Examples:
Example | Instruction stream (clickable) | Notes |
---|---|---|
(1) John saw Mary | [mary, d, see, 'v*', [john, d], 'Tpast', c] | Transitive sentence: v* values acc Case, has edge theta-position. |
(2) John runs | [run, 'v_unerg', [john, d], 'T', c] | Unergative sentence: v_unerg has edge theta-position. |
(3) A man arrives | [man, a, arrive, 'v~unacc', 'T' ,c] | Unaccusative sentence: v~unacc checks theta and has an edge feature. (See Sobin examples.) |
(4) There arrived a man | [man, a, arrive,'v~unacc', there, 'Tpast', c] | Expletive unaccusative construction: v~unacc checks theta and edge feature satisfied by merge of there. |
(5) John was arrested | [john, d, arrest, prt, 'v~', 'Tpast' ,c] | Passive sentence: v* replaced by v~ and prt. v~ checks theta. (See Sobin examples.) Passive participle prt has uCase and uPhi. (See Deriv by Phase examples.) |
(6) Who saw John? | [john, d, see, 'v*', [who, q], 'Tpast', c_Q] | Subject wh-question: CQ has uWh and uT simultaneously valued by corresponding features on who. (See Pesetsky & Torrego examples.) |
(7) What does John see? | [what:n, q, see, 'v*', [john, d], 'T', c_Q] | Object wh-question: CQ has uWh and uT valued from what and (head of) T, respectively. (See Pesetsky & Torrego examples.) |
(8) Who arrived? | [who, q, arrive,'v~unacc', 'Tpast', c_Q] | Unaccusative wh-question: CQ uWh and uT simultaneously valued by who. (See Pesetsky & Torrego examples.) |
Example | Instruction stream (clickable) | Notes |
---|---|---|
(4)(ii)(a) There are likely to be several
prizes awarded cf. There are likely to be awarded several prizes |
[prizes, several, award, prt, 'v~', there, 'Tinf', likely, v_be, 'T', c] | Expletive passive with raising verb. Leftwards TH/EX.
T values nom Case for prt and several prizes. Several prizes values uPhi for prt and T. |
(4)(ii)(b) Several prizes are likely to be awarded | [prizes, several, award, prt, 'v~', 'Tinf', likely, v_be, 'T', c] | Several prizes raises to matrix subject. |
(4)(iii)(a) We expect there to be several prizes awarded
We expect there to be awarded several prizes |
[prizes, several, award, prt, 'v~', there, 'Tinf', expect, 'v*', [we, d], 'T', c] | Expletive passive with exceptional Case Marking (ECM) transitive
verb. Leftwards TH/EX.
v* values acc Case for prt and several prizes. Several prizes values uPhi for prt and v*. |
(4)(iii) (b) We expect several prizes to be awarded | [prizes, several, award, prt, 'v~', 'Tinf', expect, 'v*', [we, d], 'T', c] | |
(15)(ii) There is likely to arrive a man | [man, a, arrive, 'v~unacc', there, 'Tinf', likely, 'v_be', 'T', c] | Raising verb with unaccusative. |
(16)(ii) We expect there to arrive a man | [man, a, arrive, 'v~unacc', there, 'Tinf', expect, 'v*', [we, d], 'T', c] | ECM transitive verb with unaccusative. |
(18)(a) There seem to have been several fish caught cf. There seem to have been caught several fish |
[fish, several, catch, prt, 'v~', there, perf, v, 'Tinf', seem, v_nop, 'T', c] | Raising verb with participial passive. |
(18)(b) We expect there to have been several fish caught cf. We expect there to have been caught several fish |
[fish, several, catch, prt, 'v~', there, perf, v, 'Tinf', expect, 'v*', [we, d], 'T', c] | ECM transitive verb with participial passive. |
(20)(b)(i) There were several fish believed to have been
caught cf. There were believed to have been several fish caught cf. There were believed to have been caught several fish |
[fish, several, catch, prt, 'v~', perf, v, 'Tinf', believe, prt, 'v~', there, 'Tpast', c] | Passivized ECM verb with participial passive. |
(20)(b)(ii) We expected there to have been several fish
believed to have been caught cf. We expected there to have been believed to have been several fish caught cf. We expected there to have been believed to have been caught several fish |
[fish, several, catch, prt, 'v~', perf, v, 'Tinf', believe, prt, 'v~', there, perf, v, 'Tinf', expect, 'v*', [we, d], 'Tpast', c] | ECM transitive verb with passivized ECM verb with participial passive. |
(21)(b) There is expected to arrive a man (21)(c) *A man is expected there to arrive |
[man, a, arrive, 'v~unacc', 'Tinf', expect, prt, 'v~', there, 'T', c] | Passive counterpart of (16)(ii).
Converges as: There is a man expected to arrive |
(22)(c) *There was placed a large book on the
table There was a large book placed on the table |
[table, the, on, [book, large, a], place, prt, 'v~', there, 'Tpast', c] | Leftwards TH/EX. Converges as There was a large book placed on the table |
(22)(e) *How many packages were there placed on the table? | [table, the, on, [packages, many, how], place, prt, 'v~', there, 'Tpast', c_Q] | Derivation converges. |
(24)(a) There were several packages placed on the table
cf. There were placed several packages on the table cf. (24)(b) There were placed on the table several packages |
[table, the, on, [packages, several], place, prt, 'v~', there, 'Tpast', c] | Leftwards TH/EX.
(24)(b) is rightwards TH/EX (not handled here). |
(31)(a) What are they selling books about? | [what:n, q, about, books, d, sell, 'v*', [they, d], prog, 'v~', 'T', c_Q] | No leftwards TH/EX.
2 derivations: optional prepositional pied-piping: About what are they selling books? |
(31)(b) *What are there books about being sold? | [what:n, q, about, books, d, sell, prt, 'v~', prog, 'v~', there, 'T', c_Q] | Wh-movement from extracted nominal (EN) in passive construction.
No leftwards TH/EX according to Chomsky. Derivation converges with books about raised to progressive be 2 derivations: optional prepositional pied-piping: About what are there books being sold? |
(38)(a) There are expected to be caught many fish
(38)(b) There are expected to be many fish caught (38)(c) There are many fish expected to be caught |
[fish, many, catch, prt, 'v~', 'Tinf', expect, prt, 'v~', there, 'T', c] | Converges as (38)(c) There are many fish expected to
be caught.
Note: in in DbyP, many fish expected to be caught in (38)(c) is analyzed as a reduced relative construction, and the complete sentence as existential there are DP. |
(38)(d) Many fish are expected to be caught | [fish, many, catch, prt, 'v~', 'Tinf', expect, prt, 'v~', 'T', c] |
Example | Instruction stream (clickable) | Notes |
---|---|---|
(5)(ii) Of which car did they find the driver?
Which car did they find the driver of? |
[car, which, of, driver, the, find, 'v*', [they, d], 'Tpast', c_Q] | Optional pied-piping: two derivations. |
(6)(ii) *Of which car did the driver cause a scandal? | [scandal, a, cause, 'v*', [car, which, of, driver, the],'Tpast', c_Q] | Subject island.
Crashes as CQ cannot access wh-DP which car. wh-DP on the stack is blocked by stack entry the driver of which car. Constraint: if [A .. [B ..]] is pushed onto the stack, and [B ..] from a substream is already on the stack, A subsumes B and renders B unavailable on the stack. [Implementation: a subconstituent check is performed whenever something is pushed on the stack. B is removed from the stack.] |
(7)(ii) Of which car was the driver awarded a prize?
Which car was the driver of awarded a prize? |
[car, which, of, driver, the, [prize, a, 'G2'], award, prt, 'v~', 'Tpast', c_Q] | Chomsky: Parallel extraction to edge of CQ and T.
Implementation: assume pair-merge for the driver of which car (object) and a prize (adjunct). Assume also that the adjunct doesn't pied-pipe here. [Normally pair-merged SOs do move together. G2 has feature no_pied_pipe, which blocks it from being placed on the stack.] CQ targets of which car T targets TOS the driver of which car. Optional pied-piping: two derivations. |
Example | Instruction stream (clickable) | Notes |
---|---|---|
(5a) What did Mary buy? (5b) *What Mary bought? |
[what:n, q, buy, 'v*', [mary, d], 'Tpast', c_Q] | (5a) T to matrix C => do-support. (5b) Mary to edge C blocked (FI: exclamatory). |
(5c) *Who did buy the book? (5d) Who bought the book? |
[book, the, buy, 'v*', [who, q], 'Tpast', c_Q] | No do-support triggered. (5c) ruled out by economy. |
(23) What will Mary buy? | [what:n, q, buy, 'v*', [mary, d], will, 'T', c_Q] | Auxilary will raises to edge of C. |
(24) What did John say that Mary will buy? | [what:n, q, buy, 'v*', [mary, d], will, 'T', c_eQ, say, v_unerg, [john, d], 'Tpast', c_Q] | Two derivations: (24) T to C realized as that. Or that can be omitted. c_eQ hosts wh-movement only. |
(28a) Who did John say will buy the book? (28b) *Who did John say that will buy the book? |
[book, the, buy, 'v*', [who, q], will, 'T', c_eQ, say, v_unerg, [john, d], 'Tpast', c_Q] | That-trace effect. (28b) ruled out by economy. |
(31) Mary thinks that Sue will buy the book
(33) Mary thinks Sue will buy the book |
[book, the, buy, 'v*', [sue, d], will, 'T', c_e, think, v_unerg, [mary, d], 'T', c] | Two derivations: (31) T to embedded C (c_e) realized as that. (33) c_e attracts subject Sue |
(43a) *What a silly book did Mary buy! (43b) What a silly book Mary bought! |
[book, silly, a, what:d, q, buy, 'v*', [mary, d], 'Tpast', c_Q] | (43a) T to C blocked by CI: exclamative. (43b) Mary in edge of C. |
(47a) Bill asked what Mary bought (47b) *Bill asked what did Mary buy (47c) *Bill asked what that Mary bought |
[what:n, q, buy, 'v*', [mary, d], 'Tpast', c_Qe, ask, v_unerg, [bill, d], 'Tpast', c] | Interrogative embedded C does not support T displacement. Note: same parse derived twice. |
Extra test cases | ||
Who was John meeting? | [who, q, meet, 'v*', [john, d], prog, 'v~', 'Tpast', c_Q] | Object wh-movement. Auxiliary be pied-piping. |
Who was meeting John? | [john, d, meet, 'v*', [who, q], prog, 'v~', 'Tpast', c_Q] | Subject wh-movement. Auxiliary be pied-piping. |
*Who do you think that saw John?
Who do you think saw John? |
[john, d, see, 'v*', [who, q], 'Tpast', c_eQ, think, 'v_unerg', [you, d], 'T', c_Q] | That-trace effect.
Economy: no T to C. |
Who do you think that John saw?
