Psychocomputational linguistics (original) (raw)
Related papers
Computational models of psycholinguistics
2008
Instead going straight into dealing with specific approaches, issues, and domains of computational cognitive modeling, it would be more appropriate to first take some time to explore a few general questions that lie at the very core of cognitive science and computational cognitive modeling. What is computational cognitive modeling? What exactly can it contribute to cognitive science? What has it contributed thus far? Where is it going? Answering such questions may sound overly defensive to the insiders of computational cognitive modeling, and may even seem so to some other cognitive scientists, but they are very much needed in a volume like this-because they lie at the very foundation of this field. Many insiders and outsiders alike would like to take a balanced and rational look at these questions, without indulging in excessive cheer-leading, which, as one would expect, happens sometimes amongst computational modeling enthusiasts. However, given the large number of issues involved and the complexity of these issues, only a cursory discussion is possible in this introductory chapter. One may thus view this chapter as a set of pointers to the existing literature, rather than a full-scale discussion. 1 1 What is Computational Cognitive Modeling? Research in computational cognitive modeling, or simply computational psychology, explores the essence of cognition (broadly defined, including motivation, emotion, perception, and so on) and various cognitive functionalities through developing detailed, process-based understanding by specifying corresponding computational models (in a broad sense) of representations, mechanisms, and processes. It embodies descriptions of cognition in computer algorithms and programs, based on computer science (Turing 1950). That is, it imputes computational processes (in a broad sense) onto cognitive functions, and thereby it produces runnable computational models. Detailed simulations are then conducted based on the computational models (see, e.g., Newell 1990, Rumelhart et al 1986, Sun 2002). Right from the beginning of the formal establishment of cognitive science around late 1970's, computational modeling has been a mainstay of cognitive science. 1 In general, models in cognitive science may be roughly categorized into computational, mathematical, or verbal-conceptual models (see, e.g., Bechtel and Graham 1998). Computational models (broadly defined) present process details using algorithmic descriptions. Mathematical models presents relationships between variables using mathematical equations. Verbal-conceptual models describe entities, relations, and processes in rather informal natural languages. Each model, regardless of its genre, might as well be viewed as a theory of whatever phenomena it purports to capture (as argued extensively before by, for example, Newell 1990, Sun 2005). 1 The roots of cognitive science can, of course, be traced back to much earlier times. For example, Newell and Simon's early work in the 60's and 70's has been seminal (see, e.g., Newell and Simon 1976). The work of Miller, Galanter, and Pribram (1960) has also been highly influential. See the chapter by Boden in this volume for a more complete historical perspective (see also Boden 2006).
Psycho-computational modelling of the mental lexicon
Word Knowledge and Word Usage
Over the last decades, a growing body of evidence on the mechanisms governing lexical storage, access, acquisition and processing has questioned traditional models of language architecture and word usage based on the hypothesis of a direct correspondence between modular components of grammar competence (lexicon vs. rules), processing correlates (memory vs. computation) and neuro-anatomical localizations (prefrontal vs. temporo-parietal perisylvian areas of the left hemisphere). In the present chapter, we explore the empirical and theoretical consequences of a distributed, integrative model of the mental lexicon, whereby words are seen as emergent properties of the functional interaction between basic, language-independent processing principles and the language-specific nature and organization of the input. From this perspective, language learning appears to be inextricably related to the way language is processed and internalized by the speakers, and key to an interdisciplinary understanding of such a way, in line with Tomaso Poggio's suggestion that the development of a cognitive skill is causally and ontogenetically prior to its execution (and sits "on top of it"). In particular, we discuss conditions, potential and prospects of the epistemological continuity between psycholinguistic and computational modelling of word learning, and illustrate the yet largely untapped potential of their integration. We use David Marr's hierarchy to clarify the complementarity of the two viewpoints. Psycholinguistic models are informative about how speakers learn to use language (interfacing Marr's levels 1 and 2). When we move from the psycholinguistic analysis of the functional operations involved in language learning to an algorithmic description of how they are computed, computer simulations can help us explore the relation between speakers' behavior and general learning principles in more detail. In the end, psycho-computational
Psychocomputational Models of Human Language Acquisition Proceedings of the Workshop
This workshop brings together scientists whose (at least one) line of investigation is to computationally model the process by which humans acquire various aspects of natural language. Progress in this agenda not only directly informs developmental psycholinguistic and linguistic research but will also have the long term benefit of informing applied computational linguistics in areas that involve the automated acquisition of knowledge from a human or human-computer linguistic environment. The scientific program consisted of two invited talks, one by Brian MacWhinney and another by Mark Steedman, and 10 paper presentations. We were especially pleased with the high quality of the submissions and would like to thank the authors for submitting their papers, as well as Mirella Lapata (ACL Workshop Committee Chair), Jason Eisner, Philipp Koehn (Publications Chairs) and Dragomir Radev (Local Arrangements Chair) who were extremely helpful (and patient) on more than one occasion. Abstract Sp...
