Syntax notes part 2 for Midterm.doc

42 Pages
123 Views
Unlock Document

Department
Linguistics
Course
LIN232H1
Professor
Alexandra Motut
Semester
Summer

Description
Binding Theory (Ch.5) • Referring Expressions (R-expressions) • Anaphors • Pronominals/Pronouns • Binding Principles A, B, C We will see one very important case where one particular structural relation is relevant, in that it helps us model where certain kinds of NPs can grammatically occur. Meaning/Interpretation & Structure • Binding Principles are another place where interpretation is affected by structure • There are consequences for meaning depending on structure. – Another example of this we’ve already seen is structural ambiguity. Language and Reference • Different kinds of reference o Four dogs, a cat, the party, my unicorn, Harry Potter o Deictic: Yesterday, there, she, now, our, themselves, this Reference  to the world;  reference to linguistic components  inside or  outside a sentence/discourse NPs and reference When an N grows up… • Syntax: Become an NP • Semantics: Make reference to some entity NPs and reference Not all nouns/NPs refer the same way • Independent • Dependent 1 Kinds of NPs… Binding Theory describes the conditions on the structural relations between NPs with respect to reference. Concerned with three types of NPs: • R-expressions (proper names, common nouns) • Pronouns (he, she, it, his, one, them, him, etc) • Anaphors (eg. himself, herself, themselves) These NPs are semantically distinct, but they also have different syntactic distributions. Referring Expressions • Vast majority of NPs in language are referring expressions (R-expressions): Ex. – Obama passed a health care reform bill. – Mayor Ford shouldn’t run in the next election. – A new restaurant is opening on Yonge Street. • Referring expressions get their meaning by referring to objects/entities in the world (real world or imagined world). Anaphors • Languages also have NPs that get their meaning from another word in the sentence: Ex. – The paramedic injured herself. – John saw himself in the photo. 2 • NPs like herself, himself, itself, myself, yourself, themselves, and each other are called anaphors. • They HAVE TO get their meaning from another NP in the sentence. Pronouns/Pronominals • Yet another type of NP; pronouns/pronominals. Ex. Kate said that she plays water polo. – Interpretation 1: she refers to Kate. – Interpretation 2: she refers to Alex (or Sarah, or Tina, etc., etc.) • Pronouns can either get their meaning from another noun/NP in the sentence, or can get their meaning from a noun previously mentioned in the discourse, or by context (some other entity. Eg. I, you, he, she, it, we, they, me, him, her, us, them, my, your, his, her, our, their, one GPS1: • Their  pronoun • Each cat  R-expression • Folk dancing  R-expression • Oneself  anaphor • Each other  anaphor • She  pronoun • Her  pronoun • Themselves  anaphor Restrictions on Use & Meaning • There appear to be restrictions on the use and meaning of anaphors and pronouns: – *Himself injured Bill. – *Herself saw Maria. 3 – He saw John. • He ≠ John. – John saw him. • John ≠ him. • We need to explain these restrictions. Binding Theory Terminology In order to describe the structural conditions (i.e., the distribution) of NPs, we need some notions – Antecedent – Indexing/Indices • Co-indexing • Disjoint Reference – Local domain/Binding Domain (we’ll get to this later.) Indices • To help us keep track of what anaphors and pronouns refer to, we’re going to use indices (subscript letters, starting at i): – Maria iaw herself. i – Maria iaw her. j • NPs with the same index refer to the same entity, while NPs with different indices refer to different entities. • NPs with the same index are called coindexed, and the two entities corefer. (1.) Mary iealized that John haj always loved her . i (Her = Mary) (2.) Mary iealized that Johnj had always loved her . k (Her ≠ Mary) (1): Mary and her are coindexed. Mary and her corefer (refer to the same entity in the world). (2): Mary and her have different indices. Mary and her have disjoint reference (refer to different entities in the world.) Antecedent Antecedent: An NP that gives its meaning to a pronoun or anaphor. Alex told herself that the field wasn’t far. herself = anaphor Kate said that she played water-polo. she = pronoun/pronominal Kate and Alex are both Antecedents Anaphors: Data Set 1 • We’ve already said that anaphors want something else in the sentence to refer to (an antecedent): 4 Ex. – John siw himself. i – *John saw himself. i j • But: – *Himself sai John. i • Hypothesis #1: Perhaps anaphors need an antecedent earlier in the sentence. – Here, himself precedes John. Anaphors: Data Set 2 • Hypothesis #1: Perhaps anaphors need an antecedent earlier in the sentence? • But: – *John siid that Bill sawjhimself. i – John siid that Bill sawjhimself. j • Notice here that the anaphor is in an embedded clause [CP that Bill saw himself]. • Hypothesis #2: Perhaps anaphors need an antecedent earlier in the same clause? Anaphors: Data Set 3 • Hypothesis #2: Perhaps anaphors need an antecedent earlier in the same clause? • But: – *[John’sjfather] sawihimself. j – [John’sjfather] sawihimself. i • ‘John’ and ‘himself’ are in the same clause but ‘himself’ can’t refer to ‘John’. • Hypothesis #3: Perhaps anaphors need their antecedent to be the first preceding NP in the same clause? Aside on Bracke ting of [John’s father] Why like this? … And not like this? [Johnj’s father]i Johnj’s fatheri The index i is indexing the constituent [John’s father], not just the word ‘father.’ Anaphors: Data Set 4 • Hypothesis #3: Perhaps anaphors need their antecedent to be the first preceding NP in the same clause? • But: – John sint [a letter] to jhimself]. i – *[The father of [that boy]] injuj i himself. j – [The father of [that boy]] injuj i himself. i 5 • Hypothesis #4: Perhaps anaphors need an antecedent in the same clause that is in a particular structural relationship with the anaphor? • Notice we’ve gone from a hypothesis about linear order (precedence) to a hypothesis about internal structure/structural relations. Structural Relati onship: C--‐command • The hypothesis we will adopt for anaphors: (Informally): Anaphors need an antecedent that is in the same clause and that c-commands them. 6 Pronouns Data Set 1 • But what about pronouns? Ex. – John siw him. j – *John saw him. i i – John siid [that we saw him]. i – John siid [that we saw him]. j • Hypothesis #1: Pronouns can’t have an antecedent in the same clause. Pronouns Data Set 2 • Hypothesis #1: Pronouns can’t have an antecedent in the same clause. • But: Ex. – [Johni’s father] saj him. i – *[Johni’s father] saj him. j – [The mother of [that girl]] saw j i. j – *[The mother of [that girl]] saw j i. i • Hypothesis #2: Pronouns can’t have an antecedent in the same clause that c- commands them. Pronouns • This is essentially the hypothesis that we will adopt for pronouns: (Informally): pronouns CANNOT have an antecedent that c-commands them in the same clause. 7 8 Some Definiti ons • Now we’re going to formalize these observations somewhat. • We’re going to give a name to the situation where one indexed NP c-commands another NP with the same index: binding. Binds: A binds B if and only if A c-commands B and A and B are coindexed.  Binding = C‐Command AND Co‐Indexing!! • We also saw that anaphors and pronominals seem to care about which clause their potential antecedents are in. • We’re going to use the term binding domain to describe this phenomenon. Binding domain: The clause containing the NP (anaphor, pronominal, or R- expression). Binding Domain: An Example of Locality • Note that the idea of a Binding Domain is another example of locality. Locality: the grammar being sensitive to how close or far two things are in a tree (where ‘closeness’ is defined structurally). Principles A & B • With those two definitions, we can formalize the requirements of anaphors and pronominals: • Binding Principle A: An anaphor must be bound in its binding domain. • Binding Principle B: A pronominal must be free (=not bound) in its binding domain.  Free = Not Bound!! Restrictions on R‐expressions • There also appear to be restrictions on the use of R-expressions: Ex. – *She iaw Mary. i – *She said [that Mary left]. i i – *Janeisaw Mary. i • This data suggests R-expressions do not want to be bound, either within their binding domain or outside of it.  