Who do you think John saw? |
[who, q, see, 'v*', [john, d], 'Tpast', c_eQ, think, 'v_unerg', [you, d], 'T', c_Q] | Two derivations. No that-trace effect with object. |
Example | Instruction stream (clickable) | Notes |
---|---|---|
(1c) There was a book taken from the shelf
*There was taken a book from the shelf |
[shelf, the, from, [book, a], take, prt, 'v~', there, 'Tpast', c] | TH/EX: a book raises to edge of PRT. |
(1d) There was a book being taken from the shelf
*There was being a book taken from the shelf ?There was being taken a book from the shelf |
[shelf, the, from, [book, a], take, prt, 'v~', prog, 'v~', there, 'Tpast', c] | TH/EX: a book raises to edge of progressive be. |
(4a) There was a book about the war taken from the shelf
*There was taken a book about the war from the shelf |
[shelf, the, from, [war, the, about, book, a], take, prt, 'v~', there, 'Tpast', c] | TH/EX: a book about the war raises to edge of PRT. |
(4b) *What was there a book about taken from the shelf
*What was there taken a book about from the shelf |
[shelf, the, from, [what:n, q, about, book, a], take, prt, 'v~', there, 'Tpast', c_Q] |
what cannot be extracted because it is formed in the
substream for a book about what and merged with
from.
By substream subsumption: what is deleted. |
(4c) *What was there taken from the shelf | [shelf, the, from, [what:n, q], take, prt, 'v~', there, 'Tpast', c_Q] | Derivation converges (incorrectly). [c_Q] is not blocked from targeting what. |
(13a) There is someone laughing
(10a) *There laughed someone |
[laugh, v_unerg, [someone, d], prog, 'v~', there, 'T', c] | Unergative laugh. Leftwards TH/EX not visible. |
(10b) *There watched someone a flying saucer
(13b) There is someone watching a flying saucer |
[saucer, flying, a, watch, 'v*', [someone, d], prog, 'v~', there, 'T', c] | TH/EX: someone raises to edge of progressive be |
(10c) *There fell a book on the floor
(13c) There is a book falling on the floor |
[book, a, fall, caus, prog, 'v~', there, 'T', c] | No visible TH/EX: assume caus selects for inchoative verbal root fall. |
(10d) There arrived a train
*There a train arrived |
[train, a, arrive, 'v~unacc', there, 'Tpast', c] | No TH/EX. |
(1b) There is a train arriving
*There is arriving a train |
[train, a, arrive, 'v~unacc', prog, 'v~', there, 'T', c] | TH/EX: a train raises to edge of progressive be |
(12) A train arrived | [train, a, arrive, 'v~unacc', 'Tpast', c] | |
(22) There was someone arrested
*There was arrested someone |
[someone, d, arrest, prt, 'v~', there, 'Tpast', c] | TH/EX: passive expletive. |
(23) There was someone being arrested
*There was being someone arrested *There was being arrested someone |
[someone, d, arrest, prt, 'v~', prog, 'v~', there, 'Tpast', c] | TH/EX: someone raises to edge of progressive be. |
(26a) There has arrived a train
(24a) *There has a train arrived |
[train, a, arrive, 'v~unacc', there, perf, v, 'T', c] | TH/EX: perfective expletive unaccusative. |
(26b) a train has arrived | [train, a, arrive, 'v~unacc', perf, v, 'T', c] | Perfective unaccusative. |
(17a) *There has someone laughed | [laugh, v_unerg, [someone, d], perf, v, there, 'T', c] | Derivation crashes. No place to merge there. |
(17b) Someone has laughed | [laugh, v_unerg, [someone, d], perf, v, 'T', c] | |
(17c) There has been someone laughing
*There has someone been laughing |
[laugh, v_unerg, [someone, d], prog, 'v~', there, perf, v, 'T', c] | TH/EX: someone raises to edge of progressive be |
(22a) There is a book available | [book, a, available, a_, 'v~', there, 'T', c] | v~ selects for stage-level predicate available. |
(22b) A book is available | [book, a, available, a_, 'v~', 'T', c] | |
(21a) One book was tiny | [[book, one], tiny, v_be, 'Tpast', c] | vbe directly selects for the individual-level predicate tiny. |
(21b) *?There was one book tiny | [[book, one], tiny, v_be, there, 'Tpast', c] | Does not converge. vbe has an edge that is filled preferentially from the stack. vbe merges TOS one book (not stream there). [Note: default policy is to prefer the stream over the stack, i.e. merge over move. The policy can be overridden from the lexicon.] |
(24a) There is a problem | [problem, a, be, v_ex, there, 'T', c] | [v_ex]: obligatory there-insertion. |
(24b) *A problem is | [problem, a, be, v_ex, 'T', c] | Derivation crashes. Obligatory there-insertion. |
(27a) *There was taken | [take, prt, 'v~', there, 'Tpast', c] | Derivation crashes. PRT is a probe but the stack is empty |
(27b) *There was taken a book | [book, a, take, prt, there, 'v~', 'Tpast', c] | Derivation crashes. In a passive expletive construction: TH/EX is obligatory. v~ finds non-theta there. |
(27c) There was a book taken | [book, a, take, prt, 'v~', there, 'Tpast', c] | TH/EX: passive expletive construction. a book raises to edge of PRT. |
(27d) A book was taken | [book, a, take, prt, 'v~', 'Tpast', c] | |
(51a) We believe there to be a book available | [book, a, available, a_, 'v~', there, 'Tinf', believe, 'v*', [we, d], 'T', c] | |
(51b) There seems to be a book available | [book, a, available, a_, 'v~', there, 'Tinf', seem, v_nop, 'T', c] | |
(51c) A book seems to be available | [book, a, available, a_, 'v~', 'Tinf', seem, v_nop, 'T', c] |
Example | Instruction stream (clickable) | Notes |
---|---|---|
(4)(a) John1 thinks he1 is smart John1 thinks that he1 is smart |
[john, d, he, d, smart, v_be, 'T', c_e, think, v_unerg, 'T', c] | Doubling constituent (DC) he-John.
Local Extent (LE) boundary (b) at embedded CeP triggers stacking of John. vunerg picks up TOS John. Two derivations due to T-to-C movement (Pesetsky & Torrego, 2001). |
(5)(a) *John1 praises him1 | [john, d, he, d, praise, 'v*', 'T', c] | Does not converge.
DC he-John. John not on the stack, not available for theta-merge. (No LE boundary, no stacking triggered.) |
(6)(a) John1 praises himself1 | [john, d, he, self, praise, 'v*', 'T', c] | Determiner self heads a LE.
Completion of DC self-he-John triggers stacking of John. [Note: stack elements introduced by self are subject to the condition that they may only be stacked once, i.e. must be "used" (merged) in the current LE. See example (8a).] v* picks up TOS John. |
(7)(a) John1 thinks that Mary likes him1
John1 thinks Mary likes him1 |
[john, d, he, d, like, 'v*', [mary, d], 'T', c_e, think, v_unerg, 'T', c] | DC he-John.
LE boundary (b) at embedded CeP triggers stacking of John. vunerg merges TOS John to its edge. Two derivations due to T-to-C movement (Pesetsky & Torrego, 2001). |
(8)(a) *John1 thinks that Mary likes himself1 | [john, d, he, self, like, 'v*', [mary, d], 'T', c_e, think, v_unerg, 'T', c] | Does not converge.
LE boundary (b) at DC self-he-John stacking of John. [Stack elements introduced by self are subject to the condition that they may only be stacked once, i.e. must be "used" (merged) in the current LE.] Edge of v* preferentially merges Mary from the stream. At the next LE boundary, embedded CeP, John cannot be re-stacked, and falls behind the LE boundary (b). Search past a LE boundary (b) is not permitted. Derivation crashes since edge of matrix v* cannot see John on the stack. |
(29) John1 thinks that Peter thinks that Mary thinks that Bill likes him1 | [john, d, he, d, like, 'v*', [bill, d], 'T', c_e, think, v_unerg, [mary, d], 'T', c_e, think, v_unerg, [peter, d], 'T', c_e, think, v_unerg, 'T', c] | DC he-John.
CeP triggers stacking of John. v* preferentially merges from the stream: Mary and Peter will be preferred over John on the stack. At each intermediate LE boundary, CeP, John will be re-stacked in front of the boundary mark (b). Matrix vunerg picks up TOS John. (No competition from the stream at this point.) Eight derivations due to T-to-C movement for each sub-clause (Pesetsky & Torrego, 2001). |
(32) ?John thinks that Peter1 thinks that Mary thinks that Bill likes him1 | [peter, d, he, d, like, 'v*', [bill, d], 'T', c_e, think, v_unerg, [mary, d], 'T', c_e, think, v_unerg, 'T', c_e, think, v_unerg, [john, d], 'T', c] | DC he-Peter.
Derivation proceeds as in (29) above, except no competition from the stream for v* theta-merge in 1st embedded clause (cf. matrix v*). |
(38) John1 considers himself1 to be intelligent | [john, d, he, self, intelligent, v_be, 'Tinf', consider, 'v*', 'T', c] | DC self-he-John forms a LE boundary.
John is stacked (but cannot be restacked). Complement of ECM verb consider is not a CP, and thus not a LE boundary. Matrix v* picks up TOS John . |
(9)(a) *John1 considers him1 to be intelligent | [john, d, he, d, intelligent, v_be, 'Tinf', consider, 'v*', 'T', c] | Does not converge.
DC he-John. The complement of ECM verb consider is not a LE boundary. No stacking triggered. John not on stack, not available for theta-merge at matrix v*. |
(40) *John1 thinks heself1 is smart | [john, d, he, self, smart, v_be, 'T', c_e, think, v_unerg, 'T', c] | Does not converge.
DC self-he-John forms a LE boundary. John is stacked (but not permitted to be restacked). At the next LE boundary, CeP, John cannot be restacked, and falls behinds the boundary marker (b). Matrix vunerg fails to find John on the stack. |
(42)(a) John1 likes his1 dog | [dog, '\'s', [john, d, he, d], like, 'v*', 'T', c] | DC he-John.
's heads a LE. DP he-John's dog triggers stacking of John. Merge likes and he-John's dog triggers stacking of he-John's dog (uCase). [Note: if [A .. [B ..]] is pushed onto the stack, and [B ..] from a substream is already on the stack, A subsumes B and renders B unavailable on the stack. (See On Phases examples.) ] John is a subconstituent of he-John's dog but John, although formed in a substream, is only stacked visibly in the mainstream. Constraint does not apply. Matrix v* picks up TOS John. |
(42)(b) *John1 likes himself1's dog | [dog, '\'s', [john, d, he, self], like, 'v*', 'T', c] | DC self-he-John marks a LE boundary.
John is stacked (but not permitted to be restacked). 's heads the next LE boundary. Since John cannot be restacked, it falls behind the boundary marker (b) and is inaccessible to matrix v*. |
(45)(a) *Hannah1 found a picture of her1 | [hannah, d, she, d, of, picture, a, find, 'v*', 'Tpast', c] | Does not converge.
DC she-Hannah. Hannah not on the stack, not available for theta-merge. (No LE boundary, no stacking triggered.) |
(45)(b) Hannah1 found a picture of herself1 | [hannah, d, she, self, of, picture, a, find, 'v*', 'Tpast', c] | Determiner self heads a LE.
Completion of DC self-she-Hannah triggers stacking of Hannah. v* picks up TOS Hannah. |
(45)(c) *Hannah found Peter1's picture of him1 | [peter, d, he, d, of, picture, '\'s', find, 'v*', [hannah, d], 'Tpast', c] | Does not converge.
DC he-Peter. Peter not on the stack, not available for theta-merge. (No LE boundary, no stacking triggered.) |
(45)(d) Hannah found Peter1's picture of himself1 | [peter, d, he, self, of, picture, '\'s', find, 'v*', [hannah, d], 'Tpast', c] | Determiner self heads a LE.
