Basic concepts of syntax

  • Roland Hausser


In this and the following two chapters, the grammar component of syntax is described. Section 16.1 analyzes the structural border between the grammar components of morphology and syntax. Section 16.2 discusses the role of valency in the syntactic-semantic composition of natural languages. Section 16.3 explains the notions of agreement and the canceling of valency positions by compatible valency fillers. Section 16.4 demonstrates the handling of free word order with an LA-grammar for a small fragment of German. Section 16.5 demonstrates the handling of fixed word order with an LA-grammar for a corresponding fragment of English.


Word Order Word Form Result Segment Combination Principle Category Segment 
These keywords were added by machine and not by the authors. This process is experimental and the keywords may be updated as the learning algorithm improves.


Unable to display preview. Download preview PDF.

Unable to display preview. Download preview PDF.


  1. 1.
    This is in contrast to the nativist variant of `generative semantics.’ It used transformations to derive word form surfaces from syntactic deep structures, e.g. persuade from cause to come about to intend (G. Lakoff 1972, p. 600).Google Scholar
  2. 2.
    This phrase is commented in G. Wahrig’s dictionary as `Adj.; nur noch in der Wendung… das ist so üblich [-= mhd. gäbe „ annehmbar“; zu geben]”.Google Scholar
  3. 3.
    In addition to the basic valency structure of a word, natural languages allow special uses with secondary valency structures. For example, the English verb to sleep is normally intransitive, but in the sentence This yacht sleeps 12 people it is used transitively. Thereby, the change to the secondary valency structure is accompanied by a profound change in the meaning of the word, which may be paraphrased as provide bunks for. There is also the inverse process of reducing the number of basic valency positions, called Detransitivierung in German. For example, give in German has a secondary use as a one-place — or arguably two-place — valency carrier, as in Heute gibt es Spaghetti (= today it gives spaghetti). Again, the secondary valency structure is accompanied by profound changes in the original word meaning.Google Scholar
  4. 4.
    In contrast, PS-grammar provides no formal means to account for the functor-argument structure of language at the level of syntax. This is reflected in the formal nature of the respective categories, which are atomic in PS-grammar, but of a composite structure in C- and LA-grammar.Google Scholar
  5. 5.
    The C-analysis is simplified in that it does not represent alternative orders of functor and argument by means of different slashes. Also, ungrammatical combinations are not excluded by the categorization. The LA-analysis is simplified in that no rules and agreement conditions are specified.Google Scholar
  6. 6.
    This format corresponds to the derivation structure (iii) in 10.3.1 as well as the automatic parsing analyses 10.3.2, 10.4.1, 10.5.3, 10.5.4, and 10.5.5. The equivalent format of LA-trees growing upward from the terminal string, illustrated in 10.5.2 and 16.2.1, on the other hand, has served only for the comparison with corresponding constituent structures, and is of no further use.Google Scholar
  7. 7.
    In German, this grammar is called LA-Beton. The English translation of Beton as concrete has a misleadingly positive ring, hence the name LA-Plaster.Google Scholar
  8. 8.
    The third position of the lemmata is indicated by a `*’ because the base forms are not referred to by the rules.Google Scholar
  9. 9.
    This freedom is utilized for other aspects of communication, such as textual cohesion.Google Scholar
  10. 10.
    Given the isolation of these variants in the context of a linguistic example, some may seem less natural than others. The reason for this, however, is not grammar, but that for some it is more difficult to imagine utterance situations suitable for their topic-comment structures.Google Scholar
  11. 11.
    LA-DI and the following LA-grammars for natural languages represent the categories and rule patterns in the form of lists, similar to the algebraic definition 10.2.1. Current applications of LA-grammar use the Malaga-System (cf. 15.2.6), which represents categories and rule patterns as feature-value structures. For development and application, Malaga is as powerful as it is comfortable. For a principled theoretical description, however, the list-based notation is more parsimonious and transparent.Google Scholar
  12. 12.
    Thus, multicats (cf. 15.2.4) and feature structures (cf. 15.2.6) — which are not really necessary for the syntactic properties to be explained here — can be omitted.Google Scholar
  13. 13.
    To a one-place verb there correspond 2 basic sentence patterns, one declarative and one interrogative, to a two-place verb altogether 4, and to a three-place verb altogether 12.Google Scholar
  14. 14.
    Accordingly, the variant of gave used in, e.g., John gave the book to Mary is based on the alternative lexical category (N’ A’ TO’ V).Google Scholar
  15. 15.
    In LA-D1, the categorial pattern of the valency carrier is (x np’ y V). The valency position to be canceled is decided by the fillers: their case marking, e.g. der Mann, dem Mann, den Mann, determine more or less unambiguously which of the available valency positions in the carrier they agree with — and which they may thus cancel.Google Scholar
  16. 16.
    For a complete description of the agreement restrictions of English see the variable definition of LA-E2 in 17.4.1.Google Scholar
  17. 17.
    The elementary formalisms of C- and PS-grammar require additional lexical readings or rule sequences for each additional constellation of agreement and each additional variant of word order. To improve the handling of word order, additional mechanisms like transformations or metarules were built into PS-grammar. To improve the handling of agreement, the mechanism of unification was added to PS- and C-grammar. These extensions have in common in that they increase mathematical complexity, but not the understanding of natural language communication.Google Scholar

Copyright information

© Springer-Verlag Berlin Heidelberg 1999

Authors and Affiliations

  • Roland Hausser
    • 1
  1. 1.Friedrich Alexander University Erlangen NürnbergErlangenGermany

Personalised recommendations