svolvaer 2012, movement - college of arts and sciences 2012... · interpretation (correspondence...
TRANSCRIPT
Simpler Syntax
Simpler Syntax on movement and structure
• Simpler Syntax lacks movement • Argument alternations in Simpler Syntax • A´chains in Simpler Syntax • Structure in Simpler Syntax
2
Simpler Syntax
Simpler Syntax on movement and structure
• Simpler Syntax lacks movement • Argument alternations in Simpler Syntax • A´chains in Simpler Syntax • Structure in Simpler Syntax
3
Simpler Syntax
Simpler Syntax lacks movement • Simpler Syntax rejects syntactic uniformity as
unnecessary (and sometimes wrong). • Syntactic uniformity motivates movement. • Thus, Simpler Syntax does not assume movement for
methodological reasons. • In the absence of arguments requiring movement,
argument alternations (such as active/passive) and A′ (filler gap) constructions are accounted for by direct interpretation (correspondence rules).
• As in all non-transformational theories. 4
Simpler Syntax
Simpler Syntax lacks movement • If there is no movement, there is no head
movement. • If there is no head movement, there can be no VP-
shells (with their attendant empirical difficulties). • In turn there can be no syntax-based derivational
morphology (with its attendant empirical difficulties; cf. Hale-Keyser), and no Pollock-style functional heads.
5
Simpler Syntax
Simpler Syntax lacks movement • The absence of head movement and invisible heads makes it
impossible to implement strictly binary branching and agreement through movement (Kayne, Larson). So:
give [NP Mary] [NP the money must be multiple branching, rather than derived from a complex
binary branching structure with invisible heads, e.g.
6
Simpler Syntax
Simpler Syntax lacks movement No head movement ⇒ No strictly binary branching ⇒ No agreement through movement and feature
checking
7
Simpler Syntax
Simpler Syntax lacks movement No movement ⇒ Many fewer constraints ⇒ No need for derivational economy as in the Minimalist Program
(but there is a need for a theory of complexity – see Culicover, Peter W. (in press). Grammar and Complexity. Oxford University Press, Oxford)
8
Simpler Syntax
Explanation without movement • The architecture of SYNTAX-CS
correspondences in Simpler Syntax permits: – Argument alternations (A movement in MGG) – Filler-gap dependencies (A´ movement in MGG)
9
Simpler Syntax
Correspondences and grammatical functions
Simpler Syntax assumes that a grammar consists of
• structural principles governing phrase structure, morphology and agreement;
• conceptual structure; • phonological structure; • correspondences rules that state correspondences between the three
representations (syntax, phonology, conceptual structure); • primitive grammatical functions (GFs) such as Subject and Object that
mediate correspondences between parts of conceptual structure and syntactic constituents;
• a lexicon that catalogues the unpredictable interfaces between phonological, syntactic, and semantic structures for particular words and phrases;
• functional principles that govern the alignment of syntactic structure with prosody and information structure.
10
Simpler Syntax
Correspondences and grammatical functions
• The focus is on describing and explaining the correspondences between syntactic structure and conceptual structure, linked through the grammatical functions (GFs).
• The GFs are primitives of the theory (as in Relational Grammar, HPSG, and LFG).
11
Simpler Syntax
Correspondences and grammatical functions
• A correspondence has basically the following form. (Phonetic form and information structure left out for simplicity.)
12
Simpler Syntax
• CS interpretations are not always one-to-one correspondences with syntactic forms – here are two particularly nice examples.
14
Simpler Syntax
Simpler Syntax on movement and structure
• Simpler Syntax lacks movement • Argument alternations in Simpler Syntax • A´chains in Simpler Syntax • Structure in Simpler Syntax
19
Simpler Syntax
Argument alternations in Simpler Syntax
• Some thematic roles link to GFs either by default
• or as a consequence of lexical specification. enter
20
Simpler Syntax
Argument alternations in Simpler Syntax
• Others link to adjuncts (e.g. oblique arguments such as PPs), perhaps with some lexical specification (e.g. for NP in the case of look for).
21
Simpler Syntax
Argument alternations in Simpler Syntax
• The GFs link to particular syntactic configurations, while the corresponding value of the associated CS argument supplies the form, as represented in the lexical entry.
• More complex possibilities exist, but this simple one will be sufficient for our present purposes.