DP self-he-Peter triggers stacking of Peter. 's edge theta-merges with TOS Peter. |
(45)(e) Hannah1 found Peter's picture of her1 | [hannah, d, she, d, of, picture, '\'s', [peter, d], find, 'v*','Tpast', c] | DC she-Hannah.
's heads a LE. DP Peter's picture of her-Hannah triggers stacking of Hannah. v* picks up TOS Hannah. |
(45)(f) Hannah1 found Peter's picture of herself1 | [hannah, d, she, self, of, picture, '\'s', [peter, d], find, 'v*','Tpast', c] | Does not converge.
Determiner self heads a LE. Completion of DC self-she-Hannah triggers stacking of Hannah. [Stack elements introduced by self are subject to the condition that they may only be stacked once, i.e. must be "used" (merged) in the current LE.] 's heads a LE Peter's picture of self-she-Hannah. Since Hannah cannot be restacked, it falls behind the boundary marker (b) and is inaccessible to matrix v*. |
Extra test cases | ||
(x1) John1 thinks he1 likes Mary
John1 thinks that he1 likes Mary |
[mary, d, like, 'v*', [john, d, he, d], 'T', c_e, think, v_unerg, 'T', c] | DC he-John.
LE boundary at CeP triggers stacking of John. vunerg picks up TOS John. Two derivations due to T-to-C movement (Pesetsky & Torrego, 2001). |
Example | Instruction stream (clickable) | Notes |
---|---|---|
(2)(a) I showed John1 himself1 in the mirror | [john, d, he, self, 'G', [mirror, the, in], show, 'v*', [i, d], 'Tpast', c] | Determiner self heads a Local Extent (LE).
Doubling constituent (DC) self-he-John triggers stacking of John. TOS John is merged to the edge of G. |
(2)(b) I showed John1 to himself1 in the mirror | [john, d, he, self, to:p, [mirror, the, in], show, 'v*', [i, d], 'Tpast', c] | (Same as (2a) but (2b) uses the lexical counterpart to G.) |
(3)(b) *I showed himself1 John1 in the mirror | [john, d, he, self, 'G', [mirror, the, in], show, 'v*', [i, d], 'Tpast', c] | Cannot be derived.
DC is he-John, cannot be John-he (John-he would result in stacking he and deriving Condition C violations.) Therefore the stream is the same as for (2a). |
(16)(a) *I showed John1 him1 in the mirror | [john, d, he, d, 'G', [mirror, the, in], show, 'v*', [i, d], 'Tpast', c] | Does not converge.
DC he-John. John not on the stack, not available for G theta-merge. (No LE boundary, no stacking triggered.) |
(16)(b) *I showed John1 to him1 in the mirror | [john, d, he, d, to:p, [mirror, the, in], show, 'v*', [i, d], 'Tpast', c] | Same analysis as in (16a). |
(4) Pictures of himself1 worry John1 | [john, d, he, self, of, pictures, d, caus_p, worry, caus_v, 'T', c] | Determiner self heads a LE.
Doubling constituent (DC) self-he-John triggers stacking of John. CAUSp does not assign Case to the DC object. TOS John theta-merges to the edge of CAUSp. CAUSv values ACC Case for John and also heads a LE. At CAUSvP, restacking of pictures of himself occurs. [Note: John is not subject to restacking. Stack elements introduced by self are subject to the condition that they may only be stacked once, i.e. must be "used" (merged) in the current LE.] T picks up TOS pictures of himself. [Note: A copy of John inside pictures of himself is raised to edge of T. However,John must be spelled out at edge of CAUSp.] |
(21) Pictures of him1 worry John1 | [john, d, he, d, of, pictures, d, caus_p, worry, caus_v, 'T', c] | Does not converge.
DC he-John. John not on the stack, not available for theta-merge to CAUSp's edge. (No LE boundary, no stacking triggered.) |
Example | Instruction stream (clickable) | Notes |
---|---|---|
(1) John is easy to please | [john, d, pro, please, 'v*', ['PRO', d0], 'Tinf', c, easy, v_be, 'T', c] |
Doubling Constituent (DC) pro-John.
At the Local Extent (LE) boundary, the embedded CP, John (with unvalued uCase) is re-stacked. T picks up TOS John. |
(2) The violin is easy to play the sonata on | [violin, the, pro, on, [sonata, the], play, 'v*', ['PRO', d0], 'Tinf', c, easy, v_be, 'T', c] |
DC pro-the-violin.
At the LE boundary, the embedded CP, the violin (with unvalued uCase) is re-stacked. Matrix T picks up TOS the violin. |
(3) *What sonata is the violin easy to play on? | [violin, the, pro, on, [sonata, what:d], play, 'v*', ['PRO', d0], 'Tinf', c, easy, v_be, 'T', c_Q] |
Does not converge.
DC pro-the-violin. what sonata is stacked (uScope). At the LE boundary, the embedded CP, what sonata (unvalued uScope) and the violin (unvalued uCase) are re-stacked. Matrix T fails to agree with TOS what sonata: nom Case clashes with acc Case assigned by embedded v*. |
(4) The sonata is easy to play on this violin | [violin, this, on, [sonata, the, pro], play, 'v*', ['PRO', d0], 'Tinf', c, easy, v_be, 'T', c] |
DC pro-the-sonata.
At the LE boundary, the embedded CP, the sonata (unvalued uCase) is restacked. Matrix T picks up TOS the sonata. |
(5) What violin is the sonata easy to play on? | [violin, what:d, on, [sonata, the, pro], play, 'v*', ['PRO', d0], 'Tinf', c, easy, v_be, 'T', c_Q] |
DC pro-the-sonata.
what violin is stacked (uScope). At the LE boundary, the embedded CP, the sonata (unvalued uCase) and what violin (unvalued uScope) are restacked. Matrix T picks up TOS the sonata. CQ finds iWh what violin. |
(6) Linguists are tough to please | [linguists, d, pro, please, 'v*', ['PRO', d0], 'Tinf', c, tough, v_be, 'T', c] |
(Same as (1).)
DC pro-linguists. At the LE boundary, the embedded CP, linguists (with unvalued uCase) is re-stacked. T picks up TOS linguists. |
(7) It is tough to please linguists | [linguists, d, please,'v*', ['PRO', d0], 'Tinf', c, tough, v_be, it, 'T', c] |
v* assigns Case to linguists.
(No restacking is done at the LE boundary, the embedded CP.) Expletive it merge to the edge of vbe. Agree(matrix T,it) obtains. |
(8) The book is easy to put on the table | [table, the, on, [book, the, pro], put, 'v*', ['PRO', d0], 'Tinf', c, easy, 'v_be', 'T', c] |
(Same as (4).)
DC pro-the-book. At the LE boundary, the embedded CP, the book (unvalued uCase) is restacked. Matrix T picks up TOS the book. |
(9) What table is the book easy to put on? | [table, what:d, on, [book, the, pro], put, 'v*', ['PRO', d0], 'Tinf', c, easy, 'v_be', 'T', c_Q] |
(Same as (5).)
what table is stacked (uScope). DC pro-the-book. At the LE boundary, the embedded CP, the book (unvalued uCase) and what table (unvalued uScope) are restacked. Matrix T picks up TOS the book. CQ finds iWh what table. |
Example | Instruction stream (clickable) | Notes |
---|---|---|
(1) John is easy to please | [john, d, please, 'v_unerg', ['PRO', d0], 'Tinf', c, easy, v_be, 'T', c] |
John (with unvalued uCase) is stacked after merge with please.
vunerg doesn't assign Case to John. Tinf assigns null Case to PRO. At the Local Extent (LE) boundary, the embedded CP, John (with unvalued uCase) is re-stacked. T picks up TOS John. |
(2) The violin is easy to play the sonata on | [violin, the, on_nc, [sonata, the], play, 'v*', ['PRO', d0], 'Tinf', c, easy, v_be, 'T', c] |
onnc (no case) selects for the
violin but does not assign it Case.
At the Local Extent (LE) boundary, the embedded CP, the violin (with unvalued uCase) is re-stacked. Matrix T picks up TOS the violin. |
(3) *What sonata is the violin easy to play on? | [violin, the, on_nc, [sonata, what:d], play, 'v*', ['PRO', d0], 'Tinf', c, easy, v_be, 'T', c_Q] |
Does not converge.
onnc (no case) selects for the violin but does not assign it Case. what sonata is stacked (uScope). At the LE boundary, the embedded CP, what sonata (unvalued uScope) and the violin (unvalued uCase) are re-stacked. Matrix T fails to agree with TOS what sonata: nom Case clashes with acc Case assigned by embedded v*. |
(4) The sonata is easy to play on this violin | [violin, this, on, [sonata, the], play, 'v_unerg', ['PRO', d0], 'Tinf', c, easy, v_be, 'T', c] |
on assigns Case to this violin.
vunerg doesn't assign Case to the sonata. At the LE boundary, the embedded CP, the sonata (unvalued uCase) is restacked. Matrix T picks up TOS the sonata. |
(5) What violin is the sonata easy to play on? | [violin, what:d, on, [sonata, the], play, v_unerg, ['PRO', d0], 'Tinf', c, easy, v_be, 'T', c_Q] |
on assigns Case to this violin.
vunerg doesn't assign Case to the sonata. At the LE boundary, the embedded CP, the sonata (unvalued uCase) and what violin (unvalued uScope) are restacked. Matrix T picks up TOS the sonata. CQ finds iWh what violin. |
(6) Linguists are tough to please | [linguists, d, please, v_unerg, ['PRO', d0], 'Tinf', c, tough, v_be, 'T', c] |
(Same as (1).)
Linguists (with unvalued uCase) is stacked after merge with please. vunerg doesn't assign Case to linguists. Tinf assigns null Case to PRO. At the LE boundary, the embedded CP, linguists (with unvalued uCase) is re-stacked. T picks up TOS linguists. |
(7) It is tough to please linguists | [linguists, d, please,'v*', ['PRO', d0], 'Tinf', c, tough, v_be, it, 'T', c] |
v* assigns Case to linguists.
(No restacking is done at the LE boundary, the embedded CP.) Expletive it merge to the edge of vbe. Agree(matrix T,it) obtains. |
(8) The book is easy to put on the table | [table, the, on, [book, the], put, v_unerg, ['PRO', d0], 'Tinf', c, easy, 'v_be', 'T', c] |
(Same as (4).)
on assigns Case to the table. vunerg doesn't assign Case to the book. At the LE boundary, the embedded CP, the book (unvalued uCase) is restacked. Matrix T picks up TOS the book. |
(9) What table is the book easy to put on? | [table, what:d, on, [book, the], put, v_unerg, ['PRO', d0], 'Tinf', c, easy, 'v_be', 'T', c_Q] |
(Same as (5).)
on assigns Case to what table. vunerg doesn't assign Case to the book. At the LE boundary, the embedded CP, the book (unvalued uCase) and what table (unvalued uScope) are restacked. Matrix T picks up TOS the book. CQ finds iWh what table. |
Example | Instruction stream (clickable) | Notes |
---|---|---|
(14a) the book that I read (14c) the book I read |
[book, d_rel, read, 'v*', [i, d], 'Tpast', c_rel, the] | Two derivations.
