<?xml version="1.0" standalone="yes"?>
<Paper uid="C96-2135">
  <Title>Yet Another Paper about Partial Verb Phrase Fronting in German</Title>
  <Section position="3" start_page="0" end_page="800" type="metho">
    <SectionTitle>
2 The Phenomena
</SectionTitle>
    <Paragraph position="0"> Ill German, it is possible to front non-maximal verbal projections} (1) a. \[Erz~hlen\] wird er seiner Tochter tell will he his daughter ein M~rchen.</Paragraph>
    <Paragraph position="1"> a fairy tail 'He will tell his daughter a fairy tale.' b. \[Erz~hlen mfissen\] wird er tell must will he seiner Tochter ein MPSrchen.</Paragraph>
    <Paragraph position="2"> his daughter a fairy tale 'He will have to tell his daughter a fairy tale.' In a series of papers, Hinrichs and Nakazawa argued for a special rule schema that combines the verbs of a so-called verbal complex before the arguments of the involved verbs are combined with the verbal complex. Because the verbal complex is build before any nonverbal argument of a verb gets saturated, it is possible to account for phenomena like auxiliary flip. As the verbal complex is analyzed as a constituent, the fi'onting of erz~ihlen miissen in (lb) can be explained as well. There is no problem with sentences like those in (1) for the standard NONLOC mechanism. Erziihlen miissen is a constituent in the non-fronted position in (2) and the same holds if the verbal complex is fronted.</Paragraph>
    <Paragraph position="3"> (2) Er wird seiner Tochter ein M~rchen \[erzPShlen miissen\].</Paragraph>
    <Paragraph position="4"> There are, however, examples where a partly saturated verbal complex is fronted.</Paragraph>
    <Paragraph position="5">  (3) a. \[Seiner Tochter ein MPSrchen erz~hlen\] wird er.</Paragraph>
    <Paragraph position="6"> b. \[Ein MPSrchen erzPShlen\] wird er seiner Tochter.</Paragraph>
    <Paragraph position="7"> c. \[Ein M~irchen erz~ihlen\] wird er seiner  d. \[Seiner Tochter erzghlen\] wird er das Mgrchen.</Paragraph>
    <Paragraph position="8"> A verb with some of its arguments may appear in the Vorfeld leaving other arguments in the Mittelfeld. null As (4) shows, it is possible that a PP in the Mittelfeld modifies a fronted verbal complex. (4) Den Kanzlerlmndidaten ermorden the chmlcellor.candi(late kill wollte die Frau mit diesem Messer.</Paragraph>
    <Paragraph position="9"> wanted the woman with this knife 'The woman wanted to kill the candidate with this knife.' Sentences like (5a) are ungrammatical. It is not possible to front parts of tile verbal comi)lex that would be located in the middle of the verbal eom~ plex in a verb tinal sentence (Sb).</Paragraph>
    <Paragraph position="10"> (5) a. * Miissen wird er ihr must will he her ein M~irchen erz/ihlen.</Paragraph>
    <Paragraph position="11"> a story tell b. , well er ihr ein Mgrchen erz'ahlen m{issen wird.</Paragraph>
  </Section>
  <Section position="4" start_page="800" end_page="801" type="metho">
    <SectionTitle>
3 The Analysis
3.1l Basic AssumI)tions
</SectionTitle>
    <Paragraph position="0"> In what follows, I assume a version of IIPSG that deviates from standard HI'SG in that the surface string of a t)hrasal sign is not determined by a relation that relates the PIION vahms of a sign to the PI:ION values of its daughters (Pollard and Sag, 1987, p. 169). Instead I will follow Reape's (1994) at)preach. Reape assumes word order domains as an additional level of representation. In such a domain, all daughters of a head occur. These domains differ from the daughter list, in that the ele, ments in a domain (signs) correspond in their serialization to the surface order of the words in the string. LP-eonstraints apply to elements of the order domain. Another basic assumption of Reape is that constituents may be discontinuous.</Paragraph>
    <Paragraph position="1"> As Hinriehs and Nakazmva (1994a) have shown, it is reasonable to assume in addition to the head complement schema a schema that licenses the -verbal complex. Hinri(:hs and Nakazawa introduced the concept of argument attraction into the HPSG framework. If a vert)al (:omI)lex is build two verbs are combined and the resulting sign inher~ its all arguments from both verbs. In their paper, Hinrichs and Nakazawa treat verbal complements as ordinary complements that, are included in the COMPS list of their heads. It has however proven 1,o be useful to distinguish the verbal complement from other complements (Rentier, 1994a; Mfiller, 1995a). The merits of this move will t)e discussed shortly, l~br the Imrpose of representing the intbrmation at)out verbal colnt)lements , the fl;ature VCOMP is introduced. Its vahle is a synscm-objeet if the verb embeds another verb and none otherwise. '.\['he entry in the stem lexicon for tile fllt;ure tense auxiliary we(den (will) is shown in (6). 1,Yore this stem the mort)hology eomt)onent pro- null duces the finite form shown in (7). In German, ahnost any complemellt of a verb can be fronted, subjects as well as objects. Therefore, fox' tinite forms the subject is in(:luded into the COMPS list, from where extraction is possible, l~br nonfinite forms the subject does not appear on COMPS hut stays in the SUBJ list. 2 Schema 1 licenses verb vird:</Paragraph>
