---
# Discussion
Why do X&Y alternatives have priming effects, but not X alternatives?
Salience-based view: Only logically stronger alternatives are used for implicatures
-
X&Y Alts:
There is a cross and a triangle.
-
X Alts:
There is a triangle.
-
Target:
There is a cross. ↝ no triangle
--
But the nouns in
Ad hoc were chosen randomly, e.g.
-
Prime 1:
There is a cross and a triangle
-
Prime 2:
There is a heart and a square
-
Target:
There is a star ⤳ no arrow
How did these priming trials prime an implicature generated by
There is a star and an arrow?
---
# Discussion (cont.)
Context adaptation theory
- X Alts and the target are of the same form ('
There is an X'). They evoke the same context to the same degree.
- X&Y Alts are more likely to be associated with contexts where the implicature of the target matters.
Under this view, it comes as no surprise that we observed priming effects with randomly chosen nouns
But no direct implications about theory of alternatives
In Exp 3, we tested what would happen if we didn't randomly choose nouns.
---
class: center, middle, inverse
# Experiment 3
---
# Exp 3: Design
Just like Exp 2 with Alt primes but Alts (X vs. X&Y) formed with same vs different nouns (n = 48 for each condition)
DifferentNoun, X&Y
- Prime 1:
There is a cross and a triangle
- Prime 2:
There is a heart and a square
- Target:
There is a star (⤳ no arrow)
SameNoun, X&Y
- Prime 1:
There is a star and an arrow
- Prime 2:
There is a star and an arrow
- Target:
There is a star (⤳ no arrow)
---
# Exp 3 Results:
Number

---
# Exp 3 Results:
Some

---
# Exp 3 Results:
Ad hoc

---
# Discussion
There does not seem to be any additional priming effect of
Same, relative to
Different.
- Puzzling under the salience-based approach
(Bott & Chemla 2016, Rees & Bott 2018)
- Expected under our context adaptation theory: X&Y Alts (`
There is a X and a Y') give rise to different expectations about the current context.
---
class: center, middle, inverse
# Conclusions
---
# Brief summary
Our
Baselines show:
- Implicature priming is
inverse preference effects
- Presence of
spillover effects
We propose
context adaptation as the mechanism behind implicature priming
- Explains symmetric effects of
Weak and
Strong
- Better explains alternative priming for
Ad hoc
-
X&Y Alt triggers effects,
X Alt doesn't
- No difference between
Same vs.
Different nouns.
---
# Next steps
Context adaptation involves reasoning about context
What aspects of context matter?
Candidate (potentially intertwined) factors
- Question under Discussion/conversational goal
- Relevance of alternatives
- Potential speakers
- Intonation patterns
- etc.