R-expressions do not want to be bound at all! (They always want to be free) 9 Principle C Binding Principle C: An R-expression must be free. • (Notice that this is not the same as saying that R-expressions cannot be coindexed with another NP.) Eg. 1.) Mary iaw herself i 2.) Mary iaid that she leit (1.) Mary is co-indexed with herself but Mary is binding herself, not the other way around (Mary is free) (2.) Mary is co-indexed with she but Mary is not bound. Mary binds she. Principles A, B, and C • Binding Principle A: An anaphor must be bound in its binding domain. • Binding Principle B: A pronominal must be free (=not bound) in its binding domain. • Binding Principle C: An R-expression must be free. An Example Problem Explain the ungrammaticality of the following sentence: state which binding principle is violated, and how. *Kate believes that Alice likes her . i • Identify all the NPs. • For each NP, identify if it is an anaphor, pronoun, or R-expression. – For each Anaphor, determine if it is bound in its domain. – For each pronoun, determine if it is free in its domain. – For each R-expression, determine if it is free. 10 X-Bar Theory (Ch.6) • Evidence for X-bar • one-replacement • do so replacement • X-bar rules • Generalizing X-bar rules • Complement vs. Adjunct distinction Draw the tree for… • Using our NP PSR (below), draw a tree for the phrase in (1): NP  ({D/NP’s}) (AdjP+) N (PP+) (CP+) 1. A clever student of English with high ambitions Now look at this data… The clever student of English with high ambitions, not the dull one. This is the one-replacement test. – It is a subtype of replacement/substitution constituency tests. – What problems does this piece of data above create for the tree we just drew? One-replacement • The clever student of English with high ambitions, not the dull one. • A clever student of English with high ambitions and a silly one. 11 • “One” seems to replace which constituent? – [student of English with high ambitions] – But is this a constituent in our tree? No! • What structure does the one-replacement test seem to argue for? – Make a modification to your tree that would explain the one-replacement test we just did. More data… • James is a clever [student of English] [with high ambitions], not a dull [one][with no ambition]. • A clever [student of English][with high ambitions] and a dull [one][with no ambition] both took the exam. –These constituency/one-replacement tests are telling us… one can not only replace [student of English with high ambitions] but also just [student of English] – Make another modification to your tree to reflect this… Even more data… 12 • The scholarship committee chose this [clever student of English with high ambitions], not that [one]. • Every [clever student of English with high ambitions] but that [one] will be awarded a fellowship. – This data indicates that [clever student of English with high ambitions] is also a constituent. – Modify your tree once more to reflect this… • We have a rule that tells us to draw the NP one way: NP  ({D/NP’s}) (AdjP+) N (PP+) (CP+) • And we have constituency tests using one-replacement that tell us the tree for this NP should look a different way. • Remember why we came up with PSRs in the first place… – As a way of formalizing our hypotheses about structure. – We came up with the rules based on the evidence we then had at our disposal. • Now we have new evidence that points towards a different structure. – So we shouldn’t let our rules, which afterall are just formalizations of our hypotheses, get in the way of our creating a new hypothesis about structure based on our analysis of evidence using constituency tests. – We can come up with new rules to represent our hypotheses In other words, we only created PSRs for a way of formalizing our hypothesis about structure and we can make a new hypothesis based on new rules We call these intermediate nodes “Bar levels” 13 • So now we are in a position to say that one-replacement targets an N’ node. One-replacement: Replace an N’ node with one. Things Wrong with PSRs -flat structure PSRs treat complements, modifiers, etc. in the same waysisters of the head -constituency/ -one-replacement PSRs predicts only some constituencies, but there are more Another example of flat structure in the NP: • Our theory (PSRs) allows us to describe what can and can’t occur in phrases (eg. in NPs), but it