Drel has Rel but cannot check book's D feature. Crel has T and Rel, and heads a Local Extent (LE). Two ways to satisfy Crel using Pesetsky and Torrego (2001): (1) T to edge of Crel => that (2) Nom Case-marked I to edge of Crel => empty At the LE boundary, book (uninterpretable D feature) is restacked. TOS book merges with CrelP. book labels (head merging with non-head). Stream the merges with SO headed by book. |
(14b) the book which I read *the book which that I read |
[book, which_rel, read, 'v*', [i, d], 'Tpast', c_rel, the] | Single derivation.
whichrel has Rel. whichrel can check T but not D feature. Crel heads a LE, and has T and Rel. Internal merge of whichrel book will satisfy both T and Rel on Crel. (By economy, T to C is not available.) At the LE boundary, book (uninterpretable D feature) is restacked. book merges with CrelP, then with the. |
(16a) the man to whom I talked
the man whom I talked to *the man whom that I talked to |
[man, who_rel, to, talk, 'v*', [i, d], 'Tpast', c_rel, the] | Two derivations.
whorel has Rel. whorel can check T but not D feature. Crel heads a LE, and has T and Rel. Internal merge of whorel man will satisfy both T and Rel on Crel. (By economy, T to C is not available.) At the LE boundary, man (uninterpretable D feature) is restacked. man merges with CrelP, then with the. Optional pied-piping of verb particle to. |
(16b) *The man to that I talked (16c) *The man to I talked The man I talked to The man that I talked to |
[man, d_rel, to, talk, 'v*', [i, d], 'Tpast', c_rel, the] |
Two derivations.
Covert Drel has Rel but cannot check man's D feature. Crel has T and Rel, and heads a LE. Two ways to satisfy Crel using Pesetsky and Torrego (2001): (1) T to edge of Crel => that (2) Nom Case-marked I to edge of Crel => empty At the LE boundary, man (uninterpretable D feature) is restacked. TOS man merges with CrelP. man labels (head merging with non-head). Stream the merges with SO headed by man. [Note: no pied-piping of verb particle to with Drel-man. [An] EC [Empty Category] disallows pied-piping (Chomsky, 2001:28)] |
(22) The man who John saw *The man who that John saw |
[man, who_rel, see, 'v*', [john, d], 'Tpast', c_rel, the] |
One derivation.
whorel has Rel. whorel can check T but not D feature. Crel heads a LE, and has T and Rel. Internal merge of whorel man will satisfy both T and Rel on Crel. (By economy, T to C is not available.) At the LE boundary, man (uninterpretable D feature) is restacked. man merges with CrelP, then with the. |
(29) The man who loves Mary *The man who that loves Mary |
[mary, d, love, 'v*', [man, who_rel], 'T', c_rel, the] |
One derivation.
whorel has Rel. whorel can check T but not D feature. The subject whorel-man is formed in a substream. man (uninterpretable D feature) will be initially stacked in the substream and transferred back to the main stream. Crel heads a LE, and has T and Rel. Internal merge of whorel man will satisfy both T and Rel on Crel. (By economy, T to C is not available.) At the LE boundary, man (uninterpretable D feature) is restacked. man merges with CrelP, then with the. [Note: under machine parameter setting substream(subsume_ss), main stream subject who-man (uCase) will subsume man (uninterpretable D feature) from a substream on the stack. To prevent this, tentatively we implement the policy of letting heads through.] |
(35) The boy *(that) called Mary | [mary, d, call, 'v*', [boy, d_rel], 'Tpast', c_rel, the] |
Covert Drel has Rel but cannot check man's D feature.
Crel has T and Rel, and heads a LE. Assume Drel cannot check T, i.e. Drel-boy, despite getting nom Case, fails to satisfy T on Crel. Then Crel's T can only be satisfied by moving T to the edge of Crel => that. |
Extra test cases | ||
(x1) What I read *What that I read |
[pro_n, what_rel, read, 'v*', [i, d], 'Tpast', c_rel, d] | Headless relative.
One derivation. Assume what is whatrel + pron (empty pronominal). whatrel has Rel. whatrel can check T but not D feature. Crel heads a LE, and has T and Rel. Internal merge of whatrel pron will satisfy both T and Rel on Crel. (By economy, T to C is not available.) At the LE boundary, pron (uninterpretable D feature) is restacked. pron merges with CrelP, then with d. |
(x2) What annoys John *What that annoys John |
[john, d, annoy, 'v*', [pro_n, what_rel], 'T', c_rel, d] | Headless relative.
One derivation. Assume what is whatrel + pron (empty pronominal). whatrel has Rel. whatrel can check T but not D feature. Crel heads a LE, and has T and Rel. Internal merge of whatrel pron will satisfy both T and Rel on Crel. (By economy, T to C is not available.) At the LE boundary, pron (uninterpretable D feature) is restacked. pron merges with CrelP, then with d. |
Example | Instruction stream (clickable) | Notes |
---|---|---|
(15a) The picture of himself1 that
Bill1 likes The picture of himself1 Bill1 likes |
[bill, d, he, self, of, picture, d_rel, like, 'v*', 'T', c_rel, the] | Two derivations.
Doubling Constituent (DC) self-he-Bill object of preposition of. Object of likes is Drel-picture-of-DC. Covert Drel has Rel but cannot check picture's D feature. Bill restacked at Local Extent (LE) boundary headed by self. v* merges TOS Bill to its edge. Crel has T and Rel, and heads a LE. Two ways to satisfy Crel using Pesetsky and Torrego (2001): (1) T to edge of Crel => that (2) Nom Case-marked Bill to edge of Crel => empty At the LE boundary, picture of self-he-Bill (uninterpretable D feature) is restacked. Head picture of the TOS picture of self-he-Bill merges with CrelP. picture labels (head merging with non-head). Stream the merges with SO headed by picture. [Spellout Note: Bill has been separately moved to the subject of the embedded clause. Thus although Drel pictures of self-he- |
(15b) The picture of Bill1 that he1
likes
The picture of Bill1 he1 likes |
[he, d, bill, d, of, picture, d_rel, like, 'v*', 'T', c_rel, the] |
Bill-he is not a regular DC (cf. he-Bill).
[Note: in a regular DC, e.g. he-John, the pronominal he triggers DC merge in which John will be stacked but not made visible to stack operations such as probe-goal and internal merge until a LE boundary is encountered later on in the derivation.] he (uCase) is (visibly) stacked when merged with Bill. Covert Drel has Rel but cannot check picture's D feature. v* merges he to its edge. Crel has T and Rel, and heads a LE. Crel satisfied by either T or he (locally nom Case-marked) to C. At the LE boundary, picture of Bill-he (uninterpretable D feature) is restacked. Extract head picture (uninterpretable D feature) and merge with CrelP. picture labels (head merging with non-head). Stream the merges with SO headed by picture. |
(19a) Pictures of himself1 are hard for John1 to like | [john, d, he, self, of, pictures, d, pro, like, 'v*', 'Tinf', for, hard, 'v_be', 'T', c] |
Doubling Constituent (DC) pro-pictures of
self-he-John with another DC
self-he-John embedded inside.
Determiner self heads a Local Extent (LE) phrase. Doubling constituent (DC) self-he-John triggers stacking of John. Embedded v* merges TOS John to its edge. Tinf does not value case for John. Assume for is a complementizer that probes and values Case. Then John's uCase feature is valued. Assume for heads a LE phase. Then at the LE boundary, pictures of self-he-John (with unvalued uCase) is re-stacked. Matrix vbe merges TOS pictures of self-he-John to its edge. [Spellout Note: John has been separately moved to the subject of the embedded clause. Thus although pictures of self-he- |
(19b) Pictures of John1 are hard for him1 to like | [he, d, john, d, of, pictures, d, pro, like, 'v*', 'Tinf', for, hard, 'v_be', 'T', c] |
DC pro-pictures of John-he with a (pseudo) DC
John-he embedded inside.
John-he is not a regular DC (cf. he-John). In particular, he (with unvalued uCase) is stacked when merged with John. [Note: in a regular DC, e.g. he-John, the pronominal he triggers DC merge in which John will be stacked but not made visible to stack operations such as probe-goal and internal merge until a LE boundary is encountered later on in the derivation.] Regular DC head pro merges with pictures of John-he. pictures of John-he is stacked but will not be available until the boundary of a LE has been built. Embedded v* merges non-TOS he to its edge. [Note: TOS pictures of John-he is not selected because it already has a theta-role assigned. Edge of v* is a theta-position.] Assume for is a complementizer that probes and values Case. Then he's uCase feature is valued. Assume for heads a LE phase. Then at the LE boundary, pictures of he-John (with unvalued uCase) is re-stacked. Matrix vbe merges TOS pictures of he-John to its edge. |
Extra test cases | ||
(x1) *The picture of him1 that Bill1 likes | [bill, d, he, d, of, picture, d_rel, like, 'v*', 'T', c_rel, the] |
Doubling Constituent (DC) he-Bill.
Bill is not restacked in time for merge to the edge of v* (no LE boundary). |
(x2) *Pictures of him1 are hard for John1 to like | [john, d, he, d, of, pictures, d, pro, like, 'v*', 'Tinf', for, hard, 'v_be', 'T', c] | DC he-John and DC pro-pictures of
he-John.
John is not restacked in time for merge to the edge of v* (no LE boundary). |
Example | Instruction stream (clickable) | Notes |
---|---|---|
Strong Crossover | ||
Who1 thinks (that) Mary likes him1 | [who, q, he, d, like, 'v*', [mary, d], 'T', c_eQ, think, v_unerg, 'T', c_Q] | Doubling Constituent (DC) he-who object of embedded verb like.
Two derivations. [Note: Need to allow c stack search to find (inactive) who.] |
Who1 thinks (that) Mary likes him2 | [he, d, like, 'v*', [mary, d], 'T', c_e, think, v_unerg, [who, q], 'T', c_Q] | No DC.
He is object of embedded verb like. Who is subject of matrix verb think. Two derivations. |
*Who1 does he1 think Mary likes | No DC analysis possible. | |
Who1 does he2 think (that) Mary likes | [who, q, like, 'v*', [mary, d], 'T', c_eQ, think, v_unerg, [he, d], 'T', c_Q] | No DC.
Who object of embedded verb like. He subject of matrix verb think. Two derivations. |
Example | Instruction stream (clickable) | Notes |
---|---|---|
ما بيروحوا ma b-yruuH-u neg indic.imperf go.they "They don't go (out)" |
[ruuH, v_unerg, [pro_henna,d], t_imperf, indic, neg, c] | (Brustad 2000:122) |
تعلمت اللغة من جيراننا tEallm-t el-lugha min jiiran-na learned.1st.sg. the-language.f from neighbors.m-our I learned the language from our neighbor |
[lugha, el, [jiiraan, poss, [niHna,d], min], tEallm, v_trans, [pro_ana,d], t_perf, c] | Pair-merge. |
بيروحو بهالعتمة b-y-ruuH-u bi-ha- el-ʕatma indic-go.3rd.pl.m in- the-darkness.f |
[aEtma, el, ha, bi, ruuH, v_unerg, [pro_henna,d], t_imperf, indic, c] | Brustad 2000:118 |
ما يشوفوا العروس لليلة العرس maa y-shuuf-u el-ʕaruus la-lelt el-ʕars neg see.3rd.pl.m the-bride.f until-night.f the-wedding.m "they used to not see the bride until the night..." |
[aEruus, el, [aErs, el, gen, lelt, d, la], shuuf, v_trans, [pro_henna, d], t_imperf, neg, kaan, c] | This is currently the abbreviated version of the full phrase kaan-u maa y-shuuf-u el-ʕaruus la-lelt el-ʕars.
