Syntax Marcel den Dikken Department of English Linguistics

Syntax Marcel den Dikken Department of English Linguistics ELTE

Syntax: What is it? Q what is ‘syntax’? A the branch of linguistic analysis that addresses the internal structure of sentences and the phrases that they include — the way phrases are arranged, put together (Gr συν ‘together’ + ταξις ‘arrangement’) to form sentences ● we will be interested in more than statements such as ‘a sentence consists of a subject and a predicate, in that order’ → such statements have a basic truth to them: many sentences of English are indeed like John kissed Mary subject predicate 2

Syntax: What is it? 3 Q what is ‘syntax’? A the branch of linguistic analysis that addresses the internal structure of sentences and the phrases that they include — the way phrases are arranged, put together (Gr συν ‘together’ + ταξις ‘arrangement’) to form sentences ● we will be interested in more than statements such as ‘a sentence consists of a subject and a predicate, in that order’ → such statements have a basic truth to them: many sentences of English are indeed like John kissed Mary → but there also plenty of sentences that are not: e. g. , John said that he’d kiss Mary, and kiss Mary, he did!

Syntax: What is it? (2) *John said that he’d kiss Mary, and kiss, he did Mary! (3) 4 John said that he’d kiss Mary, and Mary, he did kiss! (4) *John said that he’d kiss Mary, and kissed Mary, he! Q why can the entire predicate precede the subject (1) but not when it harbours -ed (4), and why can the predicate be ‘split’ as in (3) but not as in (2)? → these are important empirical questions that a theory of the syntax of English should be able to answer (1) John said that he’d kiss Mary, and kiss Mary, he did!

Syntax: What is it? 5 (1) John said that he’d kiss Mary, and kiss Mary, he did! ● note also that (1) consists of three sentences, one embedded in another (‘subordination’), and one conjoined with another (‘coordination’) → both subordination and coordination can be performed as many times as one wishes (5) John said that Bill thought that Bob believed that Frank… (6) John kissed Mary, and Bill kissed Sue, and Bob kissed Sarah, and Frank… → this illustrates an important property of human natural language: recursion

Syntax: What is it? ● → → 6 because both subordination and coordination are recursive, there can be no such thing as ‘the longest sentence of English’: take any really long sentence of English and you will find that it can always be made even longer, through subordination or coordination because there is no limit to the length and complexity of sentences, and because, as we had already seen, parts of sentences can often be ‘scrambled around’ to form new sentences (John said that he’d kiss Mary, and kiss Mary, he did!), the task of writing a syntax of English (or any other human natural language) cannot involve drawing up a list of possible sentences of the language we need a formal system capable of deriving (or ‘generating’) all possible sentences, and only those

Hierarchy Q what is ‘syntax’? A the branch of linguistic analysis that addresses the internal structure of sentences and the phrases that they include — the way phrases are arranged, put together (Gr συν ‘together’ + ταξις ‘arrangement’) to form sentences ● we will be interested in more than statements such as ‘a sentence consists of a subject and a predicate, in that order’ — we are after a principled understanding of the hierarchical organisation of the sentence and its constituent parts, the way in which this organisation comes about, and the ways in which it can be exploited by a restrictive formal theory of grammar 7

Hierarchy (1) ● ● (2) (3) 8 John said that he’d kiss Mary, and kiss Mary, he did! for (1), we will want our formal grammar to say that kiss Mary (the predicate) is a phrase that is syntactically independent of he (its subject) we will also want our grammar to say that these two phrases can be combined to form a sentence as long as some information about tense is added (*he kiss Mary) we will want the grammar to treat this tense information as relatively independent of the verb (kiss Mary, he did) and we will want the grammar to treat the direct object as a phrase independently of the verb without treating the verb as a phrase independently of the object *John said that he’d kiss Mary, and kiss, he did Mary! John said that he’d kiss Mary, and Mary, he did kiss!

Hierarchy (1) ● John said that he’d kiss Mary, and kiss Mary, he did! this suggests a particular organisation of the sentence, in the form of a hierarchical structure recognising the status of portions of the sentence as phrases → the structure grows bit by bit, phrase by phrase, until the whole sentence is complete: ‘ever increasing circles’ (2) (3) [ [he] did [kiss [Mary]]] *John said that he’d kiss Mary, and kiss, he did Mary! John said that he’d kiss Mary, and Mary, he did kiss! 9

Hierarchy 10 (1) ● John said that he’d kiss Mary, and kiss Mary, he did! this suggests a particular organisation of the sentence, in the form of a hierarchical structure recognising the status of portions of the sentence as phrases → this structure produces (3) by blanking out the phrase following kiss, and placing a copy of it to the left of he [ ] [Mary], [he] did [kiss [Mary]] (2) (3) *John said that he’d kiss Mary, and kiss, he did Mary! John said that he’d kiss Mary, and Mary, he did kiss!

Hierarchy 11 (1) ● John said that he’d kiss Mary, and kiss Mary, he did! this suggests a particular organisation of the sentence, in the form of a hierarchical structure recognising the status of portions of the sentence as phrases NB 1 in this structure, there is no phrase that consists only of the verb kiss: the smallest phrase containing kiss also contains Mary this explains why (2) is bad, assuming that only phrases can be blanked out → (2) (3) [ [he] did [kiss [Mary]]] *John said that he’d kiss Mary, and kiss, he did Mary! John said that he’d kiss Mary, and Mary, he did kiss!

Hierarchy 12 (1) ● John said that he’d kiss Mary, and kiss Mary, he did! this suggests a particular organisation of the sentence, in the form of a hierarchical structure recognising the status of portions of the sentence as phrases NB 2 in this structure, the smallest phrase containing kiss Mary and tense (did or -ed) also contains he → this explains why (4) is bad, again assuming that only phrases can be blanked out (4) [ [he] did [kiss [Mary]]] *John said that he’d kiss Mary, and kissed Mary, he!

