Overt Evidence from Left-Branch Extraction in Polish for Punctuated Paths

In Polish, wh-questions are formed by the pied-piping of an entire wh-NP or by the extraction of a left-branch wh-phrase (left-branch extraction; LBE). A question formed by a fronted wh-NP is given in (1). In this squib, I argue that an NP stranded by a wh-phrase marks a position in which the wh-NP has been merged in its derivational history; thus, the LBE facts provide evidence for successive-cyclic movement. There can be at least four such positions, all marked by a stranded NP: the base-generated position of the wh-NP, (2); the left edge of the VP, (3); the left edge of the vP, (4); and—in the case of long-distance wh-movement—the left edge of the embedded CP, (5).

1. 1. Jaki samochód Paweł kupił swojej żonie t_wh?

   what car_acc Paweł_nom bought his wife_dat

   ‘What car did Paweł buy his wife?’

2. 2. Jaki Paweł kupił swojej żonie samochód?

   what Paweł_nom bought his wife_dat car_acc

3. 3. Jaki Paweł kupił samochód swojej żonie t_wh?

   what Paweł_nom bought car_acc his wife_dat

4. 4. Jaki Paweł samochód kupił swojej żonie t_wh?

   what Paweł_nom car_acc bought his wife_dat

5. 5. ?Jaki pro myślisz samochód (*że) Paweł kupił

   what (you) think car_acc that Paweł_nom bought

   swojej żonie t_wh?

   his wife_dat

   ‘What car do you think that Paweł bought his wife?’

Interestingly, some speakers also accept a long-distance wh-question construction in which a wh-NP is stranded at the edge of the upper vP.

1. 6. %Jaki Maria samochód myślała, że Paweł

   what Maria_nom car_acc thought that Paweł_nom

   kupił swojej żonie t_wh?

   bought his wife_dat

   ‘What car did Maria think Paweł bought his wife?’

Constructions in which the movement of the left branch strands the NP in a fronted position, then, provide new evidence for successive-cyclic movement and, more broadly, for punctuated paths in syntax. In this squib, I follow the logic of McCloskey’s (2000) work on a dialectal Irish English and Barbiers’s (2002) work on Dutch, where remnants of constituents stranded in a fronted position are argued to teach us about the nature of movement.¹ In what follows, I argue that the positions marked by the stranded NP are indeed edges of phases: CP, vP, and—perhaps somewhat less obviously—VP. In sections 1 and 2, I outline the basics of word order and wh-movement in Polish. In section 3, I argue that LBE can take place from wh-NPs fronted to the edges of phases. In section 4, I show that the dislocations of wh-NPs to phase edges are truly instances of successive-cyclic movement and cannot be analyzed as scrambling.

1 The Position of Arguments in Polish

The basic word order of monotransitive constructions in Polish is S-V-O (7), and the basic word order of ditransitive constructions is S-V-IO-DO (8).

1. 7. Paweł lubi kawę.

   Paweł_nom likes coffee_acc

   ‘Paweł likes coffee.’

2. 8. Paweł dał Marii książkę.

   Paweł_nom gave Maria_dat book_acc

   ‘Paweł gave Maria a book.’

Although scrambling can change the order of arguments in Polish, there exists evidence that the S-V-IO-DO word order is indeed basic. For instance,Witkoś and Dziemianko (2006) propose that such evidence comes from the syntax of idioms. Idioms have been extensively argued to involve unmarked word orders (see Larson 1988, Svenonius 2005, and references cited therein), and the word order of Polish idioms is V-(IO_dat)-DO_acc.

1. 9.

   1. a. masz (ci) babo placek

      have (you_dat) woman_voc pie_acc

      (cf. *masz placek babo)

      ‘what bad luck’

   2. b. piłkarze gryzą trawę

      footballers_nom bite grass_acc

      (cf. *trawę gryzą piłkarze)

      ‘footballers put their hearts into the game’

The same word order is the only one attested in discontinuous idioms. As shown in (10), the core of the idiom includes the verb and...