Evidential Implicatures in Cuzco Quechua

1
Evidential Implicatures in Cuzco Quechua

• MPI lunch talk
• Martina Faller, MPI KUN

2
Overview

• Background on evidentiality
• Quechua evidentials and their implicatures
• Levinsons heuristics for calculating generalized
  conversational implicatures
• How to calculate evidential implicatures
• Proposal evidential heuristics
• Implications for the Theory of GCIs

3
Quechua

• Quechua is a language family spoken throughout
  the Andes by approximately 8 million people
  (Lefebvre Muysken 1988)
• Data used in Faller (2002) collected in Cusco,
  Peru. Peru has around 4 million Quechua speakers
  (Cerrón-Palomino 1987)
• Quechua is an agglutinative language

4
Evidentiality

• the encoding of the speakers grounds for making
  a speech act
• in assertions speakers type of source of
  information
• an evidential is a grammatical marker of
  evidentiality

5
Three main types of Evidentiality

• Direct evidence
• Reportative evidence
• Inferential evidence

Willett (1988)
6
Epistemic Modality

• The encoding of the speakers judgment of a
  proposition as true or false with a certain
  degree of certainty.
• English epistemic modals must, may

7
Epistemic Modality Evidentiality Overlap
Epistemic Modality Possibility
Necessity
Epistemic Modality Possibility
Necessity
Inference
Report Direct
Evidentiality
Based on Auwera and Plungian (1998)
8
Quechua evidential paradigm

• Para-sha-n
• rain-prog-3
• It is raining.

No evidential
9
Quechua evidentials as illocutionary modifiers

• Any assertion has the sincerity condition that
  the speaker believes p
• Quechua evidentials add a sincerity condition
  specifying how the speaker came to believe p

10
Quechua evidentials as illocutionary modifiers

• Para-sha-n.
• rain-prog-3
• pIt is raining.
• Sinc Bel(p)

• Para-sha-n-mi.
• rain-prog-3
• pIt is raining.
• Sinc Bel(p), Dir(p)

11
Two types of evidential implicatures

• I. Absence of evidential
• Para-sha-n
• rain-prog-3
• It is raining. gt Direct evidence

II. Presence of indirect evidential Para-sha-n-si
/-chá rain-prog-3 It is raining. gt
Direct evidence
12
Generalized Conversational Implicatures

• GCIs increase the informativeness of the
  coded/entailed content of a sentence in a
  predictable and regular way from the structure
  of utterances, given the structure of the
  language, and not by virtue of the particular
  contexts of utterance (Levinson 2000).

13
Generalized Conversational Implicatures

• In contrast to the coded meaning, GCIs are only
  preferred interpretations, which in certain
  circumstances can be cancelled or blocked.

14
The Q-Heuristic

• What is saliently not said, is not the case.
• The Q-heuristic is related to Grices first maxim
  of Quantity Make your contribution as
  informative as required (for the current purposes
  of the exchange)

15
The Q-Heuristic

• operates on paradigmatic expressions, which can
  be ordered according to degree of
  informativeness
• lt Strong, Weak gt

16
The Q-Heuristic

• Example lt all, some gt
• (a) All tigers are fierce.
• (b) Some tigers are fierce.
• (a) is more informative than (b), because (a)
  entails (b)

17
The Q-Heuristic

• What is saliently not said is not the case
• lt all, some gt
• Some tigers are fierce.
• Qgt not all tigers are fierce.

18
The I-Heuristic

• Unmarked, minimal expressions warrant
  interpretations to the stereotypical extensions.
• related to Grices second Maxim of Quantity Do
  not make your contribution more informative than
  required.

19
The I-Heuristic

• I-implicatures enrich/narrow/strengthen what is
  said
• Example
• boxer Igt male boxer

20
The M-Heuristic

• Marked message indicates marked situation.
• relates to Grices Maxim of Manner
• Be perspicous
• Example
• (a) Bill stopped the car. Igt
  normally
• (b) Bill caused the car to stop.
  Mgt not normally

21
Maxim of Quality

• Quality I Do not say what you believe to be
  false
• Quality II Do not say that for which you lack
  adequate evidence

22
Maxim of Quality

• Levinson does not propose any heuristics relating
  to the Maxim of Quality
• The maxim of quality is mainly appealed to for
  implicatures that arise from flouting it
• A Tehran's in Turkey, isn't it, teacher?
• B And London's in Armenia, I suppose.

23
Implicating Direct Evidence

• Absence of evidential
• Para-sha-n
• rain-prog-3
• It is raining. gt Direct evidence
• Is the DE-implicature a GCI?

24
Implicating Direct Evidence

• Direct evidence implicature can be overridden by
  context
• ...triciclu-n-ta-qa tari-ra-ka-pu-n.
• ...tricycle-3-acc-top find-hort-rfl-ben-3
• ...they found his tricycle.

25
Implicating Direct Evidence

• A sentence can only have a single evidential
  value.
• If evidential-less sentences encoded the value
  direct, it should be impossible to add an
  indirect evidential.
• Para-sha-n-si/-chá
• rain-prog-3
• It is raining.

