Ambiguities in natural language can multiply so fast that no person or machine can be expected to process a text of even moderate length by enumerating all possible disambiguations. A sentence containing $n$ scope bearing elements which are freely permutable will have $n!$ readings, if there are no other, say lexical or syntactic, sources of ambiguity. A series of $m$ such sentences would lead to $(n!)^m$ possibilities. Some alternative scopings may boil down to the same reading. The relative order in which we scope two existentially quantified noun phrases, for example, will not matter if no other material intervenes. But all in all the growth of possibilities will be so fast that generating readings first and testing their acceptability afterwards will not be feasible.

We develop an extensible description logic for stating the content of optimalitytheoretic constraints in phonology, and specify a class of structures for interpreting it. The aim is a transparent formalisation of OT. We show how to state a wide range of constraints, including markedness, input–output faithfulness and base–reduplicant faithfulness. However, output–output correspondence and ‘intercandidate’ sympathy are revealed to be problematic: it is unclear that any reasonable class of structures can reconstruct their proponents’ intentions. But our contribution is positive. Proponents of both output–output correspondence and sympathy have offered alternatives that fit into the general OT picture. We show how to state these in a reasonable extension of our formalism. The problematic constraint types were developed to deal with opaque phenomena. We hope to shed new light on the debate about how to handle opacity, by subjecting some common responses to it within OT to critical investigation.

this article appeared in Glot, the main issue agitating researchers in model theoretic syntax was the problem of the context-free barrier. We have seen that the hierarchy of logics collapses, when applied to trees, at the border of the tree languages strongly generated by context free (string) grammars, in the sense that distinctions between the different tree logics reduce to apparently superficial distinctions in how much memory allocation is hidden in the logic. The problem which researchers set themselves was not just breaking the context free barrier but remaining decidable in the process. This is a very difficult problem, and it must be admitted right off that it is somewhat artificial in that there is no a priori reason to suppose that natural languages can be described in a decidable logic. The arguments on either side are something like the following. First, the rather slight increases in computational complexity required to get the "mildly context sensitive" languages do suggest that this might be possible. The hunch here would be that the qualities that characterize the mildly context sensitive languages (polynomial parsability, constant growth property) as being like the context-free languages are going to turn out to be reflections of decidability. The problems must not be underestimated, however! It is well known that the monadic second order logic of trees is one of the most powerful decidable logics known. It seems unlikely that any primitive relations can be added to the repertoire of tree description primitives that we have already seen, without making the logic undecidable. Many attempts have been made within logic and all have failed. So it is equally tempting to conjecture that the context-free boundary coincides in some deep sense with the bounda...

This thesis aims to develop a psychologically plausible account of concepts by integrating key insights from philosophy (on the metaphysical basis for concept possession) and psychology (on the mechanisms underlying concept acquisition). I adopt an approach known as informational atomism, developed by Jerry Fodor. Informational atomism is the conjunction of two theses: (i) informational semantics, according to which conceptual content is constituted exhaustively by nomological mind–world relations; and (ii) conceptual atomism, according to which (lexical) concepts have no internal structure. I argue that informational semantics needs to be supplemented by allowing content-constitutive rules of inference (“meaning postulates”). This is because the content of one important class of concepts, the logical terms, is not plausibly informational. And since, it is argued, no principled distinction can be drawn between logical concepts and the rest, the problem that this raises is a general one.

This paper presents a previously unnoticed universal property of stress patterns in the world’s languages: they are, for small neighborhoods, neighborhood-distinct. Neighborhood-distinctness is a locality condition defined in automata-theoretic terms. This universal is established by examining stress patterns contained in two typological studies, Bailey (1995) and Gordon (2002). Strikingly, many logically possible— but unattested—patterns do not have this property. Not only does neighborhood-distinctness unite the attested patterns in a non-trivial way, it also naturally provides an inductive principle allowing learners to generalise from limited data. A learning algorithm is presented which generalises by failing to distinguish same-neighborhood environments perceived in the learner’s linguistic input—hence learning neighborhood-distinct patterns—as well as almost every stress pattern in the typology. In this way, this work lends support to the idea that properties of the learner can explain certain properties of the attested typology, an idea not straightforwardly available in Optimality-theoretic and Principle and Parameter frameworks.

This article introduces and evaluates an approach to semantic compositionality in computational linguistics based on the combination of Distributional Semantics and supervised Machine Learning. In brief, distributional semantic spaces containing representations for complex constructions such as Adjective-Noun and Verb-Noun pairs, as well as for their constituent parts, are built. These representations are then used as feature vectors in a supervised learning model using multivariate multiple regression. In particular, the distributional semantic representations of the constituents are used to predict those of the complex structures. This approach outperforms the rivals in a series of experiments with Adjective-Noun pairs extracted from the BNC. In a second experimental setting based on Verb-Noun pairs, a comparatively much lower performance was obtained by all the models; however, the proposed approach gives the best results in combination with a Random Indexing semantic space. 1

This paper presents an analysis of certain aspects of Czech sentential clitics in Higher Order Grammar. I focus on the relative order of clitics within the clitic cluster. The overall aim of the paper is to show that constraints governing Czech sentential clitics, though quite complex, can be captured relatively easily within a higher order formalism such as Higher Order Grammar.

Abstract not found

. van den Bosch (eds.), Proceedings of the 7th Belgian-Dutch Conference on Machine Learning, BENELEARN-97. 1997. Bouma, H.H.W. (1997). Learning Dutch Phonotactics with Neural Networks. Master thesis, Alfa-informatica department, University of Groningen. 140 BIBLIOGRAPHY Cairns, Charles E. and Mark H. Feinstein (1982). `Markedness and the Theory of Syllable Structure'. In: Linguistic Inquiry, 13 (2), 1982. Charniak, Eugene (1993). Statistical Language Learning. MIT Press. Chomsky, Noam (1965). Aspects of the Theory of Syntax. MIT Press. Cleeremans, Axel (1993). Mechanisms of Implicit Learning. The MIT Press. Cleeremans, A., D. Servan-Schreiber and J.L. McClelland (1989). `Finite State Automata and Simple Recurrent Networks'. In: Neural Computation, , pp 372--381, 1989. Daelemans, Walter and Antal van den Bosch (1996). `Language-Independent DataOriente

In this note we define a machine that generates nests. The basic relations commonly attributed to linguistic expressions in configurational syntactic models as well as the device of chains postulated in current transformational grammar to represent distance relations can be naturally derived from the assumption that the combinatorial syntactic procedure is a nesting machine. Accordingly, the core of the transformational generative syntactic theory of language can be solidly constructed on the basis of nests, in the same terms as the general theory of order, an important methodological step that provides a rigorization of Chomsky’s minimalist intuition that the simplest way to generate hierarchically organized linguistic expressions is by postulating a combinatorial operation called Merge, which can be internal or external. Importantly, there is reason to think that nests are a useful representative tool in other domains besides language where either some recursive algorithm or evolutionary process is at work, which suggests the unifying force of the mathematical abstraction this note is based on.