Interpreting Bare Nouns: Type-Shifting vs. Silent Heads
SALT 21
Yu Izumi, Department of Philosophy, University of Maryland
[email protected]
1. Theses
3. Argument Against Silent Heads
3.2. If null arguments (NAs) are silent pronouns
(1) Japanese bare NPs (JBNs) do not contain silent determiners.
• NAs are ellipses (Takahashi, 2008) or empty pronouns (Hoji, 1998).
• How could [NA ∅ ] in (8) have ∃ reading?
(2) JBN interpretations are the result of semantic rules that add content (and shift type).
• Either way, the silent heads ("more structure") approach fails.
☺
Correct view:
(e.g., Chierchia, 1998)
☹
dog:!β"
Reductio Argument
barked
dog:!α"
Semantic Rule
1. NA is a pro that denotes a property and gets existentially closed. e.g., [∃ pro
] ➤ Problem: Collapses into 3.1. or ∃ is a semantic type-shifting rule
3.1. Incompatible with the grammar of NP deletion
Ellipsis
Is NA anaphora in (8) and (9) elliptical or pronominal?
2. NA is a context-sensitive existentially quantified pronoun. e.g., λP∈D . ∃x[Cx & Px] (C denotes a contextually provided property)
3.2. Difficulty in explaining ∃ reading of [NA ∅ ]
Pronoun
Incorrect view:
➤ Problem: Predicts intermediate scope readings (overgeneration)
DP:!β"
(e.g., Cheng and Sybesma, 1999)
ø: !α, β"
3.1. If null arguments (NAs) are elliptical
barked
dog:!α"
More Structure
• The most promising analysis under the silent heads approach would be that NA anaphora is the projection of a silent determiner whose constituent NP gets deleted (or unpronounced).
DP (3) Any standard analysis of null argument (NA) anaphora in Japanese is incompatible with the "More Silent Structure" view.
• Bare NPs (BNs) in article-less languages such as Japanese can express a variety of interpretations. (4) Inu-ga zetumetusita. dog-nom extinct.became 'Dogs became extinct.'
ø∃
NP
øι
NP
øk
Ambiguity
Demonstratives
Quantifier
Gen Quantification
b. *Hanako-wa [a-no/arera-no/subete-no kuruma] -(o) katta. Hanako-top [that/those /every car ]-(acc) bought 'Taro bought this/these/several car(s). Hanako bought that/those/ every one(s).'
Existential Quantification • Possible Reply: NP-deletion is possible with a floating quantifier (FQ).
(7) Ippikino inu -ga heya-ni haittekita. Boku-wa inu-o mita. one.cl dog -nom room-to entered I-top dog-acc saw Object Reference 'A dog entered the room. I saw the dog.'
➤ Problem: FQ phrases are subject to several semantic restrictions (Nakanishi, 2007), whereas BNs are not. • An FQ phrase that modifies a subject disallows a collective reading (11b), whereas a BN has a collective reading (12).
KEY DATA
• The interpretations of BN antecedents and NA anaphora do not have to be equivalent. (11) a. [Otokonoko san-nin]-ga kinoo booto-o tukutta. [Boy three-cl]-nom yesterday boat-acc made (8) BN: Kind NA: Existential 'Three boys built a boat yesterday' (✓distributive, ✓collective) Americajin-ga [BN cola ]-o hatumeisi, Doitujin-ga [NA ∅ ] syohisita. American-nom cola -acc invented.and, German-nom 'An American invented cola, and Germans consumed it.' (9)
consumed
BN: Existential NA: Kind Aru Portugaljinno funanori-ga [BN dodo ]-o saisyoni tabe, a.certain Prtuguese sailor-nom dodo -acc first eat, Olandajin-ga notini [NA ∅ ] zetumetusa-seta. Dutch-nom later extinct.become-forced 'A Portuguese sailor ate a dodo first, and the Dutch people later exterminated them.'
• BNs and NAs have the same range of interpretations.
b. Otokonoko-ga kinoo [FQ san-nin] booto-o tukutta. Boy -nom yesterday [FQ three-cl] boat-acc made 'Three boys built a boat yesterday' (✓distributive, ?? collective) (12) [BN Otokonoko]-ga kinoo yondaino [BN Boy ]-nom yesterday four.cl 'A/The/Some boy(s) made four boats.' (✓distributive, ✓collective)
booto-o tukutta boat-acc made
Conclusion: alleged determiners ∅∃ etc. are not FQs. No NP-deletion with ∅∃ etc.
• An overt quantificational term dareka ('someone') indeed has an intermediate reading (13b). (13) Sorezoreno sensei-wa seito no dareka -ga okorareta each teacher-top student no someone -nom scolded.passive to kita comp heard a. For each teacher x, x heard that, [for some student y], y was scolded. (Narrowest scope) b. For each teacher x, [for some student y], x heard of y that y was scolded. (Intermediate scope) c. [For some student y], for each teacher x, x heard of y that y was scolded. (Widest scope) • But [NA ∅] in (14b) has no intermediate reading analogous to (13b).
(10) a. Taro-wa [ko-no/korera-no/ikutuka-no kuruma]-o katta. Taro-top [this/these /several car ]-acc bought
Kind Reference
(5) Inu-ga hoeru no wa atarimae-da. dog-nom bark comp top obvious-be 'It's obvious that dogs bark.'
NP
DP
DP
➤ Problem: determiner-like expressions (demonstratives and quantifiers) do not allow NP-deletion as in (10b).
2. Basic Observation
(6) Inu-ga hoeta. dog-nom barked 'Dogs barked.'
• Possible Solutions 1 and 2:
No FQ: Collective
FQ: No Collective
(14) a. Sorezoreno sensei-wa seito-ga okorareta to kita. each teacher-top student-nom scolded.passive comp heard 'Each teacher heard that some student was scolded.' b. Sikasi sorezoreno oya-wa [NA ∅ ] nagurareta to kita. But each parent-top beat.passive comp heard 'But each parent heard that they were beaten.'
4. Type-Shifting Approach • Type-shifting approach has no difficulty whatever analysis of NA anaphora turns out to be correct.
4.1. If null arguments (NAs) are elliptical • Any adequate semantic analysis of overt BNs applies to NAs as well because they are semantically equivalent. (15) ... [BN dog ] ... [NA dog ] ...
4.2. If null arguments (NAs) are silent pronouns • If JBNs uniformly denote kinds (Chierchia, 1998), then we can assume that the value of a pro can be a kind. (15) a. ... [BN dog ] ... [NA pro ] ... b. ... [BN ] ... [NA ] ...
Collective: No FQ
References Cheng and Sybesma, 1999, "Bare and not-so-bare nouns and the structure of NP". Chierchia, 1998, "Reference to kinds across languages". Nakanishi, 2007, Formal Properties of Measurement Constructions. Hoji, 1998, "Null object and sloppy identity in Japanese". Takahashi, 2008, "Quantificational null objects and argument ellipsis".