26
Implicating Direct Evidence

• GCIs usually can be cancelled overtly
• Some tigers are fierce, in fact all of them
  are.
• the DE-implicature cannot be cancelled.
• Para-sha-n, ichaqa mana riku-ni-chu
• rain-prog-3 but not see-1-neg
• It is raining, but I did not see (it).
• This is not surprising, however, since the
  DE-implicature is illocutionary

27
Which heuristic is responsible for the
DE-implicature?

• Q-heuristic operates on overt linguistic
  expressions.
• The M-heuristic operates on marked expressions
  evidential-less sentences are not marked.
• Therefore, neither Q nor M can be responsible

28
Which heuristic is responsible for the
DE-implicature?

• The DE-implicature does narrow/strengthen what is
  said so, is it an I-implicature?
• But reportative or conjectural evidence would
  also be potential enrichments.

29
Which heuristic is responsible for the
DE-implicature?

• Principle of Informativeness if an utterance
  has competing interpretations, the best one is
  the most informative one (Atlas and Levinson
  1981).
• Is direct evidence more informative than
  reportative or conjectural?
• Not if informativeness is defined as entailment

30
Calculating the DE-implicature

• Direct evidence is stronger than indirect
  evidence.
• Addressee can assume that speaker bases an
  assertion on the strongest type of evidence
  available to him or her.
• If no type of evidence is overtly expressed,
  direct evidence is implicated.

31
Implicating absence of DE

• Indirect evidentials implicate the absence of
  direct evidence
• Para-sha-n-si/-chá
• rain-prog-3
• It is raining. gt Direct evidence
• The DE-implicature arises in other languages
  with evidentials (de Haan 1998)

32
Implicating absence of DE

• De Haan DE implicature is calculated on the
  basis of a universal evidential hierarchy
• Direct gt Inferential gt Reportative
• Ordering criterion speaker preference
• Is the DE implicature a Q-implicature?

33
Implicating the absence of DE

• De Haans hierarchy is not valid for Quechua.
  Instead
• Direct -mi gt Reportative -si
• Direct -mi gt Conjectural -chá
• If the DE implicature is a Q-implicature, then
  sentences with -mi should entail the same
  sentences with -si or -chá

34
Implicating the absence of DE

• Para-sha-n-mi.
• rain-prog-3
• pIt is raining.
• Sinc Bel(p), Dir(p)

• Para-sha-n-si.
• rain-prog-3
• pIt is raining.
• Sinc Bel(p), Rep(p)

• Trivially, a sentence S with -mi entails S with
  -si but also vice versa.
• Moreover, this entailment relation does not
  include the evidential value

35
Implicating the absence of DE

• Para-sha-n-mi.
• rain-prog-3
• pIt is raining.
• Sinc Bel(p), Dir(p)

• Para-sha-n-si.
• rain-prog-3
• pIt is raining.
• Sinc Bel(p), Rep(p)

• Relevant notion is illocutionary entailment
  speech act A entails B, if A cannot be performed
  without also performing B (Vanderveken 1990).

36
Implicating the absence of DE

• Para-sha-n-mi.
• rain-prog-3
• pIt is raining.
• Sinc Bel(p), Dir(p)

• Para-sha-n-si.
• rain-prog-3
• pIt is raining.
• Sinc Bel(p), Rep(p)

• The sentence with -mi does not illocutionary
  entail the sentence with -si or -chá

37
Implicating the absence of DE

• The evidential scales are not ordered in terms of
  informativeness.
• The DE-implicature is not a Q-implicature.
• Again, the revelant notion is
• strength of evidence

38
Proposal Evidential Heuristics

• Both the DE- and the DE-implicature are arise
  because direct evidence is stronger than
  reportative or conjectural evidence.
• They exploit Grices second Maxim of Quality Do
  not say that for which you lack adequate
  evidence.

39
Proposal Evidential Heuristics

• E(vidential-)S(trength)-Heuristic
• Saliently not indicated types of evidence are
  not available to the speaker
• This heuristic operates on paradigms ordered by
  degree of strength of evidence
• Direct -mi gt Reportative -si
• Direct -mi gt Conjectural -chá

40
Proposal Evidential Heuristics

• E(vidential-)S(trength)-Heuristic
• Saliently not indicated types of evidence are
  not available to the speaker
• Para-sha-n-si/-chá
• rain-prog-3
• It is raining. gt Direct evidence

41
Proposal Evidential Heuristics

• E(vidential-)E(nrichment)-Heuristic
• Unmarked, minimal expressions warrant
  interpretations to the evidentially
  richest/strongest extension.
• Para-sha-n.
• rain-prog-3
• It is raining. gt Direct evidence

42
Implications for Theory of GCIs

• The EE-implicature is not universal
• It is raining.
• does not implicate that the speaker saw it rain.

43
Implications for Theory of GCIs

• EE-implicature hypothesis
• evidential zero marking only gives rise to the
  DE-implicature in languages that encode
  evidentiality paradigmatically.

44
Implications for Theory of GCIs

• Generalized Zero-marking-implicature hypothesis
• a. Zero marking implicates a value for feature X
  just in case there is a linguistic paradigm
  encoding the values of X.
• b. Given a linguistc paradigm for X, zero
  marking implicates the super value of X.

45
Implications for Theory of GCIs

• Levinsons heuristics all operate on the
  propositional content of an utterance.
• Quantity and Manner maxims relate to information
• The evidential heuristics operate on the
  illocutionary level of an utterance
• Quality maxims relate to sincerity

46
Implications for Theory of GCIs

• A chicken and egg problem?
• Do languages have certain morphosyntactic devices
  because their speakers adhere to the related
  heuristic, or
• Do speakes adhere to certain heuristics, because
  they happen to have the morphosyntactic devices?