Hierarchy 13 (1) ● John said that he’d kiss Mary, and kiss Mary, he did! this suggests a particular organisation of the sentence, in the form of a hierarchical structure recognising the status of portions of the sentence as phrases → but we can produce (1), by blanking out the phrase following did, and placing a copy of it to the left of he [ kiss [Mary]], [ [he] did [kiss [Mary]]]

Hierarchy (1) ● ● 14 John said that he’d kiss Mary, and kiss Mary, he did! this suggests a particular organisation of the sentence, in the form of a hierarchical structure recognising the status of portions of the sentence as phrases we need to understand this ‘blanking out’ and ‘copying’ much better before we can claim victory [ kiss [Mary]], [ [he] did [kiss [Mary]]]

Hierarchy (1) ● ● → → 15 John said that he’d kiss Mary, and kiss Mary, he did! this suggests a particular organisation of the sentence, in the form of a hierarchical structure recognising the status of portions of the sentence as phrases we need to understand this ‘blanking out’ and ‘copying’ much better before we can claim victory but hierarchical structures of this type are clearly useful as a tool in the analysis of syntactic structures [ [he] did [kiss [Mary]]] we will exploit such hierarchical structures to the full in the approach to English syntax in this department

Hierarchy ● ● Q 1 Q 2 but before we can put these hierarchical structures to optimal use, we will need to fill in some of their details the first things to figure out are how to arrive at these triangles, and how to refer to the triangles in a more efficient way than with expressions such as ‘the phrase to the right of did’ do these triangles have labels, and if so, how are these labels determined? do these triangles have their own internal structure, and if [he] did [kiss [Mary]] so, how does it come about? we will exploit such hierarchical structures to the full in the approach to English syntax in this department [ → 16 ]

The Lexicon 17 ● syntactic structures are projections of properties stored for each word or subword atom (‘morpheme’, such as -ed in our previous example) in a great pre-syntactic storage house called the lexicon ● the lexicon harbours at least three syntactically relevant types of information – category – subcategorisation – thematic roles → it is these lexical specifications of a word or morpheme that determine its syntactic distribution — the kind of phrase that it projects in the syntax

Categories ● there are four lexical categories – nouns N – verbs V – adjectives A – prepositions P 18 adoration, bath adore, bathe boastful, fond from, into → we can put these four categories into a matrix with exactly four cells by using two basic features, each with either a positive (+) or a negative (–) value → perhaps somewhat confusingly, the two categorial features standardly used in the literature are [+N] and [+V]

Categories ● there are four lexical categories – nouns N – verbs V – adjectives A – prepositions P 19 adoration, bath adore, bathe boastful, fond from, into +N –N +V Adjectives (A) Verbs (V) –V Nouns (N) Prepositions (P) → the [+V] categories (verbs and adjectives) are the pure predicates — these can only be used predicatively → the [–N] categories are the potential structural Case assigners — they can take a nominal object

Categories ● there are four lexical categories – nouns N – verbs V – adjectives A – prepositions P his adoration *(of) himself he adores (*of) himself he is very fond *(of) himself he is very much into (*of) himself → 20 adoration, bath adore, bathe boastful, fond from, into we will come back to this later the [–N] categories are the potential structural Case assigners — they can take a nominal object

Categories ● there are four lexical categories – nouns N – verbs V – adjectives A – prepositions P 21 adoration, bath adore, bathe boastful, fond from, into ● in addition to the four lexical categories, there are several functional categories, each of which belongs to one of the four lexical categories — for instance: N: V: A: P: – – quantifier tense degree particle Q T Deg Prt some, many -ed very, too up, back

Categories → we can make a formal distinction between lexical and functional categories with the aid of the additional feature [+F], for ‘functional’ –F +N –N +V Adjectives (A) Verbs (V) –V Nouns (N) Prepositions (P) +F +N –N +V Degree (Deg) Tense (T) –V Quantifier (Q) Particle (Prt) 22

Categories NB 23 there are quite a few more functional categories than the four mentioned above: while in the [–F] matrix, each cell as exactly one occupant, in the [+F] matrix there can be several candidates for each cell –F +N –N +V Adjectives (A) Verbs (V) –V Nouns (N) Prepositions (P) +F +N –N +V Degree (Deg) Tense (T) Complementiser (C) –V Quantifier (Q) Determiner (D) Particle (Prt)

Categories –F +N –N +V Adjectives (A) Verbs (V) –V Nouns (N) Prepositions (P) +F +N –N +V Degree (Deg) Tense (T) Complementiser (C) –V → Quantifier (Q) Determiner (D) Particle (Prt) a phrase like the many baths we can now analyse as: [DP the [QP many [NP baths]]] 24

Categories –F +N –N +V Adjectives (A) Verbs (V) –V Nouns (N) Prepositions (P) +F +N –N +V Degree (Deg) Tense (T) Complementiser (C) –V → Quantifier (Q) Determiner (D) Particle (Prt) a phrase like very boastful we can now analyse as: [Deg. P very [AP boastful]] 25

Categories –F +N –N +V Adjectives (A) Verbs (V) –V Nouns (N) Prepositions (P) +F +N –N +V Degree (Deg) Tense (T) Complementiser (C) –V → Quantifier (Q) Determiner (D) Particle (Prt) a phrase like up from the floor we can now analyse as: [Prt. P up [PP from [DP the [NP floor]]]] 26