23
Simpler Syntax
Alternative argument correspondences
• Passive • Applicative • Anti-passive • Movement is not the answer (-- the question is,
how do we account for argument alternations?)
24
Simpler Syntax
Passive • Here is a typical active/passive pair in English.
a. Kim kissed Sandy. b. Sandy was kissed (by Kim).
• The passive construction is characterized by the fact that the NP that would normally correspond to the Object of the verb in the canonical correspondence becomes the Subject, while the NP that would become the Subject of the verb in the canonical correspondence becomes an oblique argument, if it is expressed.
• The NPs Kim and Sandy have the same thematic roles in the two sentences; only the syntax is different.
25
Simpler Syntax
get passive • It is possible to state the get passive in the
form of a movement • but there is no reason to do that, given that the
various verbs take different complement structures.
30
Simpler Syntax
get passive • In addition, the get passive is not synonymous with
the active. (1) Otto was deliberately arrested ~ X deliberately
arrested Otto. (2) Otto got deliberately arrested ≠ X deliberately
arrested Otto. [If anyone acted with purpose, it was Otto, not X.]
• deliberately forces an Agent interpretation on the subject of get.
31
Simpler Syntax
Conclusion • A movement analysis of the passive does not
capture all of the argument realization possibilities.
• The argument realization possibilities are captured in terms of the mapping from CS to syntax.
• The active/passive alternation is only one of several possibilities.
33
Simpler Syntax
Applicative • The movement analysis of the passive and the
correspondence rule may appear to be almost notational variants.
• But there are other constructions that involve alternative correspondence rules that do not lend themselves to a movement analysis.
• One is the applicative.
34
Simpler Syntax
Conclusion • A movement analysis of the applicative could
be made to work (but it is not pretty). • Again, the argument realization possibilities
can be captured directly in terms of the mapping from CS to syntax.
39
Simpler Syntax
Dative is not applicative
41
• No overt morphological mark • Not a systematic alternation (unlike the applicative),
but a lexical one:
Simpler Syntax
Conclusion • There appears to be no natural way to derive
the anti-passive by movement, yet it has a natural analysis in terms of correspondences.
45
Simpler Syntax
Movement is not the answer • Movement does not appear to be the right way to
account for applicatives and antipassives. • In fact it is not entirely adequate for passives, either. • Challenges to the derivational approach to passive
come from languages in which the motivation for a movement analysis is weak or non-existent.
• If it turns out that we have to analyze such passives as direct correspondences with CS, then such an approach is also more plausible for the English passive.
46
Simpler Syntax
Non-movement passives Spanish a. Mis amigos comieron la torta.
my friend-PL eat-past-3PL the cake ‘My friends ate the cake’.
b. Se comió la torta (*por mis amigos). SE eat-past-3SG the cake by my friends
‘The cake was eaten.’
49
Simpler Syntax
Non-movement passives Italian
a. In Italia tutti mangiano spaghetti. in Italy everyone eat-PRES-3PL spaghetti-PL
‘In Italy everyone eats spaghetti.’ b . In Italia si mangia spaghetti
(*per tutti). in Italy SI eat-PRES-3SG spaghetti-PL (by everyone) ‘In Italy spaghetti is eaten (*by everyone).’
50
Simpler Syntax
Non-movement passives • Note that in (b) the singular verb does not
agree with the plural spaghetti, showing that spaghetti is the object and not the subject.
• The important point is that these constructions are other ways of suppressing the highest CS argument without syntactic movement and without passive morphology.
51
Simpler Syntax
German impersonal passive German
a. Es wurde getanzt. it be(come).past.3.sg dance-past.part. ‘There was dancing.’ b. Es wurde viel gelacht. it be(come).past.3.sg much laugh-past.part. ‘There was a lot of laughing.’
53
Simpler Syntax
German impersonal passive • This construction follows directly if we
assume – no movement in the passive – Object is not mapped to syntax (in this case there
is no CS argument mapped to Object) – obligatory filled Subject of TENSE-d S in German
(Dutch, etc.) • if there is no XP in this position, then it is filled by
dummy es. 55
Simpler Syntax
Simpler Syntax on movement and structure
• Simpler Syntax lacks movement • Argument alternations in Simpler Syntax • A´chains in Simpler Syntax • Structure in Simpler Syntax
56
Simpler Syntax
A´ Chains in Simpler Syntax
• What is a chain? • Correspondences and chains • Non-gap chains • Constraints
57
Simpler Syntax
derivational perspective A′ (Ā or “A bar”) movement produces an A′ chain — A′ dependency. • The constituent in A′ position is the head of the
chain while the gap is the tail. • Such constructions are also referred to as filler-gap
constructions.