Was.3rd.pl.m neg. see.3rd.pl.m the-bride.f until-night.f
of-the-wedding.m
Pair-merge. |
جاب له صحن بوظة zhaab l-u SaHn buuDHa brought.3rd.sg.m to-him plate.m (of).ice-cream.f he brought him a plate of ice cream |
[buuDHa, d, gen, saaHn, d, [huwa, d, l], zhaab, v_trans, [pro_huwa, d], t_perf, c] | A phrase from the full sentence "zhaab l-u SaHn buuDHa Saar y-aakl-u bi-el-khebez" he brought him a plate of ice cream, he began to eat it with bread. Pair Merge with spliced PP. |
يكونوا عم بيتمشّوا بالشارع y-kuun-u ʕam b-y-tmashh-u bi-el-shaarʕa was.3rd.pl.m prog. indic-walk.3rd.pl.m in-the-street.m ...they'd be walking in the street |
[shaarEa, el, bi, tmashh, v_unerg,[pro_henna,d], t_imperf, indic, 'Eam', kaan, c] | Brustad 2000:216 |
الضار ما كبيرش
eD-Dar maa kbir-a-sh the-house.f.sg. not big.f.-not the house is not big |
[dar, el, kbira, t, neg, c] |
In some dialects, e.g. Moroccan and Jordanian, negation is split
into maa + -sh.
-sh affix hops over the predicate root. |
Example | Instruction stream (clickable) | Notes |
---|---|---|
(1a) 鬼が花子を食った
oni-ga hanako-o kuu-ta ogre-NOM hanako-ACC eat-PAST (1b) 花子を鬼が食った hanako-o oni-ga kuu-ta hanako-ACC ogre-NOM eat-PAST "The ogre ate Hanako" |
[hanako, d, kuu, 'v*', [oni, d], 'Tpast', c] | Basic subject/object scrambling.
Allow internal Merge to edge of T from either subject or object position. |
(2a) 花子がサムにメールをした
hanako-ga samu-ni mēru-o suru-ta hanako-NOM Sam-DAT mail-ACC do-PAST (2b) メールを花子がサムにした mēru-o hanako-ga samu-ni suru-ta mail-ACC hanako-NOM Sam-DAT do-PAST (2c) サムに花子がメールをした samu-ni hanako-ga mēru-o suru-ta Sam-DAT hanako-NOM mail-ACC do-PAST "Hanako sent Sam an email" |
[samu, d, suru, v, ['mēru', d], 'v*', [hanako, d], 'Tpast', c] | Ditransitive frame. No FocP. v values dative Case.
Base order: Agent Theme Goal (Assume ni-marked Goal is lower than Theme.) Undergenerates: only 3 permutations possible: (a) Agent Theme Goal, (b) Theme Agent Goal, and (c) Goal Agent Theme. |
(3a) 花子がメールをサムにした
hanako-ga mēru-o samu-ni suru-ta hanako-NOM mail-ACC Sam-DAT do-PAST (3b) メールを花子がサムにした mēru-o hanako-ga samu-ni suru-ta mail-ACC hanako-NOM Sam-DAT do-PAST (3c) サムに花子がメールをした samu-ni hanako-ga mēru-o suru-ta Sam-DAT hanako-NOM mail-ACC do-PAST (3d) 花子がサムにメールをした hanako-ga samu-ni mēru-o suru-ta hanako-NOM Sam-DAT mail-ACC do-PAST (3e) メールを花子がサムにした mēru-o hanako-ga samu-ni suru-ta mail-ACC hanako-NOM Sam-DAT do-PAST (3f) サムにメールを花子がした samu-ni mēru-o hanako-ga suru-ta Sam-DAT mail-ACC hanako-NOM do-PAST (3g) 花子がサムにメールをした hanako-ga samu-ni mēru-o suru-ta hanako-NOM Sam-DAT mail-ACC do-PAST (3h) メールをサムに花子がした mēru-o samu-ni hanako-ga suru-ta mail-ACC Sam-DAT hanako-NOM do-PAST (3i) サムに花子がメールをした samu-ni hanako-ga mēru-o suru-ta Sam-DAT hanako-NOM mail-ACC do-PAST "Hanako sent Sam an email" |
[samu, d, suru, v, ['mēru', d], 'v*', [hanako, d], foc, 'Tpast', c] | Ditransitive frame. FocP with EF above v*P.
Base order: Agent Theme Goal (Assume ni-marked Goal is lower than Theme.) Overgenerates: 9 permutations, only 6 distinct: initial 3 same as (2a-c). (3d)=(3g), (3b)=(3e), (3c)=(3i) |
Example | Instruction stream (clickable) | Notes |
---|---|---|
(1a) Man mehmun davat kard-am I guest invitation do-1.sg I invited a guest |
[mehmun,d,[davat,'V'],'v*',[man,d],kard,'Tpast',c] | Assume an empty verbalizer 'V' merges with nominal 'davat'
(invitation) in morphology
space. In narrow syntax,
[V[V][N davat]] will be
treated as an atomic head.
Assume required 'kard' (do + PST) heads its own projection and is a 'v'. T's uPhi will be valued by the subject and surfaces as '-am' [1.sg]. |
(1b) U mehmun davat kard-∅ He/she guest invitation do-3.sg He/she invited a guest |
[mehmun,d,[davat,'V'],'v*',[u,d],kard,'Tpast',c] | ∅ = zero affix for [3.sg]. |
(1c) Anha mehmun davat kard-and They guest invitation do-3.pl They invited a guest |
[mehmun,d,[davat,'V'],'v*',[anha,d],kard,'Tpast',c] | |
(2) Man mehmun davat mi-kon-am I guest invitation DUR-do-1.sg I invite a guest |
[mehmun,d,[davat,'V'],'v*',[man,d],kard,dur,'T',c] | DUR = durative. DUR spells out as the prefix 'mi-'.
'kard' (do) + PRES = 'kon'. |
(3) Man dar khorshild-am mehmun davat mi-kon-am I to house-my guest invitation DUR-do-1.sg I invite a guest to my house |
[mehmun,d,[davat,'V'],[khorshid,poss1sg,dar],'v*',[man,d],kard,dur,'T',c] | Pair-merge adjunct PP 'dar khorshid-am'. |
(4) Man dar khorshild-am mehmun ro davat mi-kon-am I to house-my guest SPF invitation DUR-do-1.sg I invite a (specific) guest to my house |
[mehmun,d,ro,[davat,'V'],[khorshid,poss1sg,dar],spf,'v*',[man,d],kard,dur,'T',c] | Determiner "ro" indicates specificity.
Specific-marked objects must raise outside of vP. Head SPF hosts the final position of specific-marked objects in its edge. Pair-merge adjunct PP 'dar khorshid-am'. |
(5) mehmun ro guest-SPF a (specific) guest |
[mehmun,d,ro] | Assume specificity determiner "ro" is independent from d, and doesn't stack.
Determiner "ro" has an edge feature. As a result: inner DP raises to the edge of "ro". |
Example | Instruction stream (clickable) | Notes |
---|---|---|
(32a) the boy who told the story | [story, the, tell, 'v*', [boy, who_rel], 'Tpast', c_rel, the] | A subject relative clause.
Subject who boy attracted to edge of Crel, i.e. C with an Rel. DP who boy is marked with an Rel feature, originating on who. Note, whorel is defective, i.e. it doesn't value D feature on boy. Instead, boy raises from edge of C and its D feature is checked by the. |
(32b) the letter which Dick wrote yesterday | [letter, which_rel, write, 'v*', [dick, d], 'Tpast', [yesterday, d], c_rel, the] | An object relative clause.
Assume, substantially following (Haumann, 2007) and (Larson, 1985), that yesterday is a TP-level bare NP adverb. As an adjunct, yesterday is pair merged at TP. Defective whichrel is attracted by Crel; the entire DP which letter raising to the edge of C, see notes for (32a) above. Finally, letter within the DP raises and its D feature is valued by the. |
(32c) the man who Ann gave the present to
the man to who Ann gave the present (pied-piping) |
[man, who_rel, to, [present, the], give, 'v*', [ann, d], 'Tpast', c_rel, the] | Relativization of the indirect object of a double object
verb.
Two analyses, also produces pied-piped variant. Crel attracts who man, headed by whorel, to its edge. Defective who, can't value D on man. man raises. The uninterpretable D feature on man is valued by c-commanding the. |
(32d) the box which Pat brought the apples in
the box in which Pat brought the apples (pied-piping) |
[box, which_rel, in, [apples, the], bring, 'v*', [pat, d], 'Tpast', c_rel, the] | Assume bring...in involves a dyadic preposition
in.
Crel attracts which box, headed by whichrel, to its edge. Defective which, can't value D on box. box raises. D on box valued by c-commanding the. |
(32e) the dog which was taught by John | [dog, which_rel, teach, [[john, d], by], prt, 'v~', 'Tpast', c_rel, the] | Relativization of passivized object.
The agentive by-phrase is pair-merged to teach (not PRT or v~ due to merge over move preference). Crel attracts which dog, headed by whichrel, to its edge. Defective which, can't value D on dog. dog raises. D on box valued by c-commanding the. |
(33a) the boy who Mike writes better than
cf. *the boy than who Mike writes better |
[boy, who_rel, than, better, write, 'v_unerg', [mike, d], 'T', c_rel, the] | Relativization of the object of a comparative.
Assume than is a preposition, not a C - as needed in the case of Comparative Deletion (not implemented here: e.g. as in Mike writes better than I do/write. The pied-piped counterpart (see cf.) is blocked by a spellout rule (after the parse has been completed). Crel attracts who boy, headed by whorel, to its edge. Defective who, can't value D on boy. boy raises. D on boy valued by c-commanding the. |
(33b) the girl whose friends bought the cake | [cake, the, buy, 'v*', [friends,'\'s',[girl, who_rel]],'Tpast',c_rel,the] |
Relativization of a genitive from a subject.
Assume the entire possessive 's DP Defective whorel cannot value D on girl. This permits girl to raise out and its D feature is valued by the. Finally assume spellout rule: who + 's -> whose. |
(33b') the girl who friends of bought the cake
cf. *the girl of who friends bought the cake |
[cake, the, buy, 'v*', [girl, who_rel, of, friends, d], 'Tpast', c_rel, the] |
Relativization of a genitive from a subject.
Assume DP Crel attracts who girl, headed by whorel, to its edge. Defective who, can't value D on girl. This permits girl to raise. D on girl is valued by c-commanding the. The pied-piped counterpart is blocked by a spellout rule that asserts there is no spellout for whorel. |
(33c) the man whose house Patrick bought
the man whose house that Patrick bought (T-to-C variant) |
[house,'\'s',[man,who_rel],buy,'v*',[patrick,d],'Tpast',c_rel,the] |
Relativization of a genitive from the object.
Assume possessive 's DP This permits man to raise out and its D feature is valued by the. Finally assume spellout rule: who + 's -> whose. |
(33c') the man who Patrick bought the house of
cf. *the man of who Patrick bought the house |
[man, who_rel, of, house, the, buy, 'v*', [patrick, d], 'Tpast', c_rel, the] |
Relativization of a genitive from the object.