    <Paragraph position="3"> verbal complement (\[~). The resulting sign is a  verbal complex or a part of a verbal complex. It is marked LEX+ because it can in turn be embedded. (8) , well er ihm ein M/irchen because he him a fairy talc \[\[erz~hlen lassen\] hat\].</Paragraph>
    <Paragraph position="4"> tell let has 'because he has let him tell the story.'</Paragraph>
    <Section position="1" start_page="801" end_page="801" type="sub_section">
      <SectionTitle>
3.2 The LEX Feature
</SectionTitle>
      <Paragraph position="0"> The LEX feature in the entry for werden ensures that a matrix verb is combined with its verbal complement before the verbal complement is saturated by one of its complements. It is therefore possible to avoid multiple structures in the Mittelfeld. null  (9) a. Er wird seiner Tochter ein M/irchen \[erzghlen mfissen\].</Paragraph>
      <Paragraph position="1"> b. Er wird sciner Toehter \[\[ein M/irchen erz/ihlen\] miissen\]\].</Paragraph>
      <Paragraph position="2"> c. Er wird \[\[seiner Tochter ein Miirchen erz//hlen\] miissen\]\].</Paragraph>
      <Paragraph position="3">  But exactly those constituents that have to be avoided in the Mittclfeld are needed in the Vorreid. Very complicate mechanisms have been introduced to cope with this problem without a lot of spurious ambiguities (Nerbonne, 1994; Hinrichs and Nakazawa, 1994b). I will suggest a solution to the problem that is very simple: If it is the case that an embedded verb or verbal complex has to be LEX+ when verb and complement are tombitted locally and if it is the case that this does not hold if a nonlocal dependency is involved than the simplest solution is to view LEX not as a local feature. If one assumes that LEX lives under the path SYNSEM instead of SYNSEMILOC than the problem turns into a non-issue.</Paragraph>
      <Paragraph position="4"> Figures 1 and 2 show the analyses of the sentences in (10). 4 In the analyses of (10a), a trace flmctions as a verbal complement. In (10b) a trace for a verb is modified by an adverb.</Paragraph>
      <Paragraph position="5"> (10) a. Seiner Tochter erziihlen wird er das M/irchen.</Paragraph>
      <Paragraph position="6"> b. Vortragen wird er es morgen.</Paragraph>
      <Paragraph position="7"> Sentences like (5a) are ruled out because wird selects a complement in bse-form that has a VCOMP value none. As erziihlen does not appear in any COMPS list it is not possible for the verb to count as an argument of the fronted verbal complex that is saturated in the Mittel\]eld. This is the case in Pollards account. Hinrichs and Nakazawa have to block this case by stating type constraints 4In the original grammar, I use a binary branching schema for head-complement and verb cluster struclures. Adjuncts and complements are inserted into the domain of their head so that word order facts are accounted for. Due to space limitations, the figures show a tree for a flat head-complement structure. on lists of attracted arguments. With a separate VCOMP feature this problem disappears.</Paragraph>
    </Section>
    <Section position="2" start_page="801" end_page="801" type="sub_section">
      <SectionTitle>
3.3 The Problem of Underspeeified
</SectionTitle>
      <Paragraph position="0"/>
    </Section>
  </Section>
  <Section position="5" start_page="801" end_page="803" type="metho">
    <SectionTitle>
COMPS Lists
</SectionTitle>
    <Paragraph position="0"> In this section, I will address a problem that seems to have gone unnoticed until now. All analyses that involve argument attraction admit signs with underspecified COMPS lists. So in (1), wird is coinbined with a trace or a lexical rule is applied to it. The LOC value of the verbal complement is put into SLASII and the argmnents of the verbal complement are attracted by the matrix verb. This list of argmnents, however, is not instantiatcd in the resulting sign. It remains variable until the SLASIt element becomes bound. Therefore, the HPSG principles admit any kind of combination of totally unrelated signs. Since the COMPS list of the head is variable, any constituent is a possible complement. 5 As an HPSG theory is assumed to be a set of constraints that describe well formed descriptions of linguistic objects, this is clearly not wanted. If a grammar contains phonologically empty elements (traces, relativizers, and the like) the set of ill-formed signs will be infinite because wird -i could be combined with arbitrarily many empty elements. 6 It is clear that we want the matrix verb to behave in a very well defined way. It shall attract exactly the arguments of the fronted verbal projection that were not saturated by this projection, i.e., the matrix verb shall perform the argmnent attraction that would take place in base position, abstracting away from tile value of LEX. The desired effect can be reached if a rule schema is used for the introduction of nonlocal dependencies. To introdnce a nonlocal dependency for a verbal complex, this schema requires an additional licensing condition to be met. The extracted element is licensed by an actually existing verbal projection in the string. When a hearer of a sentence hears the words that have to be combined with a trace or introduce the nonlocal dependency in another way, he or she has already heard the phrase actually located in the Vorfeld. Therefore, the infor-Ination about the nonlocal dependency is present and can be used to license the extracted element.</Paragraph>
