Sudo 2010. Contextual evidence (B&G) Importance of contextual evidence established by B&G (as we just saw) Contextual evidence for negative answer blocks.

Presentation on theme: "Sudo 2010. Contextual evidence (B&G) Importance of contextual evidence established by B&G (as we just saw) Contextual evidence for negative answer blocks."— Presentation transcript:

1 Sudo 2010

2 Contextual evidence (B&G) Importance of contextual evidence established by B&G (as we just saw) Contextual evidence for negative answer blocks the felicitousness of a PPQ PPQs fine in neutral contexts

3 More than one type of bias… (N.B. the term evidential is not to be confused with evidentials in the sense of source marking) The above already described in B&G Sudo points out an additional type of bias: epistemic

4 Epistemic bias Outer negation questions have [-positive] evidential bias, incompatible with positive evidence They are fine in neutral contexts

5 Epistemic bias Outer negation Isn´t p? Epistemic (speaker thinks p is likely) Bouletic (speaker expects p) Deontic (speaker thinks p should be the case) Sudo groups all of these (E, B, D) into a general category of bias she proposes in addition to evidential bias: epistemic bias

6 Isn’t Bill tall? Epistemic (speaker thinks it is likely that Bill is tall) Bouletic (speaker expects Bill to be tall) Deontic (speaker thinks that it should be the case that Bill is tall) p= Bill is tall

7 Epistemic bias PPQs – no epistemic bias, ONPQs have positive epistemic bias Found in context! Found in question!

8 What about inner negation? INPQs have stronger evidential bias than ONPQs Negative evidence is necessary for INPQs to be felicitous, positive evidence incompatible

9 But if there’s no evidence, INPQs are infelicitous

10 Summary ON ON – infelicitous when there is contextual evidence for the positive answer A: This neigborhood is a hippie’s paradise! B: #Isn’t there a vegetarian restaurant around here? (ON) ON is fine in a neutral context A: Where do you want to eat tonight? B: Isn’t there a vegetarian restaurant around here? (ON) Therefore, the proposition is must be old to the speaker but new to the discourse (or unactivated in the discourse)

11 Summary IN INs require contextual evidence for the negative answer A: A hippie like Tom would never survive in this neighborhood. It’s for meatlovers. B: Isn’t there a vegetarian restaurant around here? (IN) INs are infelicitous in neutral contexts A: We can eat anywhere you want! B: #Isn’t there a vegetarian restaurant around here? (IN)

12 Summary IN Propositional content is new to the discourse, new to the speaker and old to the hearer

13 Summary PQs There is a conflict between some prior belief and the evidence that has come about in the context N.B. NOT an exhaustive list of possible combinations/parameters

14 Intonation How are these ideas applicable to intonation? Armstrong 2010: 3 contexts: Context 1 (NEUTRAL) Tu amigo: Quiero llevarte a comer esta noche. ¿Adónde quieres ir? Tú: ¿Por aquí hay algún restaurante vegetariano? Ya sabes que no como carne. ¡H* L%

15 Context 2: Como Uds. son vegetarianos, no podemos comer en este barrio. Todos los restaurantes aquí son más para los que comen carne. Tú: ¿Por aquí no hay ningún restaurante vegetariano? ¡H* L%, L* HL% Context 3: Estás visitando un amigo en NYC y te acuerdas que la última vez que lo visitaste comieron en un restaurante vegetariano Tú: ¿Por aquí no hay un restaurante vegetariano? ¡H* L%, H+L* L%

16 Contours Context 1 Ev: any Ep: any ¡H* L% Context 2 Ev: +pos Ep: +neg L* HL% Context 3 Ev: ?? Ep: +pos H+L* L%

17 Looking at contour types based on features for evidence and epistemicity reveals the nature of their pragmatic division of labor Some contours (H+L* L%, L* HL%) are more restricted than others (¡H* L%) – these are the contours that are more strictly specified for either evidence or epistemicity

18 ContourEvidenceBelief (Ep) ¡H* L%Not specified H+L* L%(?)+pos L* HL%+pos+neg

19 Because of the underspecified nature of ¡H* L%, it can be used in more contexts ¡H* L% marks “questionhood” in all contexts but production results suggest that the nature of the question (biased or not) is context-induced) (when there is +neg belief, ¡H* L% may be modified phonetically – scaling differences) L* HL% and H+L* L%, are much more restricted and were only produced for specific contexts

20 Questions How can this be tested experimentally? (talk about issues with separating evidence and belief experimentally)

21 It may be helpful to consider temporal features as well Perhaps through application of theories of information structure (i.e. Prince 1981…) tutu -Belief ¡H* L% L* HL%