More specifiers, more functional projections

    Lots of
    Functional
    Projections     		 

    Example:

      I have always love peanut butter.

    Our new tree (V->T movement and NP movement):

    Example:

      Have I always loved peanut butter.

    Our new tree (T->C Movement):

    Trends:
    1. More functional distinctions
      1. (NegP (not),
      2. AspP (have,be): PerfP,ProgP,
      3. Various adverb licensing projections to constrain adverb placement.
    2. Positions in sentences defined by heads or specs of functional projections (Spec of CP [preposed constits], C [Auxes, question particles), Spec of TP [subject], T, Spec of VP [VP-internal subject].
    3. Cross linguistic data tackled:
      1. Differences in word order correlating with presence and absence of overt T
      2. Inversion in questions; complementary distruibution with complementizers/question particles Differences in word-order between in embedded and matrix clauses (German, Dutch)
      3. Differences in adverb placement (French, English); correlation with differences in inversion

V->T Movement

    V->T  

    Parametric variation:

    1. V->T: French
    2. Affix lowering: English
    Languages are always one or the other.

    French: T->C, which implies V->T:

      Mange-t-il du pain
      Eat-∅-he of bread

      Does he eat bread?

    English: affix lowering.

      Will he eat bread? (T->C for Auxes)
      *Eats he bread? (T->C for Auxes only)
    Explanation: No V->T, which is a pre-requisite for T->C. Evidence:
      He happily ate the bread
      * He ate happily the bread.
      He will happily eat the bread.
      ? He happily will eat the bread.
    Adverbs occurs immediately to the right of Auxes, but not to the right of main verbs. Assumption:
      V' → Advp V'
      T' → T VP
    So no adverb immediately to the right of transitive verb:

    Complication: English have and be.

    1. Invert (in contrast to ordinary verbs):
      1. Has John left?
      2. Is John going?
      3. * Eats John his lunch?
      4. * Keeps John laughing?
    2. Can't be lexical Ts (like modals) because they co-occur with modals: They may (be laughing/have been laughing).

    Solution: They are verbs (or members of a special category like PerfP or ProgP) and they do V->T movement.

T->C Movement

    T->C  

    T->C movement accounts for inversion (English, Franch).

    1. Is John going?
    2. Mange-t-il des pommes? (* Eats he apples?)
    Note that French inversion works for main verbs, not just auxes. This is characteristicv of a langauge that allows main verbs to move to T (V->T), whence they can move into C (T->C).

    Incompatible with the presence of a complementizer:

    1. John wonders will he come?
    2. John wonders whether he will come.
    3. * John wonders whether will he come?
    German  

    The case of German (Problems 3 and 5 last week):

    1. V -> T
        (a) Sprechen Sie Deutsch?
        Speak you German?
        (b) Ist er nach Haus gegangen?
        Is he to house gone?
        (c) Er sitzt nicht auf diesem Tisch?
        He sits not on this table
        (d) Er soll nicht auf diesem Tisch sitzen.
        He should not on this table sit
    2. Trees
        PATTERN A
      Note: our analysis of German is that it is underlyingly verb final.
        VP -> XP V
      This is strongly supported by the fact that the verb occurs in VP final position when the aux is present (examples [b] and [d]). As we'll see below, verb-final VPs also surface in embedded clauses.

      When nothing fills T, the main verb hops from its underlying position into T (V -> T) and thence into C (T-> C):

      When an Aux fills T, the main verb stays where it is and the Aux hops into C (T-> C):

      Given the trees we've drawn so far, the data above presents a problem! If inversion were optional, we would expect the following word order to be possible in main clauses:

        * Er nicht auf diesem Tisch sitzt. (no T-> C)
      It's not. Then, if inversion were obligatory, we should get:
        * Sitzt er nicht auf diesem Tisch. (T -> C)
      This is hopeless. Instead we get. the English-like:
        Er sitzt nicht auf diesem Tisch. [=3c, p. 214, Carnie]
      This is why you may have assumed, when you did this problem, that German T' were not head final. On this assumption, Carnie (3c) is not problematic:
        PATTERN B
      Pattern B exactly captures the data (3a)-(3d) in Carnie:
      (3a) [C Sprechen[MV]] Sie Deutsch?
      (3b) [C Ist[Aux] ] er nach Hause gegangen?
      (3c) Er [T sitzt[MV] ] nicht[Neg] auf diesem Tisch.
      (3d) Sie [T soll[Modal]] nicht auf diesem Tisch sitzen[MV].
      with uninverted main clauses posing no problem. Univerted verbs and auxes go into T (which is 2nd position). Only inverted verbs and auxes go into C, which is first position.

      Nevertheless, Pattern A is empirically correct, and the problem with main clauses goes away, as some further data shows. This goes flashing briefly by in problem 5.