Three Seductions of Computational Psycholinguistics
Descript.i ve linguisLs, compuLaLional linguisu;, and psycholinguist.s have traditionally been co ncerned with different aspects of the formal study of languag~. Linguists want ~xp licit grammatical formulations to charad.eri:.~e the well-formed senlences of a language and Lo indicaLe in some systema tic wa y how the sequence of clements th at makes up an utterance en codes Lhal ut.Lerance's m e aning. They don't. parLicularly care aboul specific processing algorithms that might be used to identify well-formed s~nt~n c~s or to asso <:iat~ t h~m with th~ir nwanings, bnt this is a c~ntral concern of compuLaLional linguists. Comput.aLional linguisls are inlerested in discovering the feasible algorithms that can interpret granunatical descriptions Lo recognize or produce ut.Leran ces, and in undersLanding how the performance of these algorithms depends on various properties of grammars and machin~ ar<:hit~ctnr~s. Psy<:holingni sts ar~ also conc~rn~d with processes and algorithms, bul noL jusl with ones LhaL are feasible within conventional computational architectures. They focus on algorithms and ar<:hit~ctnr~s that mod~l or ~h1cidat~ tlw langnag~ pro<:~ssing capabilities of human speakers and listeners.
Psycho-computational modelling of the mental lexicon A discriminative learning perspective
Word Knowledge and Word Usage. A Cross-Disciplinary Guide to the Mental Lexicon, 2020
Pirrelli, V., C. Marzi, M. Ferro, F. A. Cardillo, R. H. Baayen, and P. Milin Over the last decades, a growing body of evidence on the mechanisms governing lexical storage, access, acquisition and processing has questioned traditional models of language architecture and word usage based on the hypothesis of a direct correspondence between modular components of grammar competence (lexicon vs. rules), processing correlates (memory vs. computation) and neuro-anatomical localizations (prefrontal vs. temporo-parietal perisylvian areas of the left hemisphere). In the present chapter, we explore the empirical and theoretical consequences of a distributed, integrative model of the mental lexicon, whereby words are seen as emergent properties of the functional interaction between basic, language-independent processing principles and the language specific nature and organization of the input. From this perspective, language learning appears to be inextricably related to the way language is processed and internalized by the speakers, and key to an interdisciplinary understanding of such a way, in line with Tomaso Poggio's suggestion that the development of a cognitive skill is causally and ontogenetically prior to its execution (and sits "on top of it"). In particular, we discuss conditions, potential and prospects of the epistemological continuity between psycholinguistic and computational modelling of word learning, and illustrate the yet largely untapped potential of their integration. We use David Marr's hierarchy to clarify the comple-mentarity of the two viewpoints. Psycholinguistic models are informative about how speakers learn to use language (interfacing Marr's levels 1 and 2). When we move from the psycholinguistic analysis of the functional operations involved in language learning to an algorithmic description of how they are computed, computer simulations can help us explore the relation between speakers' behavior and general learning principles in more detail. In the end, psycho-computational models can be instrumental to bridge Marr's levels 2 and 3, bringing us closer to understanding the nature of word knowledge in the brain.