Categories 27 we will come back to the position of the subject –F +N –N +V Adjectives (A) Verbs (V) –V Nouns (N) Prepositions (P) +F +N –N +V Degree (Deg) Tense (T) Complementiser (C) –V Quantifier (Q) Determiner (D) Particle (Prt) → a subordinate sentence we can now analyse as: I said [CP that [TP [DP the [NP groom]] DID [VP kiss [DP the [NP bride]]]]]

The Lexicon 28 ● syntactic structures are projections of properties stored for each word or subword atom (‘morpheme’, such as -ed in our previous example) in a great pre-syntactic storage house called the lexicon ● the lexicon harbours at least three syntactically relevant types of information ✓ – category – subcategorisation – thematic roles

Subcategorisation 29 ● for the four lexical categories, we can distinguish between several subcategories based on their selectional properties → thus, there is a subcategory of verbs that take no direct object (‘intransitive’), there is also a subcategory of verbs that do (‘transitive’), and there is even a subcategory of verbs that take both a direct and an indirect object (‘ditransitive’) – sleep category: [–F, –N, +V] subcat: [ ___ Ø] – devour category: subcat: [–F, –N, +V] [ ___ DP] – give category: subcat: [–F, –N, +V] [ ___ DP DP]

Subcategorisation 30 ● for the four lexical categories, we can distinguish between several subcategories based on their selectional properties → for the other lexical categories, the range of variation in the realm of subcategorisation is less pronounced — but we do find significant variation even there thus, there is a distinction among nouns between those that do not take an object and those that do (CP or PP) – bath category: [–F, +N, –V] subcat: [ ___ Ø] → – belief category: subcat: [–F, +N, –V] [ ___ CP/PP] his belief [CP that he is great] / [PP in himself]

Subcategorisation 31 ● for the four lexical categories, we can distinguish between several subcategories based on their selectional properties → for the other lexical categories, the range of variation in the realm of subcategorisation is less pronounced — but we do find significant variation even there is a distinction among adjectives between those that do not take an object and those that do (CP or PP) – ill category: [–F, +N, +V] subcat: [ ___ Ø] → – hopeful category: subcat: [–F, +N, +V] [ ___ CP] – independent category: subcat: [–F, +N, +V] [ ___ PP]

Subcategorisation 32 NB the subcategorisation of the adverb independently is the same as that of the adjective independent: independent(ly) of the government → this lends support for the idea that there is no separate lexical category of ‘adverbs’: -ly is a functional element that, just like -ed, does not affect the lexical category of its host – ill category: subcat: [–F, +N, +V] [ ___ Ø] – hopeful category: subcat: [–F, +N, +V] [ ___ CP] – independent category: subcat: [–F, +N, +V] [ ___ PP]

Subcategorisation 33 NB the subcategorisation of the adverb independently is the same as that of the adjective independent: independent(ly) of the government → this lends support for the idea that there is no separate lexical category of ‘adverbs’: -ly is a functional element that, just like -ed, does not affect the lexical category of its host → this is a reassuring conclusion: after all, our matrix of lexical categories already had all of its four cells filled; there would not be any room for a separate lexical category of ‘adverbs’ –F +N –N +V Adjectives (A) Verbs (V) –V Nouns (N) Prepositions (P)

The Lexicon 34 ● syntactic structures are projections of properties stored for each word or subword atom (‘morpheme’, such as -ed in our previous example) in a great pre-syntactic storage house called the lexicon ● the lexicon harbours at least three syntactically relevant types of information ✓ – category ✓ – subcategorisation – thematic roles

Theta-roles/grids ● ● → 35 separately from their subcategorisation properties, we can also make distinctions for the lexical categories in terms of the number and nature of thematic roles (or theta-roles, θ-roles) that they assign θ-roles represent the participants in the eventuality denoted by the predicate head an event of ‘sleeping’ involves just a sleeper (Agent); we add this information to the lexical entry for sleep in the form of a so-called θ-grid – sleep category: subcat: θ-grid: [–F, –N, +V] [ ___ Ø]

Theta-roles/grids ● ● → 36 separately from their subcategorisation properties, we can also make distinctions for the lexical categories in terms of the number and nature of thematic roles (or theta-roles, θ-roles) that they assign θ-roles represent the participants in the eventuality denoted by the predicate head an event of ‘giving’ involves a giver (Agent), a beneficiary (Goal), and a thing given (Theme), as registered in the θ-grid for the verb give – give category: subcat: θ-grid: [–F, –N, +V] [ ___ DP DP]

Theta-grid vs Subcat 37 NB there is a degree of overlap between θ-grids and subcategorisation frames: both register information about the predicate head’s object(s) ● but there are important differences between the two: (i) the subcat frame only lists objects: the Agent in give’s θ-grid has no match in the subcat frame (ii) the subcat frame includes information about the category of the object(s) and about their position relative to the predicate head – → give category: [–F, –N, +V] subcat: [ ___ DP DP] θ-grid: the Goal and Theme are the internal arguments of give; the Agent is the external argument of give

Theta-grid vs Subcat 38 NB there is a degree of overlap between θ-grids and subcategorisation frames: both register information about the predicate head’s object(s) ● but there are important differences between the two: (i) the subcat frame only lists objects: the Agent in give’s θ-grid has no match in the subcat frame (ii) the subcat frame includes information about the category of the object(s) and about their position relative to the predicate head → there is variation regarding the categorial properties of the internal arguments of a predicate head, and with respect to their placement vis-à-vis the head there is no such variation for the external argument: it is overwhelmingly nominal and precedes the predicate →