58
Simpler Syntax
A´-chain (in MGG) A´-chain: Two constituents X and Y form an
A´-chain when • X and Y are coindexed and • X c-commands Y and • Y is a gap.
59
Simpler Syntax
Correspondence for wh-questions (getting chains without indexing
(i) the wh-phrase appears in clause-initial position, (ii) there is a gap in the position that defines the grammatical function of the phrase containing wh-, and (iii) the sentence has an interrogative interpretation.
Interrogative operator in CS: Q for the yes-no question – Q(P) Qα for the wh-question about some argument variable α contained in P – Qα(λxP(…,x,…))(α)
60
Simpler Syntax
Yes-no question • did Sandy buy the pizza • will then have the CS-representation in (b),
a. Sandy bought the pizza. CS: BUY(AGENT:SANDY,THEME:PIZZA)
b. Did Sandy buy the pizza? CS: Q(BUY(AGENT:SANDY,THEME:PIZZA))
61
Simpler Syntax
Wh-question • and what did Sandy buy • will have the representation in (b).
a. Sandy bought the cake CS: BUY(AGENT:SANDY:THEME:CAKE)
b. What did Sandy buy? CS: Qα[λx.BUY(AGENT:SANDY:THEME:x)](α) ⇒ Qα[λx.BUY(AGENT:SANDY:THEME:α)]
62
Simpler Syntax
buy: BUY(AGENT:X, THEME:Y)
LINKING: Agent ↔ Subject Theme ↔ Object
• The subject and object correspondence rules link Sandy and [e] to Subject and Object respectively, in this example.
• These are linked to AGENT and THEME, respectively, by the general linking rule and the lexical entry for buy.
• Finally, the wh-phrase in initial position is linked to the Q operator by the correspondence rule WH-QUESTION.
65
Simpler Syntax
• The phrase structure rules of English are responsible for the syntactic structure,
• while the rules of CS are responsible for where Qα, α, λx and BUY go with respect to one another.
• The gap corresponds to x.
67
Simpler Syntax
• The correspondence for the wh-question where the wh-phrase is in an A′ position requires a wh-phrase in clause-initial position, the operator Qα in CS, and a lambda expression applied to (an expression containing a) variable α in CS, by λ-reduction.
68
Simpler Syntax
CS-chain CS-chain: Two constituents X and Y form a CS-
chain when (a) X c-commands Y and (b) X and Y are both linked to the same element
or coindexed set of elements of a CS-representation.
69
Simpler Syntax
CS-chain • When the wh-phrase is the subject of a main
clause, it appears in clause-initial position in English.
• There is no direct evidence that the wh-subject is in an A′ position that is linked to a gap in subject position.
• The WH-QUESTION correspondence rule and the SUBJECT correspondence rule both apply to the wh-phrase.
71
Simpler Syntax
Topicalization is not movement • Topic precedes wh To Sandyj, whati are you planning on giving ti tj? • Topic follows that
a. Terry claims [that Sandyi, I like ti ]. b. *Terry claims [Sandyi, that I like ti ].
75
Simpler Syntax
Iterated CP? • This is a particularly clear example of how
MGG uses empty heads, features to trigger movement, and various stipulations that rule out overgeneration to get just the observed orders.
78
Simpler Syntax
• The Simpler Syntax solution is to explicitly specify the constructional possibilities --
where XP[TOP] is an XP linked to a topic in IS. 79
Simpler Syntax
Other CS-chains involving gaps • infinitival questions (what to do [e]) • infinitival relatives (the man to see [e]; the
man for you to see [e]; the man to save us) • parasitic gaps (a book that I criticized [e]
before reading [e]) • etc.
80
Simpler Syntax
Hanging topic left dislocation a. That cari, I really love to drive iti at high speeds. b. I really love to drive iti at high speeds, that cari.