    <Paragraph position="1"> The COMPS list of the extracted element, therefore is specified. The specified COMPS are attracted by tile matrix verb and the COMPS list of the matrix verb therefore does not contain any variables and our theory does not admit signs that don't describe linguistic objects.</Paragraph>
    <Paragraph position="2"> ~The same problem exists for analysises that treat verb second as verb movement (Kiss and Wesche,  Schema 2 shows how this is implemented. A verbal complement of a matrix verb is saturated. The VCOMP value of the resulting sign is none. The LOC value of the saturated verbal complement is moved into SLASH. This LOC value is licensed by another verbal projection that meets the local requirements of the matrix verb but may be positioned in the Vorfeld. As there are no constraints for daughters to be adjacent to each other, there may be an arbitrary munber of constituents between the licensing daughter and the head daughter. The licensing daughter has licensing function only and is not inserted into the domain of the resulting sign (\[~) at this point of combination.</Paragraph>
    <Paragraph position="3"> However, an appropriate sign is inserted into the domain of its head when the nonlocal dependency is bound.</Paragraph>
  </Section>
  <Section position="6" start_page="803" end_page="803" type="metho">
    <SectionTitle>
4 Alternatives
</SectionTitle>
    <Paragraph position="0"> The drawback of the approaches of Pollard (To appear) and Nerbonne (1994) are discussed in (Hinrichs and Nakazawa, 1994b). I will not repeat the arguments against these approaches here. Instead, I will explain some of the problems of the Hinrichs and Nakazawa approach\] Hinrichs and Nakazawa changed the value of SLASII into a set of signs rather than local objects. The fronted phrase is a maximal projection with the missing constituents moved to SLASH. The fronted partial phrase is the filler for a nonlocal dependency which is introduced by their PVP-Topicalization Lexical Rule. As SLASII elements are signs, the lexical rule can refer to the SLASH set of a SLASlt element and it is thus possible to establish a relation between the COMPS list of the auxiliary and the SLASH set of the fronted verbal 7Due to space limitations, I cannot give a detailed discussion of their approach here. The interested reader is referred to (Mfiller, 1996).</Paragraph>
    <Paragraph position="1"> projection. However, the assumption that SLASH contains signs rather than local objects is a change of the basic HPSG formalism with far reaching consequences that is not really needed and that has some side effects.</Paragraph>
    <Paragraph position="2"> In the following, I discuss two problems for this approach. Firstly, it is not possible to account for cases where a modifier in the Mittelfeld modifies the fronted verbal projection without, assuming an infinite lexicon because the only way for a modifier to stay in the Mittelfeld while the modified constituent is fronted is that the modifier is contained in the SLASIt set of the fronted constituent. It, therefore had to be a member of the COMPS list. An infinite lexicon is both not very nice from a conceptual point of view and an implementational problem. Without; a complex control strategy (late evaluation) it is not possible to implement an infinite lexicon. Another problem that was pointed out by Hinrichs and Nakazawa themselves is sentences like (11).</Paragraph>
    <Paragraph position="3"> (11) * Gewuflt, daft Peter-i sehlggt, habe known that Peter hit have ich siei.</Paragraph>
    <Paragraph position="4"> I her 'I knew that Peter hit her.' In (11), sic is extracted from the complement sentence of gewuflt and than inserted into the COMPS list of babe and saturated in the Mittelfeld. The same problem arises for other constructions involving nonlocal dependencies, s (12) a. \[Da\]i hatte Karl \[-i mit\] gerechnet.</Paragraph>
    <Paragraph position="5"> this had Karl with counted 'Karl expected this.' b. * \[\[-i mit\] gerechnet \] hatte \[da\]i Karl. (13) a. Bus/ will Karl\[_/ fahren\].</Paragraph>
    <Paragraph position="6"> bus wants Karl drive 'Karl wants to go by bus.' b. * \[ -i fahren\] will Karl bus/.</Paragraph>
  </Section>
class="xml-element"></Paper>