Computational linguistics and the study
I. Two Kinds of Models T begin with, I would like to assert that computational linguistics (henceforth CL), despite its qualifying adjective, has to do with human behavior and, in particular, with that subset of human behavioral patterns that we study in linguistics. In other words, one of the tasks of c L as a science is to explain human behavior insofar as it avails itself of the possibilities inherent in man's faculty of speech. In this sense, CL and linguistics proper both pursue the same aim. However, there are differences, as we will see shortly; for the moment, let us just establish that CU can be considered as a subfield of linguistics and leave the delineation of the boundaries for later. An important notion in behavioral sciences is that of a model as a set of hypotheses and empirical assumptions leading to certain testable conclusions called predictions (cf., e.g., Braithwaite 1968, ~aumjan 1966). I would like to call this kind of model the descriptive one. "Descriptive" here is not taken in the sense that Chomsky distinguishes descriptive adequacy from explanatory adequacy; indeed, the function of the descriptive model is to explain, as will become clear below. However, there is another respect in which the descriptive model reminds one of some of the characteristics attributed to Chomskyan models: it need not be (and should not be) considered a "faithful" reproduction of reality in the sense that to each part of the model there corresponds, by some kind of isomorphic mapping, a particular chunk of "real" life. In other words, this descriptive kind of model does not attempt to imitate the behavior of its descriptum. The other kind of model I propose to call the simulative one. As the name indicates, we are dealing with a conscious effort to picture, point by point, the activities that we want to describe. Of course, the simulative model, in order to be scientifically interesting, must attempt to explain; a machina loquax, to use Ceccato's expression (1967), is no good if there is a deus in machina. Although the idea of building homunculi, robots, and whatever else they are called is not exactly a new one, the advent of the computer made it possible to conduct these experiments on a hitherto unknown scale, both with regard to dimensions and to exactitude. In fact, one of the popular views of the computer is exactly that: a man-like machine. Interestingly, the fears connected with this
A state of the art in Computational Linguistics
Linguistics today: facing a greater challenge, 2004
Computational Linguistics has a long history, dating back to the Fifties, during which it developed a whole set of computational models and implementations, theories, methodologies and applications. It is difficult to give a sensible account of its present state without going back a little to the main steps through which this discipline evolved towards its present state. Since its origins, Computational Linguistics has been in an intermediate position between Computer Science and Artificial Intelligence, Linguistics and Cognitive Science, and Engineering. Computer Science itself shares its roots with Computational Linguistics; parsing, which is central for the design of compilers for programming languages (Aho and Ullmann 1977: 6), is also the building block of any natural language processing engine, and both are the realizations of the chomskian theory of formal languages (Chomsky 1957). The same theory, together with the corresponding computational model, has given a contribution to the general hypothesis of Artificial Intelligence, that human behaviours usually judged intelligent could be simulated in a computer in a principled way. Oversimplifying, Artificial Intelligence aims at modelling a number of behaviours through three very general paradigms, theorem proving, problem solving and planning, and language understanding and production. The history of both disciplines is rich in intersections, especially between language processing and planning, as in SHRDLU (Winograd 1971) or, more recently, in ARGOT (Allen et al. 1982, Allen 1983), with all its practical and theoretical follow-ups; modern dialogue systems in all their forms and applications are derived from the dialogue model J.Allen designed for ARGOT. The commitment to "simulation of behaviour", shared by Artificial Intelligence and and a relevant part of Computational Linguistics, makes them also share the effort for "cognitive modelling" of different human behaviours, including the use of language. This is probably one of the reasons why Linguistics appears in the set of sciences originally interested in the arising of the new discipline called Cognitive Science (www.cognitivesciencesociety.org). Since the Seventies, when language technology reached a state of maturity such as to allow the realization of some applications, Engineering has been interested in some of the language processing techniques, and it appeared soon that the approach introduced by engineers was certainly less theoretically and cognitively interesting, but more effective in many ways. By now, we can say that while Computational Linguists were, and are, more interested in the correctness and plausibility of their models, Engineers were, and are, more interested in the usability of tools and techniques, even
2014
This synopsis presents the application of computational linguistic tools and approaches which were developed by the author for Descriptive Linguistics, Text Mining, and Psycholinguistics. It also describes how the computational linguistic tools, which are originally based on linguistic insights and assumptions, lead to new and detailed linguistic insights if applied to different research areas, and can in turn again improve the computational tools. The computational tools are based on models of language, predicting part-of-speech tags or syntactic attachment. These models, which were originally designed for the practical purpose of solving a computational linguistics task, can increasingly be used as models of human language processing. A large-scale syntactic parser is the core linguistic tool that I am going to use. I further also employ its preprocessing tools, part-of-speech taggers and chunkers, and approaches learning from the data, so-called data- driven approaches. The use o...