LIN5317 A, Fall 2015 Dennis Ott Assignment 2 Solution 1. he Scandinaian languages show different degrees of erb moement. [ ] onsider the following representatie data from four Scandinaian languages: (1) a. *att han erkligen läste boken (Swedish) b. *att han läste erkligen boken (2) a. *Ég spurði [af herju Helgi hefði oft lesið þessa bók]. (Icelandic) b. *Ég spurði [af herju Helgi oft hefði lesið þessa bók]. (3) a. hóast fólk ongantíð heur fingið fisk her. (Faroese) b. hóast fólk heur ongantíð fingið fisk her. (4) a. *Jeg spurgte [horfor Peter hade ofte læst den]. (Danish) b. *Jeg spurgte [horfor Peter ofte hade læst den. i. Identify the erb-moement properties of each language (how far does raise?), and gie a reason for your diagnosis. In Swedish, the finite erb (+) does not raise to, as it follows aderbs. In Icelandic, the finite erb does raise to, as it precedes aderbs. In Faroese, the finite erb optionally raises to ; this is likely to be a dialectal split. In Danish, the finite erb does not raise to, as it follows aderbs. ii. As you see in (1), Swedish has definite DPs of the form boken the book, where the en affix expresses definiteness. How could we analyze such expressions, assuming that they are DPs? What kinds of eidence could support or refute your hypothetical analysis? One option is to assume that Swedish has inherently [±definite] nouns like boken, and the [+definite] forms are selected by a null determiner: [ DP ø [ NP[+definite] boken]]. A more syntactic alternatie would be to assume that DPs like boken start out as [ DP en [ NP bok ]], and the head noun bok undergoes head moement to the D en: [ DP bok i +-en t i ]. On this kind of N-to-D moement (in other contexts), see arnie, 316f., 322. A morphophonological ariant of the second option would be to assume that D and N undergo some sort of morphological fusion under linear adjacency. It s not easy to decide between these options (or others). We could try to see if insertion of some other element between D and N has any effect. eteris paribus, insertion of some XP (e.g. an AdjP) shouldn t hae any effects on either a null determiner selecting NP or on N-to- D moement (since the HM only rules out head moement that skips another head). But it might hae an effect on morphological fusion in that it would linearly separate D and N, just like negation separates erb and in English, blocking fusion and requiring do-support. he issue remains open. If you re interested in this phenomenon (which shows some ariation within the Scandinaian languages), I ll be happy to gie you releant references with further data, including on the scenario (insertion of AdjPs) described aboe. 1
2. Assuming Koopman & Sportiche s P-internal subject hypothesis (PISH) and the erbmoement parameter, can you explain the pattern of quantifier float in (5) and the asymmetry between the English examples in (6) and the meaning-equialent French examples in (7)? (You may but need not use [schematic] trees to illustrate your answer.) (5) Les enfants (*tous) ont (tous) u (*tous) ce film. the children (*all hae seen this film (6) a. *My friends all loe Mary. b. *My friends loe all Mary. (7) a. *Mes amis tous aiment Marie. b. *Mes amis aiment tous Marie. Reminders: o In both French and English (unlike Irish/Welsh), external arguments raise to Spec-. o Howeer, only in French do both auxiliaries and main erbs undergo moement to. In (5), les enfants has raised to Spec- stranding the quantifier in Spec-, where it (the quantifier) follows the finite auxiliary (in ) but precedes the participial main erb (+). [ P [les enfants] i [ ont+ [ P [tous t i ] [ u+ [ P t ce film]]]]] he difference in (6) s. (7) is due to erb moement: in English (6b), the quantifier cannot be stranded below the main erb, since this would mean that the external argument originated in some position below (but EAs originate in Spec-). By contrast, (6a) is compatible with the assumption that the quantifier is stranded in Spec-, whereas my friends is in Spec- (cf. My friends certainly all loe Mary). Schematically: [ P [my friends] i [ P [all t i ] [ P loe Mary ]]] In French (7b), the quantifier can be stranded below the main erb aiment, since the later has raised to hence, the surface order is compatible with the quantifier being stranded in Spec-. Schematically: [ P [mes amis] i aiment [ P [tous t i ] [ P t + Marie ]]] By contrast, no position is aailable for the quantifier in (7a): it is higher (to the left of) the main erb in, but its associate mes amis is already occupying Spec-. 