    3. Verb Second (Problem 5):

      The analysis you are asked to assume:

      1. A movement rule puts a constituent (of a wide variety of types) into Spec of CP (first position of sentence)
      2. An obligatory T -> C movement puts the resident of T in C position (which is now 2nd position overall):

          [CP [XP Preposed Topic ] [C' [C V_j ] [TP [NP Subject] [T' [VP [V' [NP Object ] [V_j e] ]][T_j e]]]]]

      Examples:

        Diesen Film haben die Kinder gesehen
        This film have the children seen
        Die Kinder haben diesen Film gesehen
        The children have this film seen

    So what does this analysis say about example (3c), p. 214, which was giving us problems above?

    These work out fine now. Like (a) and (b)[the inverted examples], these examples involve T->C movement.. In fact T->C movement is obligatory in German in main clauses. The difference between inverted clauses and uninverted clauses is that in UNinverted clauses, one MORE movement has happened! [Preposing of XP to topic position].


      [CP [NP_i Er] [C' [C [V_j sitzt] ] [TP [NP_i t] [T' [NegP [Neg' [Neg nicht ] [VP [V' [PP auf diesem Tisch ] [V_j e] ]][T_j e]]]]]]]

    So the apparently English-like SVO word order of German main clauses without auxiliaries arises by 3 movements: V->T, T->C, and preposing of topical subjects.


      [CP [NP_i Er] [C' [C [T_j soll] ] [TP [NP_i t] [T' [NegP [Neg' [Neg nicht ] [VP [V' [PP auf diesem Tisch ] [V_k sitzen] ]]][T_j e]]]]]]

    Of course an important part of the motivation for this analysis is the freedom of German word order. It doesnt have to be the subject that winds up in topic position:


      [CP [PP_l auf diesem Tisch ] [C' [C [V_j sitzt] ] [TP [NP_i er] [T' [NegP [Neg' [Neg nicht ] [VP [V' [PP_l t ] [V_j e] ]]][T_j e]]]]]]


      [CP [PP_l auf diesem Tisch] [C' [C [T_j soll] ] [TP [NP_i er] [T' [NegP [Neg' [Neg nicht ] [VP [V' [PP_l t ] [V_k sitzen] ]]][T_j e]]]]]]

    The key differences between English and German:

    1. V->T movement (predicting negation and adverb placement and invertability of main verbs ["Sprechen Sie Deutsch?"])
    2. Obligatory T->C movement. C must always be filled.
    3. Obligatory Topicalization in non-questions. Something must always fill Topic position.

    None of these facts YET motivate having T' be T-final, because so far we havent seen anything SURFACE in T position. Everything that moves to T just as quickly has to move out. But there are some cases where something gets to stay in T, and then we see what it's true position is, as the main facts of the problem show.

    Turning now to the main facts of the problem.

    Examples:

      * Er sagt[ dass die Kinder haben den Film gesehen]
        he said [that the children have the film seen ]
        Er sagt[ dass die Kinder den Film gesehen haben]
        he said [that the children the film seen have ]
    Apparently, movement into C position is blocked in embedded clauses. This might be explained by saying movement is blocked when C position is already filled (here, by the complementizer dass[=that]). In such cases we ought to see German V's and German T's in their underlying position. So here finally is some direct support for the idea that German T' is head final (= 5i, p. 217, Carnie):

      [CP [C' [C dass ] [TP [NP_i die Kinder] [T' [VP [V' [NP den Film] [V_k gesehen] ]][T_j haben ]]]]]
    In this example, then, there is no movement (except possibly the movement of a VP-internal subject to spec of TP position) and the underlying word order surfaces. It turns out to be a key claim of this analysis then that German is underlyingly am SOV language!

    Back to the mechanics of the problem: Our hypothesis is that (j)

      * Er sagt dass die Kinder den Film gesehen haben.
    is ungrammatical because movement of haben to C position is blocked because C position is filled, and that correlates with blocking movement of anything to topic position

    It appears to be a robust generalization that when Comp is filled with a complementizer, nothing moves to Spec of CP. The following is definitely out:

      * Er sagt die Kinder dass t den Film gesehen haben.
    The assumption we have to make is that When something other than when something fills the spec of C position, we get verb movement, as the examples above and example (k) from, Part II shows:

      [CP [AP Gestern] [C' [C [V_k sahen] ] [TP [NP_i die Kinder] [T' [VP [V' [NP den Film] [V_k e] ]][T_k e]]]]]
    Dutch  

    Is annoying. There is still a constraint against V->C movement in embedded clauses, but somehow heeft (have) and gezien (seen) are in the "wrong" order:

      * Ik geloof[ dat de man heeft een boek gezien]
        he believe [that the man have a book seen ]
        Ik geloof[ dat de man een boek heeft gezien]
        he believe [that the man a book have seen ]

    One possibility is that Dutch has a process somewhat like affix lowering (generalized to Auxes) which allows the contents of T to lower onto the verb:


      [CP [C' [C dat ] [TP [NP_i de man] [T' [VP [V' [NP een boek] [V [V_k heeft] [V_j gezien] ]] ] [T_k e ]]]]]
    Another, of course, is that gezien raises into T (despite T being filled with heeft) and somehow manages to combine with it.