Assume DP Crel attracts who man, headed by whorel, to its edge. Defective who, can't value D on man. This permits man to raise. D on man is valued by c-commanding the. The pied-piped counterpart (see cf.) is blocked by a spellout rule (after the parse has been completed). See also (33b'). |
(33d) the boy whose brother was taught by Sandra | [brother, '\'s', [boy, who_rel], teach, [[sandra, d], by], prt, 'v~', 'Tpast', c_rel, the] |
Relativization of a genitive from the passivized object.
Assume possessive DP whorel cannot value D on boy. This permits boy to raise out and its D feature is valued by the. Finally assume spellout rule: who + 's -> whose. |
Example | Instruction stream (clickable) | Notes |
---|---|---|
(23a) the book which inspired them (Example (24a); Sag 1997: 448
(23b) *the book which that inspired them |
[they,d,inspire,'v*',[book,which_rel],'Tpast',c_rel,the] | Subject relative. DP [book, whichrel] is
attracted to the edge of Crel by Rel on
Crel. By economy, whichrel values
both T and Rel on Crel, and the T-to-C option is
impossible, i.e. *which that in (23b).
However, whichrel cannot value D so book will raise further and D is valued during the merge with the. |
(24) the person whose mother died (Example (24b); Sag 1997: 448)
|
[mother, '\'s', [person, who_rel], die, 'v~unacc', 'Tpast', c_rel, the] |
Assume whose here is whorel +
's and die is unaccusative,
i.e. [[whorel person]['s mother]] is
theta-merged at the internal argument position.
Furthermore, permit Rel on Crel to find Rel on whorel at the edge of possessive 's. [[whorel person]['s mother]] is attracted to the edge of Crel. However, whorel cannot value D on person. Next, person raises and its D is valued when merged with the. |
(25) the person whose mother's dog died (Example (24d); Sag 1997: 448) | [dog, '\'s', [mother, '\'s', [person, who_rel]], die, 'v~unacc', 'Tpast', c_rel, the] | Analysis is the same as for (24) above, except we permit Rel on Crel to recursively look for Rel at the edge of possessive 's. Internal argument is [[[whorel person] 's mother]['s dog]]. |
(26a) the person whose mother's dog we were (all) fond of (Example (24d); Sag 1997: 448)
(26b) the person whose mother's dog that we were (all) fond of (26c) *the person of whose mother's dog we were (all) fond (26d) *the person of whose mother's dog that we were (all) fond |
[dog, '\'s', [mother, '\'s', [person, who_rel]],of,fond,[we,d],v_be,'Tpast',c_rel,the] |
[[[whorel person] 's mother]['s dog]] is
attracted to the edge of Crel by Rel on
Crel. We assume, following (25), that Rel on
Crel can find recursively Rel at the edge of
possessive 's.
whorel cannot value D on person. Next, person raises and its D is valued when merged with the. (26b) is predicted as [[[whorel person] 's mother]['s dog]] is not a nominative subject, so economy does not apply. The option of T-to-C generates the complementizer that. (26c-d) are blocked by a spellout rule *of who, designed to block *the girl of who friends bought the cake. |
(27a) the person to whom they dedicated the building (Example (24e); Sag 1997: 448)
(27b) the person whom they dedicated the building to |
[person,who_rel,to,[building,the],dedicate,'v*',[they,d],'Tpast',c_rel,the] |
We assume dyadic to, i.e. the VP is [dedicate [[the
building] [to [whorel
person]]]]. [whorel person] is attracted to the
edge of Crel by Rel on Crel.
whorel cannot value D on person. Next,
person raises and its D is valued when merged with
the.
Pied-piping is permitted as whorel is overt. Note: after the two successful derivations, at step 41, the system tries unsuccessfully to employ monadic to with [dedicate [to [whorel person]]]. The problem is the merge of [the building]. At step 48, the attempted XP-YP pair merge of [the building] fails. At step 59, the corresponding XP-YP set merge fails. After that, there are a further two parallel attempts pushing [to [whorel person]] onto the stack. |
(28) [Give me] the phone number of the person whose mother's friend's sister's dog's appearance had offended the audience (Example (31a); Sag 1997: 450)
|
[audience,the,offend,'v*',[appearance,'\'s',[dog,'\'s',[sister,'\'s',[friend,'\'s',[mother,'\'s',[person,who_rel]]]]]],perf,v,'Tpast',c_rel,the,of,number,phone,the] | The DP with the relative pronoun whorel is
deeply embedded recursively within the specifier of
possessive 's for the head noun
appearance. However, it must be visible to Rel
probing.
Therefore, [[[[[[whorel person]['s mother]]['s friend]]['s sister]]['s dog]]['s appearance]] is attracted to the edge of Crel by Rel on Crel. whorel cannot value D, so person will raise further to form a new head. Determiner the merges with the NP headed by person, valuing D on person. |
(29a) [This is] the senator to whose mother's friend's sister's favorite charity the lobbyist had donated a small fortune (Example (31b); Sag 1997: 450)
(29b) [This is] the senator to whose mother's friend's sister's favorite charity that the lobbyist had donated a small fortune (29c) [This is] the senator whose mother's friend's sister's favorite charity the lobbyist had donated a small fortune to (29d) [This is] the senator to whose mother's friend's sister's favorite charity that the lobbyist had donated a small fortune |
[charity,favorite,'\'s',[sister,'\'s',[friend,'\'s',[mother,'\'s',[senator,who_rel]]]],to,[small,fortune,a],donate,'v*',[lobbyist,the],'Tpast',c_rel,the] |
Similar derivation to example (28). We wh-relativize
senator from the lobbyist donated a small fortune
to the senator's mother's friend's sister's favorite
charity.
[[[[[whorel senator]['s mother]]['s friend]]['s sister]]['s favorite charity]] is attracted to the edge of Crel by Rel on Crel. whorel cannot value D, so senator will raise further to form a new head. Determiner the merges with the NP headed by senator, valuing D on senator. There are four parses as the options of inserting the complementizer that and the pied-piping of prepositional to are both freely available. |
(30a) the time when I got drunk
(30b) *the time when that I got drunk |
[drunk,get,v_unerg,[i,d],'Tpast',[time,when_rel],c_rel,the] |
Assume the time expression [whenrel time] is
merged at TP level for semantic reasons. Both pair merge and
set merge options are tested. Because extraction must take
place for relativization, only the set merge option can
succeed. (Pair merged adjuncts are inaccessible to probing.)
[whenrel time] is attracted to the edge of Crel by Rel on Crel. whenrel cannot value D on time. Next, time raises and its uninterpretable D feature is valued when merged with the. However, whenrel can value T (as well as Rel) on Crel. By economy, no T-to-C, that in (30b), is possible. |
(31a) the town where I grew up
(31b) *the town up where I grew |
[up,grow,[town,where_rel],'v_unerg',[i,d],'Tpast',c_rel,the] |
Assume the location adverbial [whererel town]
is merged at the VP level. Both pair merge and
set merge options are tested. Because extraction must take
place for relativization, only the set merge option can
succeed. (Pair merged adjuncts are inaccessible to
probing.)
[whererel town] is attracted to the edge of Crel by Rel on Crel. whererel cannot value D on town. Next, town raises and its uninterpretable D feature is valued when merged with the. However, whererel can value T (as well as Rel) on Crel. By economy, no T-to-C (that) is possible. Finally, up does not head a PP with complement [whererel town] (merged at VP level instead). Therefore no pied-piping, as in (31b), is possible. |
(33a) someone to rely on
(33b) *someone on to rely |
[someone, d_rel, on, rely, 'v*', ['PRO', d0], 'Tinf', c_rel, d] |
[Drel someone] is attracted to the edge of
Crel by Rel on Crel.
Drel cannot value
D, so someone will raise further.
Tinf checks T on Crel.
(Drel cannot value T on Crel, T is
valued by Tinf instead.
Pied-piping is blocked by the empty determiner Drel, cf. overt whorel in (35a). (Note: this lexicon contains both simple prepositional on and dyadic on, e.g. put the book on the shelf. Step 17 onwards using dyadic on fails to derive.) |
(34a) someone for you to rely on
(34b) *someone on for you to rely |
[someone, d_rel, on, rely, 'v*', [you, d], 'Tinf', for, c_rel, d] | Assume complementizer for generally licenses an
overt subject, you, by assigning Case, cf. PRO in
(33a). Crel, containing Rel, piggy-backs onto
for, attracting [Drel someone] to its
edge.
Covert Drel in [on [Drel someone]] incompatible with pied-piping of on, so (34b) is blocked. As described in (33a), someone will raise further and merge with covert d. |
(35a) the baker in whom to place your trust (Example (63a);
Sag 1997:461)
(35b) *the baker whom to place your trust in (Example (63d); Sag 1997:461) |
[baker, who_rel, in, [ trust, '\'s', [you, d]], place, 'v*', ['PRO', d0], 'Tinf', c_rel, the] | Both (35a-b) are predicted to be grammatical here,
i.e. pied-piping is permitted. (Judgements shown are from
Sag.)
Assume in is dyadic, i.e. VP structure is [place [your trust [in [who baker]]]]. (Note: the derivations with dyadic in begin at step 51 and end at step 103. Simple preposition in is tried first, but fails to converge at step 50.) Either [in [whorel baker]] (for pied-piping) or [whorel baker], the internal argument of in, may be attracted to edge of Crel by Rel on Crel. NB: whorel is defective, i.e. it doesn't value the D feature on baker. Therefore, baker raises from edge of C and its D feature is checked by determiner the. |
(36a) the baker to place your trust in
(36b) *the baker in to place your trust |
[baker, d_rel, in, [ trust, '\'s', [you, d]], place, 'v*', ['PRO', d0], 'Tinf', c_rel, the] | No overt whorel here,
cf. (35a-b). Covert Drel is used instead,
i.e. internal argument of in is [Drel
baker]. [Drel baker] is
attracted to the edge of Crel by Rel on
Crel. Pied-piping is not an option since
Drel is an empty category, cf. (35b).
Drel cannot value D so baker will raise further and its uninterpretable D feature will be checked by the. Assume T on Crel is valued by Tinf. (Note: the derivation with dyadic in begins at step 28 and concludes at step 54. Simple preposition in is tried first, but fails to converge at steps 16 and 27 for pair and set merge, respectively.) |
(37) the person for us to visit (Example (75b); Sag 1997: 464)
|
[person, d_rel, visit, 'v*', [we, d], 'Tinf', for, c_rel, the] |
Similar to (34a), assume complementizer for
V generally licenses an overt subject, we, in a
tenseless clause by assigning Case. Crel,
containing Rel, piggy-backs onto the complementizer
for, attracting [Drel person] to its
edge.
Drel cannot value D on person. The head person raises further and merges with the, which values person's D feature. |
(38) the person to visit (Example (75c); Sag 1997: 464) | [person, d_rel, visit, 'v*', ['PRO', d0], 'Tinf', c_rel, the] |
Compared with (37), there is no for with an overt
subject; instead, we assume the covert subject is PRO.
[Drel person] is attracted to the edge of Crel by Rel on Crel. Drel cannot value D on person. The head person raises further and merges with the, which values person's D feature. |
Example | Instruction stream (clickable) | Notes |
---|---|---|
(1) the boy who told the story
(cf. Figs. 2-3 and 18a in paper; Keenan & Hawkins 1987: 63) |
[story, the, tell, 'v*', [boy, who_rel], 'Tpast', c_rel, the] | A subject relative clause.
