Newsgroups: comp.speech
Path: pavo.csi.cam.ac.uk!doc.ic.ac.uk!agate!howland.reston.ans.net!darwin.sura.net!rouge!ek
From: ek@swamp.cacs.usl.edu (Enrique Vicente Kortright)
Subject: From phonemes to English words and sentences
Message-ID: <1993Nov4.212338.10512@usl.edu>
Sender: anon@usl.edu (Anonymous NNTP Posting)
Organization: The Center for Advanced Computer Studies
Distribution: comp.ai.nat-lang, comp.speech
Date: Thu, 4 Nov 1993 21:23:38 GMT
Lines: 53

My colleague, Wan Jiancheng, and I are interested in recent work
regarding the following topics. If it is preferable, e-mail me
responses and I will summarize to the net.

- About English words:
  - Can all words be represented by the international phonetic system?
  - How many different phonemes need to be considered for practical English
    representation?

- About recognition of English sentences from strings of phonemes and speech:
  - What are the most relevant references on the subject?
  - Are there any software systems that do this?
    - What is their coverage of English grammar: words, phrases, sentences?
  - Are syllables used in the recognition process?
  - How are word boundaries found?
    - by the length of zero signal (pause)?
    - by left-to-right maximum match?
    - other methods?
  - What is the current hit-rate of the recognition of syllables and words?
  - What are the current problems in this field?
  - Can whole sentences be recognized? What is the hit-rate? Problems?
  - How do substrings affect recognition, especially for words in sentences?
  - How big is the vocabulary that current techniques can recognize?
  - How does the existence of homophones affect recognition?

- About English grammar:
  - What is the status of formal descriptions of English grammar?
    - lexical, syntactic, and semantic descriptions.
    - what are the most relevant references describing these formalisms?
  - How are words that play different grammatical roles handled formally?
    - e.g., the word 'play' can be used as a noun or as a verb.
  - Are there any special syntax structures that can be used as
    heuristics? Can they be described formally?
    - e.g., the-earlier-the-better, getting-better-and-better,
      getting-higher-and-higher.
  - Are there any studies of the complexity differences between phrases,
    sentences, and compound sentences?
  _ Can all syntactical attributes of English be described formally?
    - can they all be parsed?
  - Is there a public-domain (or commercially available) English dictionary
    which would be suitable for this type of work? (in electronic form)

Prof. Wan has worked on a similar project for the Chinese language and we
are looking for ways to apply some of his results to English.

Regards,

Enrique V. Kortright               Wan Jiancheng
Computer Science Dept.             Computer Science Dept.
Nicholls State University          Shandong University of Technology
Thibodaux, LA 70310                Jinan, Shandong 250014, P.R. China
ek@reality.nich.edu
(504) 448-4406