Structure Building ● → → 39 syntactic structures are projections of properties stored for each word or morpheme in the lexicon category subcategorisation frame θ-grid whenever we want to build a syntactic phrase, we take a word or morpheme from the lexicon and make it input to the process of syntactic structure building, respecting the lexical information registered for that word/morpheme the phrase built around a word or morpheme has as its label the category of that word or morpheme we call the element that gives its label to a phrase the head of the phrase every phrase has a head – the principle of endocentricity

Structure Building ● → 40 for any head X (where ‘X’ is a variable for any member of the set of lexical and functional categories), syntactic structures are projected in accordance with a general projection schema — X-bar Theory we can think of X-bar Theory as a manual for building projections from a head equivalently: – XP X, or X 1 equivalently: ‗ X, or X′′, or X 2, or Xmax ZP X′ X [XP ZP [X′ X YP YP ]]

Structure Building 41 ● for any head X (where ‘X’ is a variable for any member of the set of lexical and functional categories), syntactic structures are projected in accordance with a general projection schema — X-bar Theory → we can think of X-bar Theory as a manual for building projections from a head short for: XP daughter of X′, sister of X short for: daughter of XP, sister of X′ ZP specifier of XP X′ X [XP ZP [X′ X complement of X YP YP ]]

Structure Building NB 1 X-bar Theory does not demand that every X combine with a YP and a ZP to form a complex phrase → either YP or ZP, or both, can be absent; but X will still project a phrase in accordance with the X-bar schema whenever either YP or ZP, or both, are absent, X will project vacuously to the next projection level → XP ZP X′ X [XP ZP [X′ X ]] 42

Structure Building NB 1 X-bar Theory does not demand that every X combine with a YP and a ZP to form a complex phrase → either YP or ZP, or both, can be absent; but X will still project a phrase in accordance with the X-bar schema whenever either YP or ZP, or both, are absent, X will project vacuously to the next projection level → XP X′ X [XP [X′ X YP YP ]] 43

Structure Building NB 1 X-bar Theory does not demand that every X combine with a YP and a ZP to form a complex phrase → either YP or ZP, or both, can be absent; but X will still project a phrase in accordance with the X-bar schema whenever either YP or ZP, or both, are absent, X will project vacuously to the next projection level → XP X′ X [XP [X′ X ]] 44

Structure Building 45 NB 2 X-bar Theory by itself does not ensure that X′ and XP can have at most two daughters → the rules of X-bar Theory in principle allow n number of daughters for any projection (where n ≽ 1) but since seminal work by Kayne (1984), it has been standard to restrict X-bar structures to binary branching note, however, that we will not require X-bar projections to be strictly binary branching: just a single daughter is allowed (vacuous projection) → → *X′ X YP ZP

Structure Building Q 46 what to do when X seems to combine with two phrases? read [the book] [very carefully] give [the book] [to a student] / give [a student] [the book] talk [to the students] [about syntax] → there are two ways to approach such cases, a priori: (a) by making an appeal to adjunction (b) by making an appeal to specification *X′ we will talk about option (b) a bit later in the show X YP ZP

Structure Building Q 47 what to do when X seems to combine with two phrases? read [the book] [very carefully] give [the book] [to a student] / give [a student] [the book] talk [to the students] [about syntax] → there are two ways to approach such cases, a priori: (a) by making an appeal to adjunction X′ under adjunction, the label and Xbar level of the host are repeated X′ X [X′ X YP ZP YP ] ZP ]

Adjunction 48 ● X-bar Theory allows any of its three levels to be repeated as many times as necessary to add material that serves some modifying function — this is called adjunction → adding an adjunct does not raise the bar level: it repeats it instead (i) adjunction to the head X′ X Y [X′ [X Y [X examples for X=N and Y=N: arm chair maths student house rules X X ]]]

Adjunction 49 ● X-bar Theory allows any of its three levels to be repeated as many times as necessary to add material that serves some modifying function — this is called adjunction → adding an adjunct does not raise the bar level: it repeats it instead (ii) adjunction to X′ X′ NB: only a phrase (YP) can adjoin above the head YP X′ X [X′ YP [X′ X ]] examples for X′=N′, YP=AP: comfy chair smart student silly rules

Adjunction 50 ● X-bar Theory allows any of its three levels to be repeated as many times as necessary to add material that serves some modifying function — this is called adjunction → adding an adjunct does not raise the bar level: it repeats it instead XP (iii) adjunction to XP NB: the direction of adjunction (to the left or right) is free in principle, but often a function of the type of adjunct XP X′ YP example for XP=NP, YP=CP: rules that/which I told you about X [XP [X′ X ]] YP ]

Adjunction 51 ● X-bar Theory allows any of its three levels to be repeated as many times as necessary to add material that serves some modifying function — this is called adjunction → adding an adjunct does not change the X-bar structure; we should always be free to either insert or not insert an adjunct → adjuncts are optional ● adding a complement or a specifier, by contrast, does alter the X-bar structure → complements and specifiers as a rule are not optional

Projection Principle 52 Q what determines whether a given head X does or does not take a complement or specifier? A for lexical heads, this is entirely a function of their selectional properties the projection of lexical selectional properties into the syntax is regulated by the Projection Principle → Projection Principle structures are projected from the lexicon at all levels for now, we need not worry about ‘levels’

Projection Principle 53 ● consider a head that is lexically specified as follows: lie category: [–F, –N, +V] subcat: [DP PP] θ-grid: → the projection of these lexical properties into the syntax in accordance with the Projection Principle demands a core structure of the following sort: we call lie an unaccusative verb — a verb which projects all of its θ-roles inside VP (see Ch. 5, § 2. 1) VP DP they V′ V lie both DP and PP serve as internal arguments of V, projected internal to VP PP down

