Fred Karlsson

Constraints on multiple center-embedding of clauses

A common view in theoretical syntax and computational linguistics holds that there are no grammatical restrictions on multiple center-embedding of clauses. Syntax would thus be characterized by unbounded recursion. An analysis of 119 genuine multiple clausal center-embeddings from seven ‘Standard Average European’ languages (English, Finnish, French, German, Latin, Swedish, Danish) uncovers usage-based regularities, constraints, that run counter to these and several other widely held views, such as that any type of multiple self-embedding (of the same clause type) would be possible, or that self-embedding would be more complex than multiple center-embedding of different clause types. The maximal degree of center-embedding in written language is three. In spoken language, multiple center-embedding is practically absent. Typical center-embeddings of any degree involve relative clauses specifying the referent of the subject NP of the superordinate clause. Only postmodifying clauses, especially relative clauses and that -clauses acting as noun complements, allow central self-embedding. Double relativization of objects ( The rat the cat the dog chased killed ate the malt ) does not occur. These corpus-based ‘soft constraints’ suggest that full-blown recursion creating multiple clausal center-embedding is not a central design feature of language in use. Multiple center-embedding emerged with the advent of written language, with Aristotle, Cicero, and Livy in the Greek and Latin stylistic tradition of ‘periodic’ sentence composition.

SWETWOL: A Comprehensive Morphological Analyser for Swedish

SWETWOL is implemented in the framework of Koskenniemis (1983) two-level model. It contains a 48,000 item lexicon and a full inflectional description. Special attention was paid to the design of a computational analysis of productive Swedish compounds. Recall (coverage) and precision of SWETWOL meet high standards. SWETWOL has been extensively tested on various types of texts.

The formalism and environment of Constraint Grammar Parsing

This is a full documentation of the Constraint Grammar Parser CGP. The emphasis is on describing how a Constraint Grammar for a particular language is designed and developed. The general motivation, aims, and properties of Constraint Grammar Parsing were expounded in Chapter 1 (also cf. Karlsson 1989, 1990). Here, in Chapter 2, all aspects of the parsing formalism are treated. A full survey is given of the details that need to be known for properly applying Constraint Grammar to a language, and for running the parser once a description of a language is available. The basic purpose of this chapter is thus to act as a users manual. Most of the examples in this and the subsequent chapters will be drawn from the English Constraint Grammar description (ENGCG) done within the framework of the ESPRIT-II project “Structured Information Management, Processing, and Retrieval” (SIMPR). One of the aims of Constraint Grammar Parsing is to make it possible to parse unrestricted text. It is important to have a clear view of the sequencing of the whole parsing process, especially of what modules precede Constraint Grammar Parsing proper. Full-scale optimal Constraint Grammar Parsing proceeds by application of five consecutive modules:

Finnish Syntax in Text: Methodology and Some Results of a Quantitative Study

The Paper contains a report on the theory and methodology used in a project on finnish syntax as investigated in (some genres of written) text. 123 texts containing 10,149 clauses (5016 graphical sentences) were analyzed according to a syntactic coding frame containing 63 variables. The variables include sentences structure, clause structure, clause type, clause function, clause length, surface word order, constituent structure, the text semantic and referential properties of nominal constituents, and some movement and deletion transformations. The theoretical considerations behind the selection of the variables are discussed in detail, as are the principles for syntactic sampling. The usefulness of the computer for certain low-level tasks is demonstrated, such as finding correlations between variables. The relevance of quantitative data to ‘qualitative’ syntactic description is stressed.

Recursion in natural languages

The received view is that there are no grammatical constraints on clausal embedding complexity in sentences in languages of the ‘Standard Average European’ (SAE) type like English, Finnish, and Russian. The foremost proponent of this thesis is Noam Chomsky. This hypothesis of unbounded clausal embedding complexity is closely related to the hypothesis of unbounded syntactic recursion. Psycholinguistic experimentation in the 1960’s established that there are clear performance-related preferences especially regarding center-embedding. The acceptability of repeated center-embeddings (nesting) below depth 1 steeply decreases with each successive level of embedding. Not much corpus-based work has been done to find out what the empirical ‘facts’ of clausal embedding complexity are. I have conducted extensive corpus studies of English, Finnish, German, Latin, and Swedish, with the aim of determining the most complex clausal embedding patterns actually used. The basic constraint on nested center-embedding in written language turns out to be two (with a marginal cline to three), in spoken language one. There are further specific restrictions on which types of clauses may be nested. The practical limit of final embedding (right-branching) is five. Repeated initial embedding (left-branching) of clauses below depth two is not possible. These written language constraints were reached already in Sumerian, Akkadian, and Latin along with the advent of written language and have remained the same ever since. The constraints on center-embedding imply that SAE syntax is finite-state, type 3 in the Chomsky hierarchy. Clause-level recursion is thus not unbounded. The special case of right-branching relative clauses is rather an instance of depth-preserving iteration.

Is There a Crisis in Generative Linguistics

Several recent critiques have claimed that the empirical foundations of generative linguistics are weak due to the reliance on informally gathered grammaticality judgments drawn from the intuitions of the researcher. Phillips (In: Japanese/Korean linguistics, vol. 17, 2009) argued i.a. that, in order for there to be a theoretical crisis, two criteria should be fulfilled, namely (i) intuitive judgments have led to generalizations that are widely accepted yet bogus, and (ii) misleading judgments form the basis of important theoretical claims or debates. Furthermore Phillips claimed that (i, ii) have not been fulfilled. I argue that (i, ii) have in fact been satisfied because Chomsky’s (Q. Prog. Rep.—Mass. Inst. Technol., Res. Lab. Electron. 41:64–65, 1956; Syntactic structures. Mouton, The Hague, 1957) intuition-based claim that English is not a finite-state language is demonstrably false.

Finnish as an Agglutinating Language

A theoretically ideal agglutinative language has no inflectional classes, several morphotactic positions for affixes of composite word forms, clearly segmentable morphological elements, monofunctional affixes, and no morphophonological alternations in any element due to morphological processes such as affixation. Finnish detracts from this theoretical ideal somewhat as regards the occurrence of numerous nominal and verbal inflectional classes, allomorphy among the affixes, morphophonological alternations, endings expressing composite grammatical functions, and fusion of certain grammatical elements.