Questions in Discourse

Assume as a problematic null-hypthesis that the following 1:1-correlation holds between accent marking in the grammar (A), focus (F), and questions (Q): the nuclear pitch accent A of some utterance U marks the focus F of U, the function of which is to indicate the current question under discussion Q that is addressed and answered by U. That would mean that for every sentence with a nuclear pitch accent both the focus and the (implicit) question it addresses are determined. There are, however a number of reasons, why weaker correlation between A, F and Q is needed.

When dealing with this side, we typically assume that we know the function of focus. We assume that it signals the QUD.

On the Focus-Accent side, this strict mapping hypothesis has been questioned on the basis of certain empirical data that appear to be problematic for it:

…

Instances of nuclear pitch accents that do not appear to relate directly to the question under discussion.

i. Kadmon & Sevi (2011) discuss a number of examples in which the accented associate of an exclusive focus particle does not seem to relate to the explicit question under discussion.

(1) A: What's peculiar about Granny's dog? (Kadmon & Sevi 2011, ex. (8))

  B: She only likes JOHN
          H*    H*   H*L L%

If we assume with Beaver & Clark that the associate of the exclusive particle 'only' must be in focus, the focus of the utterance should be the direct object 'John', but the direct object 'John' in (1B) is not the direct answer to the explicit question under discussion in (1A). In order to save the strict mapping hypothesis one would have to assume that (1B) answers some implicit question to be accomodated in the discourse structure.

For discussion: first discussion

Another example illustrating the same problem is:

(2) A: Larry danced with Mary. (Kadmon & Sevi 2001, ex. (9))

  B: Yes, the problem is that ONLY Larry danced with Mary.
                               H* L                   L%

ii. A second well-known class of examples that have been discussed in the literature from Bolinger (1972) to Beaver & Velleman (2011) are sentences in which the position of the nuclear pitch accent does not agree with the question under discussion:

(3) A: What happened?

  B: ROBBERS have stolen from me. (Bolinger 1972)

In (3), the expected position for the nuclear accent given (3A) would be the lexical verb, giving rise to the question of whcih additional discourse factors give rise to the deaccenting of the verb in (3B). A possibility that has been proposed is the following (Kadmon & Sevi 2011, Beaver & Velleman 2011): The VP-part of the all-new sentence following the subject 'ROBBERS' is de-accented because it is weakly given, or activated, in the sense that the content of 'ROBBERS' activates the content of 'stolen from me'. If so, we would expect the accent pattern to switch back to the default pattern with accent on the indirect object in non-collocational context, in which the verb and/or the (in)direct object are less expected/activated:

(3') A: What happened?

   B1: Robbers have given me a PRESENT\.
   B2: #ROBBERS have given me a present.
   

(3'') A: What happened?

   B1: Robbers have stolen from their BEST\ friends.
   B2: #ROBBERS have stolen from their best friends.

iii. A third case of a nuclear pitch accent that appears to be only losely related to the QUD is presented by focus accents embedded inside a focus phrase (Krifka 2006). In (4Aab), the placement of the pitch accent inside the relative clause correctly indicates that the complex DP containing it (= the focus phrase) is the answer to the QUD in (4Q). However, the variable placement of the pitch accent inside the relative clause suggests that accenting has other functions over and beyond indicating the QUD. In the example at hand, it seems that the position of accent helps in constraining the restriction of the exclusive focus particle 'only':

(4) Q: Whom do you like?

  A: a. I only like [the woman that PETER introduced to Bill].
     b. I only like [the woman that Peter introduced to BILL].

Instances of question-indicating focus that are not marked by accent:

Perhaps the best-studied cases are instances of so-called second occurrence focus (SOF) (Fery & Ishihara 2010, Rooth 2010, Büring 2008). While it has been convincingly argued that run-off-the mill SOFs are still marked by other prosodic means (Fery & Ishihara, Beaver et al), the question remains as to how to integrate SOF-sentences into the discourse structure:

(5) A: John only bought a chocolate bar.

   B: Even BILL only bought a chocolate bar.

