On Non-Head Non-Movement

In: G. Goerz (ed.): KONVENS 92, Springer Verlag, 1992, pp 218-227

Klaus Netter ?

Deutsches Forschungszentrum fur Kunstliche Intelligenz GmbH

Stuhlsatzenhausweg 3, D-6600 Saarbrucken 11, Germany

e-mail: [email protected]

The paper discusses different approaches to the analysis of finite verb position in German

within the framework of HPSG. It argues that among three parameters (a) standard subcatego-

rization and adjunction, (b) uniform structural representation and (c) clause initial finite verbs

as heads, only two can be maintained simultaneously, unless additional devices are introduced.

We abandon (c) and present an approach which allows to derive a binary right-branching struc-

ture for the Mittelfeld" also for V-initial structures without modifications of general principles

of the theory. The key to the analysis is the postulation of an empty node as a functional head

which takes the finite verb as a complement.

Die Arbeit diskutiert mehrere Analysen fur die Stellung des finiten Verbums im Deutschen

im Rahmen der HPSG. Es wird argumentiert, dass von den drei Parametern (a) Standard Subkat-

egorisierung und Adjunktion, (b) uniforme strukturelle Representation und (c) initiale Verben

als Kopfe, ohne eine Erweiterung der Theorie um spezielle Mechanismen nur zwei aufrechter-

halten werden konnen. Wir geben (c) auf und diskutieren eine Analyse, die auch bei V-initialen

Strukturen ein binar verzweigendes Mittelfeld erzeugt, ohne dass generelle Modifikationen die

Theorie erforderlich werden. Der Schlussel fur die Analyse ist die Annahme eines leeren Knoten

als funktionalen Kopf des Satzes, der das finite Verbum als Komplement subkategorisiert.

1 Introduction

The initial and final position of the finite verb in languages such as German or Dutch

presents an interesting problem for grammar frameworks such as Head Driven Phrase

Structure Grammar (HPSG) and related Categorial Grammar style theories. Since the

finite verb as the head of a clause may either precede or follow its complements and

adjuncts (1)/(2), the question arises whether the two types of clauses can and should be

assigned a uniform structural representation (modulo the position of the finite verb and

the complementizer).1

(1) [H Bringt ] [C er ] [A morgen ] [C dem Kunden ] [ A punktlich ] [C die Fracht ]

(2) dass [C er ] [A morgen ] [C dem Kunden ] [ A punktlich ] [C die Fracht ] [H bringt ]

?The research underlying this paper was supported by a research grant, FKZ ITW 9002 0, from the

German Bundesministerium fur Forschung und Technologie to the DFKI project DISCO. We would like

to thank Sergio Balari, John Nerbonne, Maike Paritong and Hans Uszkoreit for dicussions and comments.

The described approach has been implemented in the DISCO system on the basis of the type description

language TDL.

1Throughout this paper we will abbreviate the abstract grammatical functions head, complement

and adjunct as H, C and A respectively. In our discussion we will largely abstract away from the

phenomena of V-second or topicalization and from word order variations in the Mittelfeld". Both of

these phenomena we consider to be quite unrelated to the problem of the position of the finite verb and

for both a range of independent modular solutions can be envisaged.

218