3. In Kannada (spoken in India), causatie forms (such as (8b)) are morphologically more complex than inchoatie ones (such as (8a)). Gie clause-structure trees for (8a,b), and explain how they differ. (Assume that Kannada is head-final and has erb moement to.) (8) a. Neer kud-itu. water.nom boil-ps.1sg he water boiled. b. Naanu neerannu kud-is-ide. I.NOM water.a boil-aus-ps.1sg I boiled the water. 2
See figures 1, 2 on p. 6. Note that I e inerted the order: figure 1 = (8b), figure 2 = (8a). he causatie morpheme -is- in (8b) is analyzed as the light erb (), which incorporates into ia head moement; the resulting + complex in turn incorporates into, yielding the morpheme order -- (since head moement = left-adjunction to a higher head). he DP naanu receies an Agent role in Spec- before it raises to Spec-. By contrast, no is present in (8a), so no external argument is projected and no A ase is assigned, hence the object DP neer raises to Spec- to receie NOM ase. incorporates into. his is simply an unaccusatie deriation as we know it. Note: here and in what follows, I e indicated traces of heads ia a category subscript, i.e. the trace of a head X is written as t X, whereas I e used the familiar coindexation for traces of XPs. his allowed me to preent an unreasonable proliferation of indices in the more complex trees later on. Anything s fine as long as the representations are unambiguous! 4. onsider the following complex sentence from Breton, a eltic language spoken in Lower Brittany (France). (9) Ar merc hed a laare e wele Yann Mona. the women said that saw Yann Mona he women said that Yann saw Mona. i. Draw a tree for this sentence, indicating any releant moements and for each DP, where it receies its θ-role and ase, respectiely. (reat the sequence a laare as a single erb.) See figure 3 on p. 7. ii. How does Breton set Koopman & Sportiche s parameter concerning the assignment of Nominatie ase? NOM is assigned downward (under goernment/c-command) to the external argument in Spec-, as in Irish or Welsh. While we don t see this in the 2 main clause, where the subject has moed to initial position (Spec-), we see it in the embedded clause, where the subject is below, hence presumably in its base/theta position Spec-. From this we can infer that the main-clause subject, too, was both theta- and ase-marked in Spec-. (Note that this tells us that Breton doesn t need to satisfy an EPP on, at least not in all configurations; this is why I assume no intermediate stopoer in Spec- of the main-clause subject.) 5. In Icelandic (and other Scandinaian languages), possessie pronouns are syntactic reflexies, i.e. subject to Principle A of the Binding heory (hence glossed REFL for reflexie below). onsider now the following cases of Icelandic ditransities: (10) a. Ég sendi [ DP Rauðettui ] i [ PP til ömmu sinnari i ]. (Icelandic) I sent Little Red Riding Hood to grandma her.refl I sent Little Red Riding Hood i to her i grandma. 3
b. *Ég sendi [ DP ínið sitti i ] [ PP til ömmui i ]. *I sent wine her.refl to grandma * I sent her i wine to the grandma i. i. Explain why Principle A is satisfied in (10a) but not in (10b). he DP argument (heme) c-commands the PP argument (Goal), hence the former can bind into the latter but not ice ersa. his is consistent with Larson s English facts. he erb precedes both DP and PP object because it raises to (and subsequently to, since this is Icelandic, and on to, since this is a 2 main clause). ii. Relying on Larson s insights and the analysis of 2 we discussed in class (Icelandic is a head-initial 2 language), gie a tree for (10a) that shows its deriation. See figure 4, p. 8. (I used dashed arrows to indicate moement of the erb, for readability.) 