” d’une jeune bourgeoise } //
let’s say from a young bourgeois girl
(4)
KPS021 - KIParla
PKP126: quindi l~ la linea tra finzione e realtà
so the line between fiction and reality
cioè tra verità non verità non ho ancora capito [dove sta]
I mean between truth and non-truth I still haven’t understood where it lies
PKP125: [più tra de]tto e non detto [x]
more between said and unsaid
B. Speaker B adds a new dependency: Speaker 2 realizes a syntactic dependency that is not already projected by Speaker 1, thereby extending the syntactic structure initiated by them.
(5)
Ono and Thompson (1996: 81)
M: they must know each other.
H: . . . very well.
(6)
KIParla - BOA3017
BO139: quando si parlano sopra
when they speak one over the other
devi mettere delle parentesi quadre
you have to put square parentheses
intorno alle parole
around words
BO146: ah okay
oh okay
BO139: che si sovrappongono
that overlap
BO136: come nei sottotitoli
like in subtitles
In cases of overlaps, the guiding criterion is whether it is possible to apply the ‘if you can link, link’ principle: if a tree can be linked to a previous tree uttered by a different speaker or by the same speaker but not consecutively, then use the coconstruct relation; if a tree can be linked to a previous tree uttered by the same speaker consecutively, then merge the trees. If the intervention by Speaker 2 is fully overlapping, that is Speaker 2 does NOT interrupt Speaker 1, then merge and move Speaker-2’s unit after.
Annotation of cases of completion (A1) and additional dependency (B)
When a projected dependency remains unrealized in Speaker-1’s utterance (cases A1 and B), promote the unfinished element to fill the syntactic position that would normally be occupied by its missing complement. Then, annotate the element or subtree whose syntactic projection is unifinished with two additional MISC features:
Scrap=YesPromotion=⟨deprel⟩, specifying the relation that was not realised.
The missing element may be the head of a dependecy relation, as in example (7):
(7)
Speaker 1: she has a very nice
Speaker 2: attitude
In cases like this:
- promote nice as the
objof has and give it the feature Promotion=amod; - annotate a as
detof nice with a feature Head=Position, and very asadvmodof nice with a feature Head=Word.
Annotation of parallel (A.2) and subsequent realisations (A.3) - stacking
Stacking can be annotate with different deprels.
- For repetitions or repair use
repairas the deprel label, it is then transformed intoreparandumin dependency view; - For reformulations (when the first conjunct is complete and the
second conjunct is different) use the newly introduced relation:
conj:reform.
Reformulations may be distinguished from coordinated conjuncts because, in reformulation the conjuncts are different denotations of the same referents, contrary to coordinated conjunts.
2. Wh-questions
Consider answers to wh-questions (ex. 8) as con-constructions, since they are syntactically dependent on the question, and thus part of the same rectional unit. Answers to polar questions should not be treated coconstructions, since their response items do not depend syntactically on any element of the question nor fill a pre-existing syntactic slot.
(8)
KIParla - PTA007
TOR001: dove vai ad arrampicare?
where do you climb?
TOI007: al bi side4 vicino alla colletta
at the bi side close to colletta
The head of the coconstruction is the interrogative element (pronoun, adverb etc..)
3. Backchanneling
Backchannels (BC) are short productions uttered by one participant in the conversation when the other participants occupy the floor (back-channel vs front-channel). They may be verbal (yes) or paraverbal (mhmh) and mainly serve to signal attention or alignment. At their core, these verbal reactions show that the speaker has heard the contribution of the partner, often adding that it has been understood and accepted (see Mereu et al. 2024 and Ward and Tsukahara 2000).
In order to be recognised as BC, a specific utterance must:
- be addressed to the content of the utterance produced by the other speaker
- not be required or expected based on previous turn (e.g. answers to questions are expected and required, so they cannot be considered BC, see ex. 10)
- not require a reaction from the other speaker (see ex. 9, where the speaker continues)
(9)
KIParla - BOA3017
BO145:[ma] per[ché mamma c’ha dei <pre]giudizi nei miei confronti> da quando (sono) nata
but because mum has had some prejudices towards me since I was born
p[enso]
I think
BO139: [mhmh]
BO145: e poi daniela non devi avere pregiudizi su di me io
(10)
KIParla - BOD2018
BO118: [sì] sì ma anch’io:: però era proprio l’esigenza [de sta da sola in dei momenti]
yes yes me too I also have the need to be alone sometimes
BO140: [sì di stare da so:la di non] parla~
yes to be alone not to talk
In dependency view, tokens that form BCs should be identifiable by an attribute in the MISC column (Backchannel=Yes).
In dependency-based view, tokens are linearly placed based on time alignment to their closest token in time, when such information is available. Syntactically, backchannelling enters the syntactic flow and is linked by a discourse relation.
discourse:backchannel is introduced as a new dependency sub-label, but should only be used in the cross-speaker case (i.e. when the BC has a different speaker ID to its syntactic parent).
In case of BC, we consider that the main speaker keeps the speech turn, its utterance is not interrupted and has not to be segmented. The BC will be placed in a separate sentence and attached to its triggering element (where possible, otherwise, the root of the utterance)
TO-DISCUSS: can the backchannel be used to take the floor? Where is the boundary between proper feedback and backchannel?