Abstract
This chapter outlines a grammar which satisfies the Criteria of Psychological Well-Foundedness. Section 3.1 presents a syntactic theory which satisfies the Derivational Order Hypothesis. Section 3.2 illustrates how the syntactic analysis of Left-Associative Grammar is displayed by the associated NEWCAT parsers. Section 3.3 describes a semantic interpretation for this syntax which satisfies the Criteria of Psychological Adequacy. Section 3.4 illustrates the system with the analysis of a “garden path sentence.” Section 3.5 summarizes the descriptive, methodological and heuristic advantages of left-associative grammar.
Access this chapter
Tax calculation will be finalised at checkout
Purchases are for personal use only
Preview
Unable to display preview. Download preview PDF.
References
The time-linear derivation order characteristic of LA-grammar must be distinguished from Hock-ett’s (1966) notion of a “linear generative grammar,” which is simply a phrase structure grammar without transformations.
The Newcat derivation of 3.1.3 is given in 3.2.2.
See Section 6.2 for the algebraic definition of LA-grammar.
Obviously, LA-grammar is not restricted to Lisp; the rules of an LA-grammar can be parsed in any general purpose programming language. For example, there exists a version of LA-grammar in Prolog.
The rule r-2 is part of the LA-grammar for the context sensitive language akbkck defined in 6.4.3.
The rule-compiler was implemented by Gerald Klix (University of Munich). For use on a PC, the LA-grammar shell was translated into PC-SCHEME because it is considerably smaller than Golden Common Lisp.
Structured lists are easier to print and much faster to display than tree structures.
This category represents a V (sentential expression containing the finite verb) which still needs a D (dative) and an A (accusative) to become a complete sentence.
Examples of ungrammatical sentences are marked with an asterisk (“*”), a convention which dates back at least to Bloomfield (1933).
Since Montague Grammar doesn’t construct a bidirectional surface-meaning mapping, it also fails to satisfy the Criteria of Procedural Adequacy (2.4.1).
See Montague (1974), pp. 222 – 246.
E.g., a semantic type or a kind of model-theoretic denotation.
Hausser (1984a).
For example, we don’t have to build a minimal database from the tree, because the structure was built as a database in the first place.
That the hierarchial analyses of LA-grammar use slot names like SUBJ, DIR-OBJ, etc., is partly a concession to popular usage, and partly intended to distinguish the semantic representation from the syntactic analysis, which uses cases.
Carbonell and Joseph (1986), Nyberg (1988).
If a slot does not yet exist, add-to-value will create it.
The first digit also shows which word contributed what to the semantic representation. For example, the filler “SG-4” (for singular) in 3.3.5 has the word-number 4, because this aspect of the noun phrase was determined by word 4.
This distinction of frames representing different readings is based on a frame copying function which was written by Jaime Carbonell as an addition to the FrameKit software.
This sentence has the same meaning as The man read that a book was given to Mary.
The first digit of SENT-3_3_3 is “3” because this frame is derived from the third word in the sentence, i.e. read.
The third index digit in 3.4.5 is 4 rather than 3 because 3.4.5 is a new derivation, for the sake of demonstration, thus incrementing the sentence counter by 1.
In this way the number of readings is apparent when the first parse appears on the screen.
At present, LA-grammar treats all possible continuations as equal. However, as explained in Hausser (1986), p. 50, a weighing of readings based on the order of rules in the rule packages would be straightforward.
See Sections 5.4 and 5.5 for further discussion.
Berwick & Weinberg (1984), p. 41.
The computational complexity of LA-grammer is analyzed in Chapter 10. 33 See the analysis and discussion of a “garden path” sentence in Section 3.4.
Author information
Authors and Affiliations
Rights and permissions
Copyright information
© 1989 Springer-Verlag Berlin Heidelberg
About this chapter
Cite this chapter
Hausser, R. (1989). Outline of Left-Associative Grammar. In: Computation of Language. Symbolic Computation. Springer, Berlin, Heidelberg. https://doi.org/10.1007/978-3-642-74564-5_4
Download citation
DOI: https://doi.org/10.1007/978-3-642-74564-5_4
Publisher Name: Springer, Berlin, Heidelberg
Print ISBN: 978-3-642-74566-9
Online ISBN: 978-3-642-74564-5
eBook Packages: Springer Book Archive