Title: Finite State
Techniques in Computational Morphology and Computational Phonology
Lecturers: Dipl.-Inform. Anne Schiller (Xerox, Grenoble), Dr. Julie Carson-Berndsen (University College Dublin)
Course Description:
Finite state techniques are employed nowadays in many areas of computational linguistics. This course is concerned with two such areas, computational morphology and computational phonology, which together comprise the principal application domain of practical finite state techniques in computational linguistics.
The first part of the course provides an introduction to two-level morphology and different approaches for its realisation as finite-state automata. For practical sample applications, we will present and use the Xerox Finite-State Tool (xfst). The second part of the course is concerned with computational models of non-linear phonology. Finite state descriptions of phonotactic constraints (phonotactic automata) and their application in the area of speech technology will be presented. Phonotactic automata describe constraints on overlap and precedence relations between phonological features and are used to interpret multilinear representations of speech utterances. In this connection, methods for acquisition, representation and evaluation of such finite state models will be discussed.
The course will have both theoretical and practical components. Participants will be given the opportunity to develop and test their own descriptions.
References:
Carson-Berndsen, J. (2000). Finite State Models, Event Logics and Statistics in Speech Recognition. In: K.I.B. Spärck Jones, G.J.M. Gazdar & R.M. Needham (eds.). Computers, Language and Speech: Integrating formal theories and statistical data. Philosophical Transactions of the Royal Society, Series A, Volume 358, issue no. 1769, pp. 1255-1266.
[http://cspeech.ucd.ie/~cliste/group/publications/2000FSMEventLogStatsSpeechRecROYSOC.pdf]
Carson-Berndsen, J. & Walsh, M. (2000). Interpreting Multilinear
Representations in Speech. In the Proceedings of the 8th Australian Conference
on Speech Science and Technology, Canberra, December 2000: 472-477.
[http://cspeech.ucd.ie/~cliste/group/publications/2000InterpretMultilinRepSpeechSST.pdf]
Karttunen, Lauri & Kenneth R. Beesley (2001): A Short History of Two-Level Morphology, ESSLLI-2001 Special Event. Aug. 23, 2001, Helsinki, Finland. [http://www2.parc.com/istl/members/karttune/publications/esslli-2001/twol-history/twol-history.html]
Karttunen, Lauri (2000): Applications of Finite-State Transducers in Natural Language Processing, Proceedings of CIAA-2000. Lecture Notes in Computer Science. Springer Verlag.
[http://link.springer.de/link/service/series/0558/bibs/2088/20880034.html]
Karttunen, Lauri; Tamás Gaál, & André Kempe (1997):Xerox Finite-State Tool. Technical Report. Xerox Research Centre Europe.
[http://www.xrce.xerox.com/competencies/content-analysis/fssoft/docs/fst-97/xfst97.html]
1. Do the exercices presented with in the lectures about finite-state morphology (1 submission)
2. Do the exercices presented with in the lectures about finite-state phonology (1 submission)
3. Complete one (out of several) projects and submit the programs and a short report before 30 November.
Titel:
Finite-State Techniken in der Computermorphologie und Computerphonologie
Dozentinnen: Anne
Schiller, M.A. (Xerox, Grenoble), Dr. Julie Carson-Berndsen (University College
Dublin)
Kursbeschreibung:
Finite-state Techniken werden heutzutage in verschiedenen Bereichen der
Computerlinguistik eingesetzt. Dieser Kurs beschäftigt sich mit zwei solcher
Bereiche, die gemeinsam den Hauptanteil der praktischen finite-state
Anwendungen in der Computerlinguistik bilden.
Der erste Teil des Kurses gibt eine Einführung in die 2-Ebenen-Morphologie
und verschiedene Ansätze zur Realisierung als endlichen Automaten. Für
praktische Übungsbeispiele wird das Xerox Finite-State Tool (xfst) vorgestellt
und angewandt. Der zweite Teil des Kurses beschäftigt sich mit Computermodellen
der nicht-linearen Phonologie. Insbesondere werden finite-state Beschreibungen
der Phonotaktik, die sogenannten "Phonotaktischen Automaten", und
ihre Anwendungen in der Sprachtechnologie erläutert. Phonotaktische Automaten
beinhalten Überlappungs- und Präzedenzrelationen zwischen phonologischen
Merkmalen, die bei der Interpretation
multilinearer Repräsentationen von Sprachäußerungen eingesetzt werden können.
In diesem Zusammenhang werden Methoden zur Akquisition, zur Repräsentation und
zur Evaluierung solcher finite-state Modelle besprochen.
Der Kurs wird sowohl theorie- als auch praxisbezogen sein. Teilnehmern wird
die Gelegenheit geboten, eigene Beschreibung zu entwickeln und zu testen.
Literaturangaben:
Carson-Berndsen, J. (2000). Finite State Models, Event Logics and Statistics in Speech Recognition. In: K.I.B. Spärck Jones, G.J.M. Gazdar & R.M. Needham (eds.). Computers, Language and Speech: Integrating formal theories and statistical data. Philosophical Transactions of the Royal Society, Series A, Volume 358, issue no. 1769, pp. 1255-1266.
[http://cspeech.ucd.ie/~cliste/group/publications/2000FSMEventLogStatsSpeechRecROYSOC.pdf]
Carson-Berndsen, J. & Walsh, M. (2000). Interpreting Multilinear
Representations in Speech. In the Proceedings of the 8th Australian Conference
on Speech Science and Technology, Canberra, December 2000: 472-477.
[http://cspeech.ucd.ie/~cliste/group/publications/2000InterpretMultilinRepSpeechSST.pdf]
Karttunen, Lauri & Kenneth R. Beesley (2001): A Short History of Two-Level Morphology, ESSLLI-2001 Special Event. Aug. 23, 2001, Helsinki, Finland. [http://www2.parc.com/istl/members/karttune/publications/esslli-2001/twol-history/twol-history.html]
Karttunen, Lauri (2000): Applications of Finite-State Transducers in Natural Language Processing, Proceedings of CIAA-2000. Lecture Notes in Computer Science. Springer Verlag.
[http://link.springer.de/link/service/series/0558/bibs/2088/20880034.htm]
Karttunen, Lauri; Tamás Gaál, & André Kempe (1997):Xerox Finite-State Tool. Technical Report. Xerox Research Centre Europe.
[http://www.xrce.xerox.com/competencies/content-analysis/fssoft/docs/fst-97/xfst97.html]
1. ...