Projection Principle 54 ● consider now a head that is lexically specified as follows: laugh category: [–F, –N, +V] subcat: [(PP)] θ-grid: → the projection of these lexical properties into the syntax in accordance with the Projection Principle demands a core structure of the following sort: we call laugh an intransitive (or unergative) verb — a verb with no nominal argument inside VP (see Ch. 5, § 2. 5) VP V′ V laugh (PP) at me

Projection Principle 55 ● consider now a head that is lexically specified as follows: laugh category: [–F, –N, +V] subcat: [(PP)] θ-grid: → the projection of these lexical properties into the syntax in accordance with the Projection Principle demands a core structure of the following sort: VP because only arguments included in the subcat frame get projected inside VP, the Agent cannot be projected internal to VP: it is an external argument V′ V laugh how do we introduce the external argument? (PP) at me

External Arg & ‘little v’ ● → 56 to answer this question, let us go back to lie category: [–F, –N, +V] subcat: [DP PP] θ-grid: lie itself has no external argument, but we can provide it with one in syntax in either of two ways: (a) addition of a ‘light verb’, make (b) lexical causativisation (realised as ‘ablaut’) VP DP they V′ V lie how do we introduce the external argument? PP down

External Arg & ‘little v’ (a) addition of the ‘light verb’ make v. P v′ DP we v make note that we are clustering in v (a) the ability to assign an external θ-role and (b) the ability to assign accusative Case VP DP they them Burzio’s Generalisation a verb that assigns no θext V cannot assign accusative Case lie V′ they must now be changed into them —v assigns accusative Case to this DP PP down 57

External Arg & ‘little v’ (b) lexical causativisation v. P v′ DP we v ABLAUT also, lay must now be pronounced at v — we will return to this shortly VP DP them V′ V lay lie changes into lay (via ‘ablaut’) in the phonological component PP down 58

External Arg & ‘little v’ 59 ● at this point, we can return to laugh: laugh category: [–F, –N, +V] subcat: [(PP)] θ-grid: → we can exploit the ‘light verb’ v now to add the external argument (Agent) for laugh VP because there is no DP in V’s subcat frame, the Agent cannot be projected internal to VP: it is an external argument V′ V laugh how do we introduce the external argument? (PP) at me

External Arg & ‘little v’ v. P the ‘light verb’ v is itself silent here — a null morpheme (Ø) v′ DP they v Ø VP V′ V laugh (PP) at me 60

External Arg & ‘little v’ 61 ● the ‘light verb’ v is also very useful for ditransitive verbs: give category: [–F, –N, +V] subcat: [DP DP] θ-grid: → the two internal arguments are introduced in VP… note that the Spec. VP position in this tree is where the Goal is projected, with the Theme projected in DP V’s complement — them but previously we always used Spec. VP for the Theme VP V′ V give this raises an issue for UTAH (p. 107); see p. 187 for a possible way to deal with it DP candy

External Arg & ‘little v’ → 62 … and the ‘light verb’ v introduces the external argument v. P v′ DP we v Ø a case of movement again, the ‘light verb’ v is itself silent — but we can use v to our advantage to get the word order right, by getting V spelled out at v VP DP them V′ V give DP candy

External Arg & ‘little v’ → 63 … and the ‘light verb’ v introduces the external argument v. P v′ DP we v give a case of movement … leaving a trace of the moved verb behind in its original (‘Dstructure’) position VP DP them V′ V t DP candy

Case and Inflection ● 64 for we give them candy, this tree now seems complete — but is this truly the whole story? v. P them can get accusative Case from v under government: ‘go up one node, and then look down until you find another governor’ v′ DP we v give VP ✓ DP them V′ V t DP candy

Case and Inflection ● 65 for we give them candy, this tree now seems complete — but is this truly the whole story? v. P ditransitive verbs like give are lexically equipped with the ability to assign accusative Case to their complement v′ DP we v give VP ✓ DP them V′ V t ✓ DP candy

Case and Inflection ● 66 for we give them candy, this tree now seems complete — but is this truly the whole story? v. P → NO! v′ DP we v give this is a problem! for the Case Filter demands that all DPs be assigned Case (see p. 112) but nothing in this structure is capable of assigning nominative Case to we VP ✓ DP them V′ V t ✓ DP candy

Case and Inflection v. P v′ DP we v give 67 note also that if we had chosen a third person singular subject, or a past tense environment, we would have missed out on some important morphology VP ✓ DP them V′ V t ✓ DP candy

Case and Inflection * v. P v′ DP he v give 68 note also that if we had chosen a third person singular subject, or a past tense environment, we would have missed out on some important morphology VP ✓ DP them V′ V t ✓ DP candy

→ Case and Inflection 69 to turn a ‘small clause’ v. P into a complete sentence, we need to add inflection (tense, agreement) *yesterday… v. P note also that if we had chosen v′ DP we v give a third person singular subject, or a past tense environment, we would have missed out on some important morphology VP ✓ DP them V′ V t ✓ DP candy

70 Building the Sentence → to turn a ‘small clause’ v. P into a complete sentence, we need to add inflection (tense, agreement) → for inflection, we need to create space in the structure John said that he would kiss Mary, and[kiss Mary], John claims that he adores Mary, and[adore Mary], → he did he does inflection (T, Agr) is syntactically autonomous vis-à-vis the verb phrase

Building the Sentence → to turn a ‘small clause’ v. P into a complete sentence, we need to add inflection (tense, agreement) → 71 for inflection, we need to create space in the structure *John said that he would kiss Mary, and[kissed Mary], he (did) *John claims that he adores Mary, and[adores Mary], he (does) → inflection (T, Agr) is syntactically autonomous vis-à-vis the verb phrase and we need to remove the subject from v. P we need to separate inflection from the v. P

