<?xml version="1.0" standalone="yes"?>
<Paper uid="C02-1034">
  <Title>A quantitative model of word order and movement in English, Dutch and German complement constructions</Title>
  <Section position="5" start_page="1" end_page="2" type="metho">
    <SectionTitle>
CTYP Decl_finite
</SectionTitle>
    <Paragraph position="0"> Other clause constituents receive their landing site (cf. Table 1) in a similar manner. Figure 2 depicts the configuration after Fronting of NP Kim.</Paragraph>
    <Paragraph position="1"> Figure 3 below includes a paraphrase where the focus on Kim is stressed prosodically rather than by Fronting. This is indicated by the disjunctive set carrying the tag 4 . In sentence generation, the Read-out module selects one alternative, presumably in response to pragmatic an other context factors. In parsing mode, one or the alternatives is ruled out because it does not match word order in the input string.</Paragraph>
    <Paragraph position="2"> The formalism defined so far yields unordered hierarchical structures. However, the values of the TPL features enable the derivation of ordered output strings of lexical items. As indicated above in connection with Figure 2, we assume that this task can be delegated to a simple Read-out module that traverses the clause hierarchy in a depth-first manner and processes the topologies from left to right  .</Paragraph>
    <Paragraph position="3"> If a slot is empty, the Reader jumps to the next slot. If a slot contains a lexical item, it is appended to  A slot may contain more than one phrase (e.g., Direct and Indirect OBJect in slot M3; cf. Table 1). We assume they have been ordered as part of the append operation, according to the sorting rule associated with the slot. the current output string and tagged as already processed. It follows that, if a slot happens to be shared with a lower topology, its contents are only processed at the higher clause level, i.e., undergo  promotion.</Paragraph>
    <Paragraph position="4"> 4. Linearization of complement clauses in English, Dutch and German  The PG formalism developed above provides a simple quantitative linearization method capturing both within-clause and between-clause phenomena. The assignment of constituents to topology slots (including, e.g., scrambling in Dutch and German) has been dealt with in Kempen &amp; Harbusch (in press; forthcoming). In the present paper we focus on promotion in complement constructions -- a domain where the three target languages exhibit rather dissimilar ordering patterns. We highlight the fact that PG enables highly similar treatments of them, differing only with respect to the settings of some quantitative parameters.</Paragraph>
    <Paragraph position="5"> The movement (promotion) phenomena at issue here depend primarily on the values assigned to sharing parameters LS and RS in five different types of complement clauses. These settings are shown Table 2. They are imported from the lexicon and control the instantiation of the TPL feature of the root S-node of the complement. We begin with some illustrations from English.</Paragraph>
    <Paragraph position="6">  The non-finite complements of do and have in sentence (1) below are both declarative. (Cf. the paraphrase &amp;quot;For which person x is it the case that I have to call x&amp;quot;, which highlights the scope of who.) It follows that LS=3 in both complements. Notice that do is treated as a 'Verb Raiser', have (in have to) as a VP Extraposition verb.</Paragraph>
    <Paragraph position="7">  know Dana hates (cf. Figure 1) and We know Dana hates  Kim. The versions correspond to different options of the topology value associated with the CoMPlement (indicated within curly brackets). Empty slots are not shown in the TPL features.</Paragraph>
    <Paragraph position="9"> who to call In example (2), the lower clause is finite and declarative --cf. the paraphrase &amp;quot;For which person x is it the case that you said that John saw x&amp;quot;. The scope of who exceeds its 'own' clause and includes the matrix clause. In (3), on the other hand, the scope of the interrogative pronoun does not include the main clause (&amp;quot;I know for which person x it is the case that John saw x&amp;quot;). Therefore, the complement is interrogative and cannot share its F1 slot with that of the main clause.</Paragraph>
    <Paragraph position="10">  The system predicts 'island effects' as in (4).</Paragraph>
    <Paragraph position="11"> (4) a. Who did you claim that you saw? b.*Who did you make the claim that you saw? The lexical frame of the verb claim includes an S-CMP-S segment identical to that of know above (repeated here for convenience):</Paragraph>
  </Section>