The problem for modelling consists in the fact that the exclusive particle 'only' seems to form part of the (implicit) question under discussion addressed by (4B), but at the same time, the accent on BILL seems to indicate the presenece of a different question under discussion. Hence, we should ask, how the two questions needed to analyse (5B) relate to each other: the question triggered by 'only' is (5'a), and the question triggered by 'even' is (5'b):

(5') a. What did Bill buy? b. Who else only bought a chocolate bar?

The following example illustrates the same point with an exclusive occurring in an explicit question:

(6) A: Who only likes bittersweet? (Kadmon & Sevi 2011)

  B: MARY only likes bittersweet.

Here the associate of 'only' triggers (6'a), and the focus-accent on 'Mary' triggers (6'b). Again, how do the two questions relate to each other? Are they both available in the discourse?

(6'a) What does Mary/Who like? (6'b) Who likes only bittersweet?

On this side, we assume that the focus marking the question under dicussion is already solved.

On the focus-question side,

it has been shown/discussed that the function of focus might not consist in determining the question under discussion directly, but rather in imposing constraints on the QUD. For example consider the Domain of questions:

(6) Who does John love, Mary or Ann? (7) Who does John loves, Mary, Ann or Jane? (8) John loves MARY.

Here the same focus in the answer (8) is compatible with two different questions.

Alternative analyses: i. Give up on the notion of focus altogether and account for the empirical data by means of some notion of givenness (Schwarzschild 1999, Kadmon & Sevi 2011) ii. Assume that two IS-distinctions of focus-background and given-new interact in bringing about the observable accent patterns (Féry & Ishihara 2010, Katz & Selkirk 2011)

Another indeterminacy is the focus projection problem. A focus on the direct object typically does only impose very slight constraints on the QUD.

Finally, the bigger questions to be addressed are:

• It seems that neither relation: F-A and F-Q is fixed, but that their solution depends on each other.
  See difference between the view of Rooth (who doesn't assume questions but only alternative sets) and Beaver (who explicitly assumes questions) and Büring (who doesn't assume either).
• Do we need the notion of focus at all, or are the empirical data better captured by some notion of givenness and/or a QUD-accent rule.
• Do we need more than one factor (e.g. focus+givenness) in order to predict the mapping between accenting and IS in intonation languages.
• What to do with the relation between focus and focus marking in other languages that do not rely on intonation for marking focus cf. (Zimmermann & Onea 2011).

• Detailed investigations and analyses of some of the empirical puzzles above: sentences with focus particles and ill-matched explicit QUDs; sentences with unexpected focus accenting (Robbers-sentences)…
• Presentations on the relation between focus, its structural realization, and the question under discussion in non-intonation languages (Chadic, DGS), or in languages that use both intonation and word order (e.g. Czech): (i.) Are there parallel mismatches to those found in German/English between the structural realization of focus and the QUD? (ii.) Are there instances of - what appears to be - focus realization without focus, or focus without a particular structural realization?
• The nature of verum focus: Does the phenomenon of verum focus in intonation languages really involve focusing of a covert verum operator in the sentential periphery (Höhle 1992, Romero & Han 2004), or does the deaccenting of the core vP reflect the fact that the entire proposition is given because it has been introduced into the preceding discourse (Hole & Zimmermann 2008).
• Additional focus accents in wh-questions: What combinations are licit? What higher question strategies are indicated by additional focus accents?
• Focus Combinations: Which combinations of focus accents are licit, which ones are not? E.g., what would be suitable contexts for the following sentences:

(1) MARY\ KISSED\ Peter. (???QUD: Who did something do Peter, and what did she do to him?) (2) Mary DID\ invite PETER\.

• A systematic Comparison of different focus models: focus = alternatives; focus = QUD; advantages and disadvantages.
• What additional assumption are required for maintaining the hypothesis of a tight A - F - Q-correlation in light of the mismatches in B? Accomodation of iplicit QUDs etc.?
• What would be decisive data in order to decide between a one-factor (focus or givenness) and a two-factor (focus and givenness) model?
• How convincing/problematic is the notion of p-givenness of Kadmon & Sevi? Can it really account for the entire range of focus phenomena?

• Talk 1 -
• Talk 2 -

Always upload file when editing this.

1. Roberts et al 2010 - SALT paper
2. Beaver 2012 - Meeting 1 paper