Building the Sentence LEXICON and thus … and pick derive (for Inflection) I' I so let’s go back to the lexicon … now we project I v. P DP v' John v+V DP kiss Mary now we merge in v. P as a complement 72

Building the Sentence suppose that we choose future tense, i. e. , will I' v. P I will DP v' John we want to get John to the left of will v+V DP kiss Mary WRONG ORDER! 73

Building the Sentence so let’s project I' Spec … and derive I' IP v. P I will now, we can merge a specifier DP v' John v+V DP kiss Mary we haven’t completed the I-projection yet 74

Building the Sentence IP Spec I' v. P I will DP v' John v+V DP kiss Mary we can now move the DP of John into Spec. IP 75

NP-movement 76 IP DP I' what about this position? John I will Projection Principle structures are projected from the lexicon at all levels we cannot just get rid of it: the DP of John gets its θ-role here! v. P v' θ-roles must remain visible throughout the derivation v+V DP kiss Mary

77 NP-movement IP DPi co-indexation we leave a trace of the moved phrase behind a chain (DPi, ti) v. P I' John I will DP v' ti the head of the chain v+V DP kiss Mary the foot of the chain

NP-movement IP DPi I' θ-Criterion (cf. p. 109) a θ-role is assigned to one and only one argument an argument is assigned one and only one θ-role v. P a chain (DPi, ti) v' Spec. IP is a θ'-position John I will no θ-role 78 DP ti θ-role v+V DP kiss Mary a well-formed chain, uniform and in full compliance with the θ-Criterion

NP-movement IP DPi I' John v. P I will no θ-role DP v' ti θ-role v+V DP kiss Mary now we know that movement of the subject from Specv. P to Spec. IP is legitimate 79

NP-movement IP DPi all instances of movement have a trigger by ‘trigger’ we mean some (morpho)syntactic driving force — a ‘deep’ answer to the question of why things happen I' John v. P I will no θ-role DP v' ti θ-role v+V DP kiss Mary BUT… we don’t know yet why the subject moves from Specv. P to Spec. IP 80

81 Trigger for Movement IP DPi all instances of movement have a trigger I' John v. P I will no θ-role DP v' why isn’t the subject content to stay in its θ-position? what is it that the subject lacks in its base position and gains in its derived position? ti θ-role v+V DP kiss Mary a θ-role? plainly not!

Trigger for Movement IP DPi the subject entertains a privileged relation with the I-node of the finite clause I' John v. P I will DP v' ti v+V DP kiss Mary 82

Trigger for Movement we will want to relate the movement of the subject to Spec. IP to these privileges the subject entertains a privileged relation with the I-node of the finite clause checking of Agreement features against Infl John {is/*am/*are} stroking the cat assignment of nominative Case by Infl {he/*him} is stroking the cat Case Filter all DPs must be assigned Case 83

A little aside… abstract or structural Case is present on a DP even where there are no morphological case distinctions abstract or structural Case (always written with an initial capital) is to be distinguished morphological case checking of Agreement features against Infl John {is/*am/*are} stroking the cat assignment of nominative Case by Infl {he/*him} is stroking the cat Case Filter all DPs must be assigned Case 84

Trigger for Movement we will want to relate the movement of the subject to Spec. IP to these privileges to figure out how to do so, we have to know how I agrees with the subject and assigns it nominative Case checking of Agreement features against Infl John {is/*am/*are} stroking the cat assignment of nominative Case by Infl {he/*him} is stroking the cat Case Filter all DPs must be assigned Case 85

Trigger for Movement HOW DOES I AGREE WITH THE SUBJECT AND ASSIGN IT NOMINATIVE CASE? to figure out how to do so, we have to know how I agrees with the subject and assigns it nominative Case checking of Agreement features against Infl John {is/*am/*are} stroking the cat assignment of nominative Case by Infl {he/*him} is stroking the cat Case Filter all DPs must be assigned Case 86

Trigger for Movement HOW DOES I AGREE WITH THE SUBJECT AND ASSIGN IT NOMINATIVE CASE? agreement and nominative Case assignment happen under Spec-Head agreement checking of Agreement features against Infl John {is/*am/*are} stroking the cat assignment of nominative Case by Infl {he/*him} is stroking the cat Case Filter all DPs must be assigned Case 87

Trigger for Movement HOW DOES I AGREE WITH THE SUBJECT AND ASSIGN IT NOMINATIVE CASE? agreement and nominative Case assignment happen under Spec-Head agreement IP DPi this now gives us the trigger for movement to Spec. IP I' he Ii v. P is ti stroking the cat 88

Structural Case Assignment BUT… Case is assigned under government via Spec-Head agreement with I, the subject satisfies the Case Filter all DPs must be assigned Case 89

Structural Case Assignment BUT… not very helpful for us: then I would assign Case under government to the subject in Specv. P! (‘go up one node, and look down until you find another governor’) Case is assigned under government IP DPi I' he Ii v. P is ti stroking the cat 90

Structural Case Assignment → ● → a government-based approach to nominative Case assignment will lose us Case as a trigger for movement of the subject from Specv. P to Spec. IP more seriously, it will also make an incorrect empirical prediction from the sentences below we know that accusative Case assignment to the object (which takes place under government) is subject to an adjacency condition (p. 204) John speaks English fluently *John speaks fluently English → 91 but I need not be adjacent to Specv. P, indicating that I does not assign Case to Specv. P under government [IP hei will probably [v. P ti stroke the cat]]