  <Section position="6" start_page="2" end_page="2" type="metho">
    <SectionTitle>
CTYP Decl_finite
</SectionTitle>
    <Paragraph position="0"> The feature matrices of root and foot nodes of this segment both specify a TPL feature referencing the slot F1. This enables insertion of coreference tag 2 and thus promotion of the filler of slot F1.</Paragraph>
    <Paragraph position="1"> However, the complement segment of the noun claim is rooting in an NP node, which cannot have a TPL feature with type F1t.</Paragraph>
    <Paragraph position="2">  is meaningless here, ruling out promotion in (4b).</Paragraph>
    <Paragraph position="3"> Turning now to Dutch, we first refer to Table 3, which specifies some important landing sites for major clause constituents. Because of the similarity with German, we combine the two languages. First, we illustrate question formation.</Paragraph>
    <Paragraph position="4"> Dutch interrogative main clauses feature Subject-Verb inversion without the equivalent of doinsertion (cf. 5).</Paragraph>
    <Paragraph position="5">  Because the complement in (6) is interrogative, the sharing rule in Table 2 prohibits left-peripheral sharing (LS=0).</Paragraph>
    <Paragraph position="6">  (6) Zij vroeg of ik hem kende  German). Precedence between constituents landing in the same slot is marked by &amp;quot;&lt;&amp;quot;.</Paragraph>
    <Section position="1" start_page="2" end_page="2" type="sub_section">
      <SectionTitle>
Finite Complement
</SectionTitle>
      <Paragraph position="0"> The subordinate clause in (7) features clause union, causing the auxiliary zal to intervene between the Direct hem the latter's governor bellen.</Paragraph>
      <Paragraph position="1"> The left-peripheral sharing area may vary between 4 and 6 slots (LS=4:6). Because hem lands in M3, i.e. in the shared area, it is promoted. The remainder of the lower topology, including the HeaD bellen itself, occupies E1 -- one of the options of the complement of a Verb Raiser. The other option, with the complement in M5 (giving bellen zal) is also allowed.</Paragraph>
      <Paragraph position="2"> (7) ...dat ik hem zal bellen that I him will phone '...that I will phone him'</Paragraph>
      <Paragraph position="4"> hem bellen Sentence (8) illustrates the treatment of 'particle hopping'. The positions marked by &amp;quot; [?] &amp;quot; are grammatical alternatives to the particle (op) position mentioned in the example; no other positions are allowed. Given LS=4:6 for complements of Verb Raisers, it follows that hem is obligatorily promoted into the higher topology:  However, sharing of the fifth slot (M4) is optional. If this option is realized in the middle topology, the order zou op hebben gebeld ensues. If, in addition, the middle topology shares M4 with its governor, the string comes out as op zou hebben gebeld. The treatment of cross-serial dependencies is exemplified in (9). In order to deal with this construction, we need to make an additional assumption about the order of constituents that land in the same slot but originate from different levels in the clause hierarchy. We stipulate that constituents from more deeply embedded clauses trail behind constituents belonging to higher clauses. This ordering can be determined locally within a slot if we equip every constituents in the hierarchy with a numerical 'clause depth' index (for instance, a Gorn number; Gorn, 1967). Given this index (not shown in the topology diagram accompanying (9)), the order hem de fiets results.</Paragraph>
      <Paragraph position="5"> (9) ... dat ik hem de fiets wil helpen maken that I him the bike want-to help repair '... that I want to help him to repair the bike'</Paragraph>
      <Paragraph position="7"> de fiets maken We now turn to German, concentrating on structures usually labeled &amp;quot;VP Extraposition&amp;quot; (10) and &amp;quot;Third Construction&amp;quot; (11).</Paragraph>
      <Paragraph position="8"> (10) ... dass er uns zwingt es zu tun that he us forces it to do '... that he forces us to do it'  (11) a. ... dass er uns verspricht es zu tun  that he us promises it to do '... that he promises us to do it' b. ... dass er es uns zu tun verspricht  The verb zwingen allows its complement to share slot F1 only (LS=1). This prevents promotion of the Direct OBJect es. Third Construction verbs like versprechen allow a great deal of variation in the size of the left-peripherally shared topology area (LS=1:6), thereby licensing optional promotion of es. However, since es is a personal pronoun, it only takes M2 as its landing site (see Table 3). The latter constraint is violated in (11e).</Paragraph>
    </Section>
  </Section>
class="xml-element"></Paper>