Subject who boy attracted to edge of Crel, i.e. C with an Rel. DP who boy is marked with an Rel feature, originating on who. Note, whorel is defective, i.e. it doesn't value D feature on boy. Instead, boy raises from edge of C and its D feature is checked by the. |
(2) the book that I read
the book I read (cf. Figs. 4-5 in paper, Gallego 2006:151) |
[book, d_rel, read, 'v*', [i, d], 'Tpast', c_rel, the] | Two derivations.
Covert Drel has Rel but cannot check book's D feature. Crel has T and Rel, and heads a Local Extent (LE). Two ways to satisfy Crel using Pesetsky and Torrego (2001): (1) T to edge of Crel => that (2) Nom Case-marked I to edge of Crel => empty At the LE boundary, book (uninterpretable D feature) is restacked. TOS book merges with CrelP. book labels (head merging with non-head). Stream the merges with SO headed by book. |
(3) The man that I talked to
The man I talked to *The man to that I talked *The man to I talked (cf. Figs. 6-7 in paper) |
[man, d_rel, to, talk, 'v*', [i, d], 'Tpast', c_rel, the] |
Two derivations.
Covert Drel has Rel but cannot check man's D feature. Crel has T and Rel, and heads a LE. Two ways to satisfy Crel using Pesetsky and Torrego (2001): (1) T to edge of Crel => that (2) Nom Case-marked I to edge of Crel => empty At the LE boundary, man (uninterpretable D feature) is restacked. TOS man merges with CrelP. man labels (head merging with non-head). Stream the merges with SO headed by man. [Note: no pied-piping of verb particle to with Drel-man. [An] EC [Empty Category] disallows pied-piping (Chomsky, 2001:28)] |
(4) The boy who called Mary
*The boy called Mary *The boy who that called Mary (cf. 15a-b in paper) |
[mary, d, call, 'v*', [boy, who_rel], 'Tpast', c_rel, the] |
One derivation.
whorel has Rel. whorel can check T but not D feature. The subject whorel-boy is formed in a substream. boy (uninterpretable D feature) will be initially stacked in the substream and transferred back to the main stream. Crel heads a LE, and has T and Rel. Internal merge of whorel boy will satisfy both T and Rel on Crel. (By economy, T to C is not available.) At the LE boundary, boy (uninterpretable D feature) is restacked. boy merges with CrelP, then with the. |
(5) The boy that called Mary
*The boy called Mary (cf. Fig. 8 in paper) |
[mary, d, call, 'v*', [boy, d_rel], 'Tpast', c_rel, the] |
Covert Drel has Rel but cannot check
man's D feature.
Crel has T and Rel, and heads a LE. Assume Drel cannot check T, i.e. Drel-boy, despite getting nom Case, fails to satisfy T on Crel. Then Crel's T can only be satisfied by moving T to the edge of Crel => that. |
(6) the book which I read
*the book which that I read (cf. Fig. 9 in paper; Gallego 2006:151) |
[book, which_rel, read, 'v*', [i, d], 'Tpast', c_rel, the] | Single derivation.
whichrel has Rel. whichrel can check T but not D feature. Crel heads a LE, and has T and Rel. Internal merge of whichrel book will satisfy both T and Rel on Crel. (By economy, T to C is not available.) At the LE boundary, book (uninterpretable D feature) is restacked. book merges with CrelP, then with the. |
(7) The man who John saw
*The man who that John saw (cf. Fig. 10 in paper; Gallego 2006:154) |
[man, who_rel, see, 'v*', [john, d], 'Tpast', c_rel, the] |
One derivation.
whorel has Rel. whorel can check T but not D feature. Crel heads a LE, and has T and Rel. Internal merge of whorel man will satisfy both T and Rel on Crel. (By economy, T to C is not available.) At the LE boundary, man (uninterpretable D feature) is restacked. man merges with CrelP, then with the. |
(8) The man who loves Mary
*The man who that loves Mary (cf. Fig. 11 in paper; Gallego 2006:151) |
[mary, d, love, 'v*', [man, who_rel], 'T', c_rel, the] |
One derivation.
whorel has Rel. whorel can check T but not D feature. The subject whorel-man is formed in a substream. man (uninterpretable D feature) will be initially stacked in the substream and transferred back to the main stream. Crel heads a LE, and has T and Rel. Internal merge of whorel man will satisfy both T and Rel on Crel. (By economy, T to C is not available.) At the LE boundary, man (uninterpretable D feature) is restacked. man merges with CrelP, then with the. |
(9) the man who(m) I talked to
the man to who(m) I talked *the man who(m) that I talked to (cf. Figs. 12-13 in paper; Gallego 2006:152) |
[man, who_rel, to, talk, 'v*', [i, d], 'Tpast', c_rel, the] | Two derivations.
whorel has Rel. whorel can check T but not D feature. Crel heads a LE, and has T and Rel. Internal merge of whorel man will satisfy both T and Rel on Crel. (By economy, T to C is not available.) At the LE boundary, man (uninterpretable D feature) is restacked. man merges with CrelP, then with the. Optional pied-piping of verb particle to. |
(10) the time when I got drunk
*the time when that I got drunk (cf. Fig. 14 in paper, Reviewer example) |
[drunk,get,v_unerg,[i,d],'Tpast',[time,when_rel],c_rel,the] |
Assume the time expression [whenrel time] is
merged at TP level for semantic reasons. Both pair merge and
set merge options are tested. Because extraction must take
place for relativization, only the set merge option can
succeed. (Pair merged adjuncts are inaccessible to probing.)
[whenrel time] is attracted to the edge of Crel by Rel on Crel. whenrel cannot value D on time. Next, time raises and its uninterpretable D feature is valued when merged with the. However, whenrel can value T (as well as Rel) on Crel. By economy, no T-to-C, that in (30b), is possible. |
(11) the baker in who(m) to place your trust
(cf. Fig. 15 in paper; Sag 1997: 461) *the baker whom to place your trust in (Example (63d); Sag 1997: 461) |
[baker, who_rel, in, [ trust, '\'s', [you, d]], place, 'v*', ['PRO', d0], 'Tinf', c_rel, the] |
Both examples are predicted to be grammatical here,
i.e. pied-piping is permitted. (Judgments displayed are from
the Sag paper.)
Assume in is dyadic, i.e. VP structure is [place [your trust [in [who baker]]]]. (Note: the derivations with dyadic in begin at step 51 and end at step 103. Simple preposition in is tried first, but fails to converge at step 50.) Either [in [whorel baker]] (for pied-piping) or [whorel baker], the internal argument of in, may be attracted to edge of Crel by Rel on Crel. NB: whorel is defective, i.e. it doesn't value the D feature on baker. Therefore, baker raises from edge of C and its D feature is checked by determiner the. |
(12) the person to visit
(cf. Fig 16 in paper; Sag 1997: 464) |
[person, d_rel, visit, 'v*', ['PRO', d0], 'Tinf', c_rel, the] |
Compared with (13), there is no for with an overt
subject; instead, we assume the covert subject is PRO.
[Drel person] is attracted to the edge of Crel by Rel on Crel. Drel cannot value D on person. The head person raises further and merges with the, which values person's D feature. |
(13) the person for us to visit
(cf. Fig. 17 in paper; Sag 1997: 464) |
[person, d_rel, visit, 'v*', [we, d], 'Tinf', for, c_rel, the] |
Assume complementizer for
generally licenses an overt subject, we, in a
tenseless clause by assigning Case. Crel,
containing Rel, piggy-backs onto the complementizer
for, attracting [Drel person] to its
edge.
Drel cannot value D on person. The head person raises further and merges with the, which values person's D feature. |
(14) What I read
*What that I read (cf. Fig. 18 in paper) |
[pro_n, what_rel, read, 'v*', [i, d], 'Tpast', c_rel, d] |
Headless relative.
One derivation. Assume what is whatrel + pron (empty pronominal). whatrel has Rel. whatrel can check T but not D feature. Crel heads a LE, and has T and Rel. Internal merge of whatrel pron will satisfy both T and Rel on Crel. (By economy, T to C is not available.) At the LE boundary, pron (uninterpretable D feature) is restacked. pron merges with CrelP, then with d. |
(15) What annoys John
*What that annoys John (cf. Fig. 19 in paper) |
[john, d, annoy, 'v*', [pro_n, what_rel], 'T', c_rel, d] |
Headless relative.
One derivation. Assume what is whatrel + pron (empty pronominal). whatrel has Rel. whatrel can check T but not D feature. Crel heads a LE, and has T and Rel. Internal merge of whatrel pron will satisfy both T and Rel on Crel. (By economy, T to C is not available.) At the LE boundary, pron (uninterpretable D feature) is restacked. pron merges with CrelP, then with d. |
(16) the letter which Dick wrote yesterday
(cf. 23b in paper; Keenan & Hawkins 1987: 63) |
[letter, which_rel, write, 'v*', [dick, d], 'Tpast', [yesterday, d], c_rel, the] | An object relative clause.
Assume, substantially following (Haumann, 2007) and (Larson, 1985), that yesterday is a TP-level bare NP adverb. As an adjunct, yesterday is pair merged at TP. Defective whichrel is attracted by Crel; the entire DP which letter raising to the edge of C. Finally, letter within the DP raises and its D feature is valued by the. |
(17) the man who Ann gave the present to
the man to who Ann gave the present (cf. 23c in paper; Keenan & Hawkins 1987: 63) |
[man, who_rel, to, [present, the], give, 'v*', [ann, d], 'Tpast', c_rel, the] | Relativization of the indirect object of a double object
verb.
Two analyses, also produces pied-piped variant. Crel attracts who man, headed by whorel, to its edge. Defective who, can't value D on man. man raises. D on man valued by c-commanding the. |
(18) the box which Pat brought the apples in
the box in which Pat brought the apples (cf. 23d in paper; Keenan & Hawkins 1987: 63) |
[box, which_rel, in, [apples, the], bring,'v*', [pat, d], 'Tpast', c_rel, the] | Assume bring...in involves a dyadic preposition
in.
Crel attracts which box, headed by whichrel, to its edge. Defective which, can't value D on box. box raises. D on box valued by c-commanding the. |
(19) the dog which was taught by John
(cf. 23e in paper; Keenan & Hawkins 1987: 63) |
[dog, which_rel, teach, [[john, d], by], prt, 'v~', 'Tpast', c_rel, the] | Relativization of passivized object.
The agentive by-phrase is pair-merged to teach (not PRT or v~ due to merge over move preference). Crel attracts which dog, headed by whichrel, to its edge. Defective which, can't value D on dog. dog raises. D on box valued by c-commanding the. |
(20) the boy who Mike writes better than
*the boy than who Mike writes better (cf. Fig. 20 in paper; Keenan & Hawkins 1987: 63) |
[boy, who_rel, than, better, write, 'v_unerg', [mike, d], 'T', c_rel, the] |
Relativization of the object of a comparative.
Assume than is a preposition, not a C - as needed in the case of Comparative Deletion (not implemented here: e.g. as in Mike writes better than I do/write. The pied-piped counterpart (see cf.) is blocked by a spellout rule (after the parse has been completed). Crel attracts who boy, headed by whorel, to its edge. Defective who, can't value D on boy. boy raises. D on boy valued by c-commanding the. |
(21) the girl whose friends bought the cake
(cf. Fig. 21 in paper) |
[cake, the, buy, 'v*', [friends,'\'s',[girl, who_rel]],'Tpast',c_rel,the] |
Relativization of a genitive from a subject.