Structural Case Assignment 92 NB note that by the same logic, we should not decide to have I assign nominative Case under government to Spec. IP either → not only would this require ‘stretching’ the reach of the government relation (from ‘one node up’ to ‘all the way up to the maximal projection’), it would also face difficulty in the face of the adjacency condition on Case assignment under government → the second example below is just as good as the first: I does not need to be adjacent to either the foot or the head of the chain of the subject [IP hei will probably [v. P ti stroke the cat]] [IP hei probably will [v. P ti stroke the cat]]

Structural Case Assignment Case is assigned under government for nominative Case, we will not follow this line, and instead exploit the link between NOM Case and agreement checking of Agreement features against Infl John {is/*am/*are} stroking the cat assignment of nominative Case by Infl {he/*him} is stroking the cat nominative Case is assigned under Spec-Head agreement 93

Structural Case Assignment 94 SUMMARY • • structural nominative Case is assigned under Spec-Head agreement by [+Agr] in I → see Portuguese (i) a. b. c. será difícil [eles aprovarem a proposta] will. be difficult they approve. 3 PL the proposal * será difícil [eles aprovar a proposta] será difícil [aprovar a proposta]

Structural Case Assignment 95 SUMMARY • • structural nominative Case is assigned under Spec-Head agreement by [+Agr] in I • • structural accusative Case is assigned under government by a [–N] lexical category (ii) a. b. c. d. I saw him I looked at him I am fond *(of) him I am the eldest brother *(of) him [–N, +V] [–N, –V] [+N, +V] [+N, –V]

Structural Case Assignment 96 • one last thing before we leave this topic • English cannot produce sentences like Portuguese (ia), with a nominative subject of the infinitive (i) • a. será difícil [eles aprovarem a proposta] but English can mimic (ia), as in (iiia) (iii) a. b. c. it’ll be hard [for them to approve the proposal] * it’ll be hard [them to approve the proposal] it’ll be hard [to approve the proposal] what is this for, and how does it assign accusative Case?

Structural Case Assignment 97 for assigns accusative Case under government • we see this Case-assigning for also in (iva) • (ivb) shows that for+him is not a constituent here • and (ivc) shows that the adjacency condition on accusative Case assignment is in effect here (iii) a. b. c. it’ll be hard [for them to approve the proposal] * it’ll be hard [them to approve the proposal] it’ll be hard [to approve the proposal] (iv) a. b. c. I prefer for him to tell me the truth * for him, I prefer to tell me the truth I prefer (sincerely) for (*sincerely) him to tell me the truth

Structural Case Assignment for assigns accusative Case under government V' V CP ? ? a complementiser prefer ? ? IP for DPi I' him I v. P to ti tell me the truth 98

Complementiser Phrases we had already discovered another example of a CP earlier V' V CP prefer C IP for DPi I' him I v. P to ti tell me the truth 99

Complementiser Phrases V' V CP think C IP that DPi I' he I v. P -ed ti tell me the truth 100

Complementiser Phrases in both examples, the complementiser itself is optionally present: C can be occupied by a silent morpheme V' V CP think C IP that DPi I' he I v. P -ed ti tell me the truth 101

Complementiser Phrases in both examples, the complementiser itself is optionally present: C can be occupied by a silent morpheme V' V CP think C IP Ø DPi I' he I v. P -ed ti tell me the truth 102

Complementiser Phrases in both examples, the complementiser itself is optionally present: C can be occupied by a silent morpheme V' V CP prefer C IP Ø DPi I' him I v. P to ti tell me the truth 103

External Arg & ‘little v’ regarding silent morphemes, recall the discussion of v v. P v′ DP they v Ø VP V′ V laugh (PP) at me 104

Complementiser Phrases back to CP and overt instances of the complementiser — what other examples can we think of? V' V CP think C IP that DPi I' he I v. P -ed ti tell me the truth 105

Complementiser Phrases back to CP and overt instances of the complementiser — what other examples can we think of? V' V CP wonder ✓ C IP if DPi I' he I v. P -ed ti tell me the truth 106

Complementiser Phrases back to CP and overt instances of the complementiser — what other examples can we think of? V' V CP wonder * C IP whether DPi I' he I v. P -ed ti tell me the truth 107

Complementiser Phrases I wonder [whether or not] he told me the truth * I wonder [if or not] he told me the truth * I think [that or not] he told me the truth V' V whether is clearly phrasal in ways that and if are not CP wonder * C IP whether DPi I' he I v. P -ed ti tell me the truth 108

Complementiser Phrases I wonder [whether or not] he told me the truth * I wonder [if or not] he told me the truth * I think [that or not] he told me the truth V' V CP wonder * C IP whether DPi for true complementisers such as that and if, it makes sense that they cannot be coordinated with not, which itself projects a phrase (cf. certainly not) I' he the Law of Coordination of Likes I v. P -ed ti tell me the truth 109

Complementiser Phrases I wonder [whether or not] he told me the truth * I wonder [if or not] he told me the truth * I think [that or not] he told me the truth V' V CP wonder * C IP whether DPi now the fact that whether can be coordinated with not strongly suggests that whether is, like not, a phrase — if so, it cannot occupy the C-position I' he the Law of Coordination of Likes I v. P -ed ti tell me the truth 110

Complementiser Phrases I wonder [whether or not] he told me the truth * I wonder [if or not] he told me the truth * I think [that or not] he told me the truth V' V CP wonder * C IP whether DPi now the fact that whether can be coordinated with not strongly suggests that whether is, like not, a phrase — if so, it cannot occupy the C-position I' he Q: where can we put whether, then? I v. P -ed ti tell me the truth 111

Complementiser Phrases I wonder [whether or not] he told me the truth * I wonder [if or not] he told me the truth * I think [that or not] he told me the truth V' V CP wonder * C IP whether DPi note that we had left the C-projection rather incomplete up to this point: we projected C straight up to CP, for simplicity; but there is, of course, also a C′ I' he Q: where can we put whether, then? I v. P -ed ti tell me the truth 112

