Stranding Quirky Case: Revisiting Case Connectivity in Icelandic

The distribution of quirky case in control and raising structures has been taken as a strong argument against Hornstein’s 1999 movement analysis of control (see for example, Landau 2003 amongst many others). In this squib, I argue that it is possible to adopt a movement analysis of control, while at the same time capturing the distribution of quirky case. 

.pdf version

The paradigm in question is give below, from Landau 2013:

(1) Control:

strákarnir-i    vonast   til að     PRO-i   vanta   ekki   alla  ı́   skólann

the.boys.NOM hope  for to     PRO.ACC to.lack   not   all.ACC   in  school

“The boys hope not to be all absent from school.”

(2) Raising:

strákana-i virðist    t-i vant       ekki alla          ı́ skólann

the.boys.ACC    seem      to.lack not all.ACC  in the.school

“The boys seem not to be all absent from school.”

In (1) the matrix subject bears nominative, while the embedded PRO bears quirky accusative, as shown by the case marking on the floated quantifier. The quirky accusative is assigned by the embedded verb “to lack”. In (2), the matrix subject and the embedded subject both bear quirky accusative. 

Landau (2013: 27) puts the generalization as follows: “The point of current interest is that case-concord often reveals that the null subject of a control complement bears a distinct case from its controller. However, the null subject of a raising complement uniformly displays the case of its antecedent, the matrix raisee.”

Based on this data, Landau (2003: 492) draws the following conclusion: “Hornstein' s theory is unable to deal with this contrast. The Icelandic facts furnish such a compelling argument against the reduction of control to raising because they so clearly reflect the chain structure of the construction.”

Under the standard theory (where raising is analyzed by DP movement, and control is a relation between an antecedent and PRO), the difference in (1) and (2) is straightforward. The analysis is shown below (putting aside many interesting issues in the syntax of Icelandic infinitival complement clauses):

(3) DP1  seem  [TP <DP1>    to <DP1>  Vquirky ]

(4) DP1    hope  [TP PRO1    to  <DP1>  Vquirky ]

In (3), the quirky case marked DP moves from the embedded clause to the matrix clause. In (4), there is no movement, rather the matrix subject controls the embedded subject.  So there is no way for the matrix subject to take on quirky case (although the embedded PRO has quirky case).

In order to give a movement account of this data (in the spirt of Hornstein 1999), suppose that a quirky case DP is really KP (where K is the quirky case suffix): [KP K DP]. This representation will allow DP to strand K under certain circumstances.

In a raising construction, KP can raise from the embedded clause to the matrix clause. The analysis is given in (5):

(5) KP1  seem  [TP    <KP1>    to    <KP1> Vquirky ]

I take the derivation in (5) to be unproblematic, deriving the sentence in (1).

Now consider how to analyze typical cases of control, as in example (2). In this case, since KP is the external argument of hope, it will have to raise to Spec vP of the matrix clause (in the spirit of Hornstein 1999):

(6) KP1  [vP <KP>   v [VP hope  [TP <KP1>    to <KP1> Vquirky ]]]

I propose that the problem here is that KP occupies Spec vP. Consider the following principle:

(7)

Suppose that H is an argument introducing head (assigns a theta-role). 

If KP is projection of quirky case, and KP is in Spec HP,

then H licenses the quirky case KP.

Ultimately, I take (7) to follow from the Theta-Criterion (e.g., as in Chomsky 1986). Little v assigns a theta-role to DP in (6), but DP is separated by the quirky case K from v. 

Now, in the structure in (6), KP occupies Spec vP, and v does not license quirky case (because hope does not have a quirky case external argument). And so the derivation crashes.

So KP itself cannot raise to Spec matrix vP with control verbs. But perhaps in certain circumstances, the DP in KP can move, stranding K:

(8) K-Stranding

….DP1…..[KP K <DP1>]…..

If such stranding is possible, then cases like (2) can be reanalyzed as:

(9) DP1  [vP <DP1> v  [VP hope [TP    [KP K  <DP1>]    to <KP> Vquirky ]]]

In (9), KP is licensed as the argument of the quirky case marking verb in the embedded clause. Then, KP moves to Spec TP of the embedded clause to satisfy the EPP. Then DP1 strands K and raises to Spec of the matrix vP (picking up the external theta-role of hope). Lastly DP1 raises to Spec of the matrix TP. At no point is there a violation of (7).

But if K-stranding is OK in (9), then what prevents K-stranding with a raising verb like seem? In other words, why isn’t the following derivation allowed:

(10) DP1  [VP seem  [TP [KP K <DP1>]    to KP Vquirky ]]

In (10), once again the quirky case KP raises to Spec TP of the embedded clasue. Then, DP1 strands K in the embedded clause and moves to the matrix Spec TP. In this example, the matrix subject would presumably have nominative case, and not the quirky case determined by the embedded verb.

I propose that (10) is ruled out by minimize search. The matrix Spec TP must be filled. But internal Merge of KP into matrix Spec TP does not violate any constraints. Therefore, KP is internally Merged into matrix Spec TP. And so it is impossible to search for the deeper DP within KP to fill matrix Spec TP. Therefore, K standing is blocked in (10).

In (6), on the other hand, movement of KP to Spec vP violates (7), and so it is blocked. Under these circumstances, movement of the DP to Spec vP stranding K is allowed.

Acknowledgments:

Thanks to Gary Thoms for discussing the ideas in this paper with me.

References

Hornstein, Norbert. 1999. Movement and Control. Linguistic Inquiry 30, 69-96.

Landau, Idan, 2003. Movement out of Control. Linguistic Inquiry 34, 471-498.

Landau, Idan. 2013. Control in Generative Grammar. Cambridge University Press, Cambridge.