6. onsider the following example: (11) Fiskurinn reyndist [era aleg nýr]. (Icelandic) the fish.nom proed [be completely fresh i. What are the two possible readings of (11), one of which is considerably more plausible than the other? [ ] How do these two readings differ in terms of θ-assignment? Pragmatically plausible reading: the fish is the subject of be fresh in the embedded clause, raises into the matrix clause to get ase (but no theta-role!) there. his will yield the reading corresponding to It turned out that the fish was fresh. Schematically: [ P [the fish] i proed Raising [ P t i ]] Pragmatically implausible reading: the fish receies a theta-role in the matrix clause (Agent of proe); by the heta riterion, this means that it cannot also receie a theta-role in the embedded clause, where consequently the subject of be fresh must be silent PRO. he result is the reading where the fish is a proer controlling the subject of be fresh in the embedded clause. Schematically: [ P [the fish] i proed ontrol [ P PRO i ]] ii. Draw trees illustrating the deriations underlying (11), and indicate which tree yields which reading. Bear in mind that Icelandic is 2 in main clauses! reat era as the main erb of the infinitial clause, whose complement is the AdjP aleg nýr. See figures 5, 6 on p. 9f. Somewhat contrary to my own suggestion, I treated era as a light erb that takes an AdjP as its complement; I beliee this is somewhat more in line with arnie s treatment, but nothing really hinges on this. What s important is that the DP the fish starts out in a local configuration with its predicate be fresh. Note also that I omitted the arrows indicating erb moement, to reduce clutter. 4
iii. Why does the sentence he fish i proed that it i was completely fresh differ in meaning from (11) in being unambiguous? Since the embedded clause here is finite and contains a subject on its own (it), the fish must receie its (Agent) theta-role in the matrix clause. Hence, there is no way this ersion could hae the (pragmatically plausible) Raising reading without iolating the heta riterion. 7. On the last page of the Week 4 handout, we briefly saw but didn t discuss further the fact that unergatie erbs like dance but not unaccusatie erbs like remain can be passiized in languages that allow impersonal passies: Dutch It was danced (by the young people) (32b) s. *It was remained (by the children) (33b). In light of our distinction between unaccusaties and unergaties in terms of (as summarized on the bottom of p. 2 and the last page of the last, Week 6 handout), can you think of an explanation for this behaior? Unergaties are, as per our assumptions, composite erbs consisting of and (the latter introducing an external argument). Unaccusaties hae no external argument, hence no. In passie, A and the external argument are suppressed (Burzio!), which we modeled as suppression of (meaning: it s not merged into the structure). [ P DP θ [ P (P) ]] à passie [ P (P) ] In unaccusaties, there simply is no to suppress they are inherently passie (arnie). If we were to merge an unaccusatie to a passie auxiliary and furthermore merge an expletie into Spec-, to get It was remained, the heme theta-role assigned by remain is not assigned, and the resulting structure will necessarily iolate the heta riterion. [ P EXPL [ P à θ?? ] Note that the problem persists een if we merge an adjunct by-phrase to P (It was remained by the children), since adjuncts neer receie theta-roles. 5
P DP k naanu P t k P DP t t kud is ide neerannu Figure 1: I boiled the water P DP k neer P t k t kud itu Figure 2: he water boiled 6
P DP i ar merc hed P t P a laare PS t i θ Agent NOM t P t P P e P DP wele PS Yann θ Agent NOM t t P DP Mona θ heme A Figure 3: he women said that Yann saw Mona. 7
P DP k ég P t k t P sendi PS t k t P DP Rauðettui t PP til ömmu sinnari Figure 4: I sent Little Red Riding Hood i to her i grandma. 8
P DP i fiskurinn P t i t P reyndist PS t P t i P [+INF] t i era t AdjP aleg nýr Figure 5: he fish proed to be completely fresh. pragmatically plausible reading 9
P DP i fiskurinn P t i t P reyndist PS t i t P t P P PRO i P [+INF] t PRO Figure 6: he fish proed to be completely fresh. pragmatically implausible reading era t AdjP aleg nýr 10