Complementiser Phrases I wonder [whether or not] he told me the truth * I wonder [if or not] he told me the truth * I think [that or not] he told me the truth CP ? ? * C' C IP whether DPi note that we had left the C-projection rather incomplete up to this point: we projected C straight up to CP, for simplicity; but there is, of course, also a C′ I' he Q: where can we put whether, then? I v. P -ed ti tell me the truth 113

Complementiser Phrases I wonder [whether or not] he told me the truth * I wonder [if or not] he told me the truth * I think [that or not] he told me the truth CP Spec. CP is the perfect place to put phrasal Spec * whether C' C IP whether DPi I' he Q: where can we put whether, then? I v. P -ed ti tell me the truth 114

Complementiser Phrases I wonder [whether or not] he told me the truth * I wonder [if or not] he told me the truth * I think [that or not] he told me the truth CP Spec. CP is the perfect place to put phrasal ✓ whether C' C now the C-position remains empty again IP Ø DPi I' he I v. P -ed ti tell me the truth 115

Complementiser Phrases I wonder [why/when/how often] he told me the truth CP Spec. CP also comes in handy for other phrasal expressions beginning with wh- — so-called wh -phrases C' C IP Ø DPi I' he I v. P -ed ti tell me the truth 116

Complementiser Phrases 117 I wonder [why/when/how often] he told me the truth CP Spec C Ø for whether, we assumed that it is born in Spec. CP — it does not bind a trace lower down the tree C' what about these other wh-elements? for why and when, the answer is uncertain IP but how often must clearly bind a trace DPi I' he I v. P -ed ti tell me the truth

Complementiser Phrases 118 I wonder [how often] he t told me the truth I think he [very often] told me the truth CP Spec C' C Ø what about these other wh-elements? for why and when, the answer is uncertain IP but how often must clearly bind a trace DPi I' he I v. P -ed ti tell me the truth

Complementiser Phrases I wonder [how often] he t told me the truth I think he [very often] told me the truth CP how oftenj C' C an adjunct to v. P IP Ø DPi I' he I v. P -ed tj ti tell me the truth 119

Complementiser Phrases 120 I wonder [what] he told me t I think he told me [the truth] CP Spec C' C the need for a trace is of course even clearer when the wh-phrase serves as an argument of the verb IP Ø DPi I' he I v. P -ed ti tell me the truth

Complementiser Phrases 121 I wonder [what] he told me t I think he told me [the truth] CP whatk C' C the Projection Principle demands that the direct object position of tell, which tell assigns the Theme θ-role to, must be occupied at all levels of syntax IP Ø DPi Projection Principle structures are projected from the lexicon I' at all levels he I v. P -ed ti tell me tk

Complementiser Phrases I wonder [what] he told me t I think he told me [the truth] CP whatk C' C IP the trace not only helps us comply with the Projection Principle, it also allows the chain (whatk, tk) to satisfy the Case Filter Ø DPi I' he I v. P -ed ti tell me tk 122

Complementiser Phrases [what] did he tell me t ? he told me [the truth] CP whatk wh-questions are not always embedded under verbs like wonder: they can also occur as main clauses C' C IP DPi now C cannot remain empty I' he I v. P -ed ti tell me tk 123

Complementiser Phrases 124 [what] did he tell me t ? he told me [the truth] CP whatk I must raise to C, giving rise to ‘subject-auxiliary inversion’ — so called because only auxiliaries can undergo I-to-C C' C IP what has he told me t? I' what will he tell me t? DPi now C cannot remain empty *what told he me t? he I v. P -ed ti tell me tk

Complementiser Phrases 125 [what] did he tell me t ? he told me [the truth] CP whatk whenever there is no finite auxiliary in the sentence, the services of the dummy auxiliary do are called upon C' C when I does not raise to C, I can merge with v+V under adjacency IP *what told he me t? what has he told me t? I' what will he tell me t? DPi he I v. P -ed ti tell me tk

Complementiser Phrases [what] did he tell me t ? he told me [the truth] CP whatk whenever there is no finite auxiliary in the sentence, the services of the dummy auxiliary do are called upon C' C IP DPi I' he but when I raises to C, merger with v+V fails; I is spelled out as do+-ed I v. P -ed ti tell me tk 126

Complementiser Phrases [what] did he tell me t ? he told me [the truth] CP whatk C' C IP In C DPi did [Q] note that I adjoins to C rather than being substituted for C: C in interrogatives is not totally empty; it has the feature [Q], for ‘question’ he we can treat [Q] as a bound morpheme and use this as the trigger for I-to-C (p. 255) I' I v. P tn ti tell me tk 127

Complementiser Phrases did he tell me the truth? (yes, ) he told me the truth for so-called yes–no questions, we can now use the same analysis CP ? ? C' C IP In C DPi did [Q] plausibly, Spec. CP is occupied by a null operator (Op) he we can treat [Q] as a bound morpheme and use this as the trigger for I-to-C (p. 255) I' I v. P tn ti tell me the truth 128

Complementiser Phrases did he tell me the truth? (yes, ) he told me the truth CP Op … the silent counterpart to whether! C' C IP Ik C DPi did [Q] plausibly, Spec. CP is occupied by a null operator (Op) I' he I v. P tn ti tell me the truth 129

Complementiser Phrases • we can use null operators profitably in a range of other constructions as well, including relative clauses that do not have an overt wh-operator — the book (that) I read CP Op we will talk about this and many other things in the more advanced classes C' C IP Ik C DPi did [Q] plausibly, Spec. CP is occupied by a null operator (Op) I' he I v. P tn ti tell me the truth 130