Assume the entire possessive 's DP Defective whorel cannot value D on girl. This permits girl to raise out and its D feature is valued by the. Finally assume spellout rule: who + 's -> whose. |
(22) the man whose house Patrick bought
the man whose house that Patrick bought (cf. Fig. 22 in paper; Keenan & Hawkins 1987: 63) |
[house,'\'s',[man,who_rel],buy,'v*',[patrick,d],'Tpast',c_rel,the] |
Relativization of a genitive from the object.
Assume possessive 's DP This permits man to raise out and its D feature is valued by the. Finally assume spellout rule: who + 's -> whose. |
(23) the boy whose brother was taught by Sandra
(cf. 27 in paper) |
[brother, '\'s', [boy, who_rel], teach, [[sandra, d], by], prt, 'v~', 'Tpast', c_rel, the] |
Relativization of a genitive from the passivized object.
Assume possessive DP whorel cannot value D on boy. This permits boy to raise out and its D feature is valued by the. Finally assume spellout rule: who + 's -> whose. |
(24) the girl who friends of bought the cake
*the girl of who friends bought the cake (cf. Fig. 23 in paper) |
[cake, the, buy, 'v*', [girl, who_rel, of, friends, d], 'Tpast', c_rel, the] |
Relativization of a genitive from a subject.
Assume DP Crel attracts who girl, headed by whorel, to its edge. Defective who, can't value D on girl. This permits girl to raise. D on girl is valued by c-commanding the. The pied-piped counterpart is blocked by a spellout rule that asserts there is no spellout for whorel. |
(25) the man who Patrick bought the house of
*the man of who Patrick bought the house (cf. 29 in paper) |
[man, who_rel, of, house, the, buy, 'v*', [patrick, d], 'Tpast', c_rel, the] |
Relativization of a genitive from the object.
Assume DP Crel attracts who man, headed by whorel, to its edge. Defective who, can't value D on man. This permits man to raise. D on man is valued by c-commanding the. The pied-piped counterpart is blocked by a spellout rule (after the parse has been completed). |
(26) the boy who the brother of was taught by Sandra
*the boy of who the brother was taught by Sandra (cf. 30 in paper) |
[boy, who_rel, of, brother, the, teach, [[sandra, d], by], prt, 'v~', 'Tpast', c_rel, the] |
Relativization of a genitive from a passivized object.
Passivization applies first. (Here we basically follow the analysis in Sobin (2014).) Tpast attracts Crel attracts who boy, headed by whorel, to its edge. Defective who, can't value D on boy. This permits boy to raise. D on boy is valued by c-commanding the. |
(27) [Give me] the phone number of the person whose
mother's friend's sister's dog's appearance had offended the
audience
(cf. Fig. 24 in paper; Sag 1997: 450) |
[audience,the,offend,'v*',[appearance,'\'s',[dog,'\'s',[sister,'\'s',[friend,'\'s',[mother,'\'s',[person,who_rel]]]]]],perf,v,'Tpast',c_rel,the,of,number,phone,the] |
The DP with the relative pronoun whorel is
deeply embedded recursively within the specifier of
possessive 's for the head noun
appearance. However, it must be visible to Rel
probing.
Therefore, [[[[[[whorel person]['s mother]]['s friend]]['s sister]]['s dog]]['s appearance]] is attracted to the edge of Crel by Rel on Crel. whorel cannot value D, so person will raise further to form a new head. Determiner the merges with the NP headed by person, valuing D on person. |
Example | Instruction stream (clickable) | Notes |
---|---|---|
(28) the town where I grew up
*the town up where I grew (Reviewer example) |
[up,grow,[town,where_rel],'v_unerg',[i,d],'Tpast',c_rel,the] |
Assume the location adverbial [whererel town]
is merged at the VP level. Both pair merge and
set merge options are tested. Because extraction must take
place for relativization, only the set merge option can
succeed. (Pair merged adjuncts are inaccessible to
probing.)
[whererel town] is attracted to the edge of Crel by Rel on Crel. whererel cannot value D on town. Next, town raises and its uninterpretable D feature is valued when merged with the. However, whererel can value T (as well as Rel) on Crel. By economy, no T-to-C (that) is possible. Finally, up does not head a PP with complement [whererel town] (merged at VP level instead). Therefore no pied-piping is possible. |
(29) the baker to place your trust in
*the baker in to place your trust |
[baker, d_rel, in, [ trust, '\'s', [you, d]], place, 'v*', ['PRO', d0], 'Tinf', c_rel, the] | No overt whorel here. Covert Drel is used instead,
i.e. internal argument of in is [Drel
baker]. [Drel baker] is
attracted to the edge of Crel by Rel on
Crel. Pied-piping is not an option since
Drel is an empty category.
Drel cannot value D so baker will raise further and its uninterpretable D feature will be checked by the. Assume T on Crel is valued by Tinf. (Note: the derivation with dyadic in begins at step 28 and concludes at step 54. Simple preposition in is tried first, but fails to converge at steps 16 and 27 for pair and set merge, respectively.) |
(30) the book which inspired them
*the book which that inspired them (Sag 1997: 448) |
[they,d,inspire,'v*',[book,which_rel],'Tpast',c_rel,the] | Subject relative. DP [book, whichrel] is
attracted to the edge of Crel by Rel on
Crel. By economy, whichrel values
both T and Rel on Crel, and the T-to-C option is
impossible, i.e. *which that.
However, whichrel cannot value D so book will raise further and D is valued during the merge with the. |
(31) the person whose mother died
(Sag 1997: 448) |
[mother, '\'s', [person, who_rel], die, 'v~unacc', 'Tpast', c_rel, the] |
Assume whose here is whorel +
's and die is unaccusative,
i.e. [[whorel person]['s mother]] is
theta-merged at the internal argument position.
Furthermore, permit Rel on Crel to find Rel on whorel at the edge of possessive 's. [[whorel person]['s mother]] is attracted to the edge of Crel. However, whorel cannot value D on person. Next, person raises and its uninterpretable D feature is valued when merged with the. |
(32) the person whose mother's dog died
(Sag 1997: 448) |
[dog, '\'s', [mother, '\'s', [person, who_rel]], die, 'v~unacc', 'Tpast', c_rel, the] | Analysis is the same as for (31) above, except we permit Rel on Crel to recursively look for Rel at the edge of possessive 's. Internal argument is [[[whorel person] 's mother]['s dog]]. |
(33) the person whose mother's dog we were (all) fond of
(33') the person whose mother's dog that we were (all) fond of (33") *the person of whose mother's dog we were (all) fond (33"') *the person of whose mother's dog that we were (all) fond (Sag 1997: 448) |
[dog, '\'s', [mother, '\'s', [person, who_rel]],of,fond,[we,d],v_be,'Tpast',c_rel,the] |
[[[whorel person] 's mother]['s dog]] is
attracted to the edge of Crel by Rel on
Crel. We assume, following (32), that Rel on
Crel can find recursively Rel at the edge of
possessive 's.
whorel cannot value D on person. Next, person raises and its uninterpretable D feature is valued when merged with the. (33') is predicted as [[[whorel person] 's mother]['s dog]] is not a nominative subject, so economy does not apply. The option of T-to-C generates the complementizer that. (33") and (33"') are blocked by a spellout rule *of who, designed to block *the girl of who friends bought the cake. |
(34) the person to whom they dedicated the building
the person to whom they dedicated the building (Sag 1997: 448) |
[person,who_rel,to,[building,the],dedicate,'v*',[they,d],'Tpast',c_rel,the] |
We assume dyadic to, i.e. the VP is [dedicate [[the
building] [to [whorel
person]]]]. [whorel person] is attracted to the
edge of Crel by Rel on Crel.
whorel cannot value D on person. Next,
person raises and its uninterpretable D feature is valued when merged with
the.
Pied-piping is permitted as whorel is overt. Note: after the two successful derivations, at step 41, the system tries unsuccessfully to employ monadic to with [dedicate [to [whorel person]]]. The problem is the merge of [the building]. At step 48, the attempted XP-YP pair merge of [the building] fails. At step 59, the corresponding XP-YP set merge fails. After that, there are a further two parallel attempts pushing [to [whorel person]] onto the stack. |
(35) [This is] the senator to whose mother's friend's
sister's favorite charity the lobbyist had donated a small
fortune
(35') [This is] the senator to whose mother's friend's sister's favorite charity that the lobbyist had donated a small fortune (35") [This is] the senator whose mother's friend's sister's favorite charity the lobbyist had donated a small fortune to (35"') [This is] the senator to whose mother's friend's sister's favorite charity that the lobbyist had donated a small fortune |
[charity,favorite,'\'s',[sister,'\'s',[friend,'\'s',[mother,'\'s',[senator,who_rel]]]],to,[small,fortune,a],donate,'v*',[lobbyist,the],'Tpast',c_rel,the] |
Similar derivation to example (34). We wh-relativize
senator from the lobbyist donated a small fortune
to the senator's mother's friend's sister's favorite
charity.
[[[[[whorel senator]['s mother]]['s friend]]['s sister]]['s favorite charity]] is attracted to the edge of Crel by Rel on Crel. whorel cannot value D, so senator will raise further to form a new head. Determiner the merges with the NP headed by senator, valuing D on senator. There are four parses as the options of inserting the complementizer that and the pied-piping of prepositional to are both freely available. |
(36) someone on whom to rely
*?someone who to rely on (Reviewer example, reviewer judgement) |
[someone, who_rel, on, rely, 'v*', ['PRO', d0], 'Tinf', c_rel, d] | |
(37) someone to rely on
*someone on to rely (Reviewer example) |
[someone, d_rel, on, rely, 'v*', ['PRO', d0], 'Tinf', c_rel, d] |
[Drel someone] is attracted to the edge of
Crel by Rel on Crel.
Drel cannot value
D, so someone will raise further.
Tinf checks T on Crel.
(Drel cannot value T on Crel, T is
valued by Tinf instead.
Pied-piping is blocked by the empty determiner Drel. (Note: this lexicon contains both simple prepositional on and dyadic on, e.g. put the book on the shelf. Step 17 onwards using dyadic on fails to derive.) |
(38) someone for you to rely on
*someone on for you to rely (Reviewer example) |
[someone, d_rel, on, rely, 'v*', [you, d], 'Tinf', for, c_rel, d] | Assume complementizer for generally licenses an
overt subject, you, by assigning Case, cf. PRO in
(37). Crel, containing Rel, piggy-backs onto
for, attracting [Drel someone] to its
edge.
Covert Drel in [on [Drel someone]] incompatible with pied-piping of on. As described in (37), someone will raise further and merge with covert d. |
Example | Instruction stream (clickable) | Notes |
---|---|---|
(rcx1) the boy John thinks called Mary
the boy John thinks that called Mary |
[mary, d, call, 'v*', [boy, d_rel], 'Tpast', c_e, think, v_unerg, [john, d], 'T', c_rel, the] | Relativize from embedded subject position. |
(rcx2) the student who lives here who studies English | [[here, d, live, 'v*', [student, who_rel], 'T', c_rel], [english, d, study, 'v*', [student, who_rel], 'T', c_rel], the] | Doubly stacked relative clause. Uses FormSet. |
(rcx3) the student who lives here who studies English who I know | [[here, d, live, 'v*', [student, who_rel], 'T', c_rel], [english, d, study, 'v*', [student, who_rel], 'T', c_rel],[student, who_rel, know, 'v*', [i, d], 'T', c_rel], the] | Triply stacked relative clause. Uses FormSet. |