81
Simpler Syntax
No gap, no constraints a. That cari, I met a guy [who wanted to sell iti to me for $3000]. b. *That cari, I met a guy [who wanted to sell ti to me for $3000]. c. That cari, I just received a report [(that) iti was in a major crash]. d. *That cari, I just received a report [(that) ti was in a major crash].
82
Simpler Syntax
Hanging topic • The hanging topic must appear in a specified
position in the structure. • It must be associated with a particular discourse
function, e.g., TOPIC, that is parallel to the function/argument structure.
• A θ-role is assigned to the pronoun, which is in the argument position.
• The hanging topic specifies the reference of what bears this θ-role.
• Hence the hanging topic itself is not assigned a θ-role.
83
Simpler Syntax
Dislocation without typical chain a. (As for) that cari, I really think that the windshield needs to be replaced. b. (As for) Johni, there are very few people that really like that guyi.
• There is no proform or gap, but there is a CS-chain.
88
Simpler Syntax
Conclusions • It is possible to represent the filler-gap chain without
movement. • Movement is only one way to create a gap. (See
Koster 1978, HPSG, LFG, etc.) • There are chains that don’t involve gaps. • There are filler-gap chains that don’t have fillers in
A´ position (zero-relatives, comparatives, tough). • Movement doesn’t help us with these. • ⟹ No A´ movement.
91
Simpler Syntax
Simpler Syntax on movement and structure
• Simpler Syntax lacks movement • Argument alternations in Simpler Syntax • A´chains in Simpler Syntax • Structure in Simpler Syntax
92
Simpler Syntax
Structure • In MGG, configuration is fundamental
– GFs defined in terms of configuration – All relations between constituents defined in terms
of configuration (e.g. c-command, Spec-head agreement, etc.)
– Linear order defined in terms of configuration and movement (antisymmetry [Kayne 1994])
93
Simpler Syntax
Structure • Simpler Syntax asks
– Can we simplify configuration by making GFs primitive?
– and by allowing explicit statements in the grammar about linear order (rather than having linear order follow from configuration)
– Answer: yes, yes, (yes!)
94
Simpler Syntax
Structure • Well, then
– must GFs be primitive? – must there be statements in the grammar about
linear order? – Answer: yes, yes
95
Simpler Syntax
GFs • With primitive GFs, we can state argument
alternations independently of structure. E.g., the general passive correspondence is
96
Simpler Syntax
GFs • The way that the Subject and passive form of
V are realized, and the linearization, is independently defined for the language (like a device driver).
97
Simpler Syntax
GFs • With GFs we can account for anaphor binding
without coreference • Mary behaved herself/*John. • John perjured himself/*Mary. • The lexical entries for these Vs can stipulate
that their Subject and Object GFs are coindexed, therefore the Object must be a reflexive.
98
Simpler Syntax
GFs • With GFs we can explain ‘raising’ without
movement or ‘exceptional case marking’.
99
Simpler Syntax
Simpler syntax • This allows us to maintain a very simple
formulation of English VP structure
• VP→ V (NP) ( )
• All of the details are taken care of by lexical entries for raising and control.
103
Simpler Syntax
Linear order • If we have linear order, how complex do
statements about structure need to be? • Not very complex.
104
Simpler Syntax
Linear order • First position • Second position • Last position • Precedes/follows • Adjacent
105
Simpler Syntax
Simpler Syntax Hypothesis (SSH) • The most explanatory syntactic theory is one
that assumes the minimum structure necessary to mediate between phonology and meaning.
106
Simpler Syntax
What is ‘minimum’ structure? • There are no invisible elements except those
that have to be postulated to explain the grammatical properties of sentences and their meanings.
108
Simpler Syntax
What is ‘evidence’ of structure? • ‘Evidence of structure’ would be behavior of
CD (but not BC) that shows that it is a unit distinct from B. E.g., – dislocation, e.g,
• topicalization • scrambling
109
Simpler Syntax
What is ‘evidence of structure’? – agreement, e.g.
• C & D share a morphological feature • (maybe this feature is dependent on B)
– CD can fulfill a range of grammatical functions (such as subject & object)
110
Simpler Syntax
What is ‘evidence of structure’ • On this view, the structure of the English
sentence is is essentially
111
Simpler Syntax
• with a few additional details for adverbs, variable order (scrambling?) in VP, membership in ‘...’.
• and special constructions (e.g. inversion) that require different structures.
112