treats all complements, modifiers, etc. in the same way  as sisters of the head: Another example: constituency • But, there’s evidence that this type of structure is not correct. • For instance, this structure predicts the following constituency: 14 Another example: one-replacement • If this is correct, other groups of words should not form constituents. But some do. Ex. – Mary likes the tall [prince of Spain] with green eyes but not the tall [one] with blue eyes. – Mary likes the tall [prince of Spain with green eyes] but not the short [one]. – Mary likes this [tall prince of Spain with green eyes] but not that [one]. this works • But, in addition to these: – the tall [prince of Spain] with green eyes – the tall [prince of Spain with green eyes] – this/that [tall prince of Spain with green eyes] • there seems to be another possible constituent: – Mary likes the [tall prince of Spain] with the green eyes but not the [one] with the blue eyes. • So how can [tall prince of Spain] undergo one replacement; it’s not a constituent in our tree. So, we need a slightly different tree… 15 Notice the two readings: a.) of the tall princes of Spain; the one with green eyes b.) of the princes of Spain with green eyes, the tall one New Rules for NPs: • Data suggests new rules for NPs: – NP  (D) N’ – N’  (AdjP) N’ or N’ (PP) – N’  N (PP) • These rules incorporate an important property called recursion. Recursion: the ability to embed objects iteratively inside one another. in theory, allows an infinite number of modifiers. NP  (D) N’ N’  (AdjP) N’ or N’ (PP/CP) N’  N (PP) NP  (D) N’ N’  (AdjP) N’ or N’ (PP/CP) N’  N (PP) 16 Do So Replacements Structure of VPs • Draw a tree for [eats rice with a spoon] using the VP rule: VP  (AdvP+) V (NP) ({NP/CP}) (AdvP+) (PP+) (AdvP+) Some data to consider: • Alex [eats rice] [with a spoon] and Derek [does so] [with a fork]. • Alex [eats rice] [with a spoon] and Derek [does so] [without a spoon]. • Does so appears to be replacing the constituent [eats rice]. • Is [eats rice] a constituent in this tree? • Revise your tree to account for the constituency we have evidence for here. More data: • Alex [[eats rice] and [stirs porridge]] [with a spoon]. • Also evidence for this intermediate projection. Let’s call these intermediate VP nodes V’ (V-Bar) nodes 17 • It seems that do so replacement targets a V’ (V-Bar) node. Do-so Replacement: Replace a V’ node with do so (or do so too or do too). Another example… Mary hung pictures in the museum on Monday. – Mary [hung pictures in the museum on Monday] and Bill [did so] (too/also/as well). – Mary [hung pictures in the museum] on Monday and Bill [did so] on Tuesday. – Mary [hung pictures] in the museum on Monday and Bill [did so] in the gallery on Tuesday. • Draw a tree structure for this VP using V’ levels to make sure each of the phrases in blue is a constituent. • …[hung pictures] in the museum on Monday. • …[hung pictures in the museum] on Monday. • …[hung pictures in the museum on Monday]. 18 New Rules for VPs • So, new rules for VPs: – VP  V’ – V’  V’ (PP) or V’ (AdvP) – V’  V (NP) • (We don’t actually have evidence for the VP  V’ rule right now, but we’ll need it in a future lecture/chapter and for parsimony we want these rules to mirror our NP rules.) Are there Bar-levels in AdjPs? • Draw a tree using PSRs for the phrase: – [very bright blue hat] • Constituency tests also show evidence for intermediate levels in adjective phrases: – [very [bright blue]] hat – very [bright blue] and [dark green] hat Using Adj’… New Rules for AdjPs 19 • Data suggests a similar set of rules: – AdjP  Adj’ – Adj’  (AdvP) Adj’ – Adj’  Adj (PP) • (Although, as with VP, we haven’t see evidence for the AdjP  Adj’ rule…) Are there Bar-levels in AdvPs? • We don’t really have the same evidence for adverbs; partly because it sounds strange to stack modifiers on adverbs: Eg. – ?very exceedingly quickly • So… we’re going to just assume they have a similar structure: (Parsimony and simplicity…assume they are structured like the other phrases.) – AdvP  Adv’ – Adv’  (AdvP) Adv’ or Adv’ (PP) – Adv’  Adv Are there Bar-levels in PPs? • Similar evidence for intermediate structures in PPs: – [right [off the table]] – right [off the table] and [onto the floor] P’ New rules for PPs: • Similar set