CALLHOME American English Lexicon (PRONLEX) Second Edition

                                 June 24, 2024

                          Linguistic Data Consortium


1. Overview
===========
This is an updated release of the CALLHOME American English Lexicon (LDC97L20),
which was originally compiled by the Linguistic Data Consortium in
support of the project on Large Vocabulary Conversational Speech Recognition
(LVCSR), sponsored by the U.S. Department of Defense. This re-release updates
the directory structure, file formats, and documentation to modern standards.


2. Directory structure
======================
- data/lexicon.tsv  --  lexicon in TSV format
- data/lexicon.dict  --  a pronunciation dictionary derived from lexicon;
  in CMUdict format
- docs/README.txt  --  this file; a top-level documentation of release
- docs/PRONUNCIATION.txt  --  a file documenting the transcription
  principles


3. lexicon.tsv
==============
This is a UTF-8 encoded TSV version of the lexicon originally distributed with
LDC97L20. It contains 90,988 lexical entries and includes coverage of WSJ30,
WSJ64, Switchboard, and CallHome English.

As specified in "PRONUNCIATION.txt", this release contains only
citation-form pronunciations, without notating predictable variation due to
dialect or variable reduction; furthermore, no attempt has been made to present
realistic pronunciations for function words (marked with #FUNC in the comment
field).  However, "PRONUNCIATION.txt" notes systematic dialectal variants which
may be generated by rule, and alternate pronunciations are given in the lexicon
for words whose pronunciation varies by part of speech (e.g., abstrAct,
Abstract), or in less systematic but salient ways (especially names).

The file format is UTF-8 encoded TSV with three tab-delimited fields:

- word  --  orthographic representation of word; e.g., "abacus"
- pron  --  pronunciation(s) of the word; multiple pronunciations are separated
  by "||".
- comments  -- (OPTIONAL) comment on the entry
  * comments should be comprehensive, but some are preliminary and qualified with
    a "?" at the end of the string
  * shorthand used in comments is documented in "docs/PRONUNCIATION.txt"

Neither words nor their pronunciations contain any internal white space.

The allophone set used is documented in "docs/PRONUNCIATION.txt". This file
notes systematic dialectal variants which may be generated by rule, and
alternate pronunciations are given in the lexicon for words whose
pronunciation varies by part of speech (e.g., abstrAct, Abstract), or in less
systematic but salient ways (especially names).


4. lexicon.dict
===============
This is a CMUdict format version of "lexicon.tsv". It consists of one
pronunciation per line, each line having the form:

    <WORD> <PRON>

where:

- WORD  --  orthographic representation of word
- PRON  --  a single pronunciation of the word, expressed as a sequence of
  phone symbols


4.1 Multiple pronunciations
---------------------------
In cases where a word has multiple pronunciations listed in "lexicon.tsv",
these are split into separate lines, one per pronunciation, in "lexicon.dict".


4.2 Phoneset
------------
Pronunciations consist of space-delimited sequences of LONG form versions of
the phones described in "PRONUNCIATION.txt". Stress is indicated following
vowels and may take on one of three values:

- 0  --  lack of stress
- 1  --  primary stress
- 2+  -- non-primary stress


5. Contacts
===========
If you have questions about this data release, please contact the following
LDC personnel:

    Neville Ryant
    <nryant@ldc.upenn.edu>