of rules for PPs: – PP  P’ – P’  (AdvP) P’ or P’ (PP) – P’  P (NP) X-Bar Theory • What we’ve been doing today is called X-Bar Theory. X is a variable that stands for any category (N, V, P, etc.) • Intermediate categories are indicated by writing a “bar” over the category: N (• Notational equivalencies: – Phrase level: NP = N” = Nmax – Intermediate level: N’ = N – Word/Head level: N = N°) 20 A Brief History of Syntax New Rules NP  (D) N’ N’  (AdjP) N’ or N’ (PP) N’  N (PP) VP  V’ V’  (AdvP) V’ or V’ ({AdvP/PP}) V’  V (NP) AdjP  Adj’ Adj’  (AdvP) Adj’ Adj’  Adj (PP) PP  P’ P’  (AdvP) P’ or P’ (PP) P’  P (NP) Generalizing X-Bar Rules • What do all these rules have in common? – Every phrase has a head that is the same category as the phrase this property is called endocentricity (it’s why there are no rules of the form: NP  V (AP).) – Every phrase has an intermediate level, ie. the bar level – Except for the determiner (which we’ll come back to next lecture), all non-head material is phrasal and optional. ie. anything that isn’t a head must be an optional phrase 21 Generalization 1: 3 types of rules For each major category there are 3 types of rules: • SPECIFIER RULE: A rule that generates the phrase NP  (D) N’ • ADJUNCT RULE: A rule that iterates: N’  (AP) N’! • COMPLEMENT RULE: A rule that introduces the “head” N’  N (PP) Generalization 2: Headedness In each rule the only item that is obligatory is the item that gives its category to the node that dominates it: • NP  (D) N’ • N’  (AP) N’ • N’  N (PP) There are no rules of the form NP  V AP !! ⇒ because of endocentricity! Generalization 3: Optionality All non-head material is both phrasal and optional (with the exception of determiners, dealt with in chapter 7), • NP  (D) N’ • N’  (AP) N’ • N’  N (PP) Some Goals of X-bar Theory Simplify the system of rules Capture intermediate structure Capture the cross-categorial generalizations. We use VARIABLES to do this. A variable is a category that can stand for any other category. X, Y, W, Z are variables that can stand for ANY of N, V, A, P X-bar Rules Revised • Essentially, all the rules capture the following: – SPECIFIER RULE: XP  (YP) X’ – ADJUNCT RULE: X’  (ZP) X’ or X’ (ZP) – COMPLEMENT RULE: X’  X (WP) • (where X is a variable for the categories N, V, P, Adj, Adv) 22 DEFINITIONS 23 Summary • Constituency tests give evidence of intermediate structures. • X-bar theory allows us to make generalizations across syntactic categories. – same 3 rules for each category – phrases must contain heads – modifiers are phrasal and optional • (exceptionality of D will be fixed in the next lecture) Complements, Specifiers, Adjuncts • When I draw a tree, how do I tell if something is a complement, specifier, or adjunct? Now when you draw your trees for sentences, you will always need to be thinking, should this constituent be a specifier, a complement, or an adjunct? Complement vs. Adjunct Distinction • Complements are often required/selected by heads, while adjuncts can usually be omitted: – I devoured the cake at the party. – *I devoured at the party. – I devoured the cake. • Complements usually occur closer to heads than adjuncts: – I ate breakfast in the morning. – *I ate in the morning breakfast. • Also, tests like do so replacement target V’ level and thus must replace complements: – Mary [ate a cake] and Bill [did so] (too/also). – *Mary [ate] a cake and Bill [did so] a pie. 24 • Similarly, one replacement tends to include complements (but somewhat less reliably): – */?The [king] of Spain and [the one] of England left. Let’s look at why this is… • Tests like do so replacement target V’ level and thus must replace complements: 1. Mary [ate a cake] and Bill [did so] (too/also). • Grammatical: “did so” replaces V + com
More Less

Related notes for LIN232H1

Log In


OR

Join OneClass

Access over 10 million pages of study
documents for 1.3 million courses.

Sign up

Join to view


OR

By registering, I agree to the Terms and Privacy Policies
Already have an account?
Just a few more details

So we can recommend you notes for your school.

Reset Password

Please enter below the email address you registered with and we will send you a link to reset your password.

Add your courses

Get notes from the top students in your class.


Submit