Moving towards a Phon Query Language

[ghedlund]

Current query filter options are general, but often confusing. These general filters should be expanded to include additional 'canned' filters with tighter options.

Query filters are categorized by domain: group, word, or syllable. The query domain is defined by the existence of a filters. For backwards compatibility, the current filters for each domain should be retained (but hidden to the user as 'advanced' options.) In place, one ore more statements of the form: Groups/Word/Syllable must contain X, Group/Word/Syllable must have X sub-elements, etc.

The following is a list of possible statements:

• Group
  • Group must equal X, X is a filter statement (plain text, regex, or phonex, word pattern, stress pattern)
  • Group must contain X, X is a filter statement (plain text, regex, or phonex, word pattern, stress pattern)
  • Group must have X words, X is a number
  • Group must have at least X words, X is a number
  • Group must have no more than X words, X is a number
  • Group must have between X and Y words, X and Y are numbers
  • Word
  • Word must equal X, X is a filter statement (plain text, regex, or phonex, word pattern, stress pattern)
  • Word must contain X, X is a filter statement (plain text, regex, or phonex, word pattern, stress pattern)
  • Word must have stress pattern X, X is a stress pattern (could also be done with a pattern filter? should we provide a list of common stress patterns?)
  • Word must have X syllables, X is a number
  • Word must have at least X syllables, X is a number
  • Word must have no more that X syllables, X is a number
  • Word must have between X and Y syllables, X is a number
  • Word must be first in group
  • Word must be last in group
  • Word must be in middle of group
  • Word must be singleton
• Syllable
  • Syllable must equal X, X is a filter statement (plain text, regex, or phonex, word pattern, stress pattern)
  • Syllable must contain, X is a filter statement (plain text, regex, or phonex, word pattern, stress pattern)
  • Syllable must be stressed
  • Syllable must be primary stressed
  • Syllable must be secondary stressed
  • Syllable must have CV pattern X (maybe a list of common CV patterns should be provided as this is covered by the pattern filter?)
  • Syllable must be first in word
  • Syllable must be last in word
  • Syllable must be in middle of word
  • Syllable must be singleton

(Please edit list above as necessary)

[ghedlund]

This could be seen as moving towards a query language: (e.g.,)

// Single session context
FIND PHONEX /\c/ IN
SELECT "IPA Target"
WHERE GROUP {
    CONTAINS "p"
}
WHERE WORD {
    "IPA Actual" CONTAINS PHONEX /\c\v/
}
WHERE SYLLABLE {
        HAS PRIMARY STRESS
}
INCLUDE ALIGNED PHONES
INCLUDE "IPA Actual (Word)", "Alignment (Word)"
INCLUDE PCC()

// Single session context with record selection
FIND PHONEX /\c/ IN
SELECT "IPA Target"
WHERE GROUP {
    "IPA Target" CONTAINS "p"
}
WHERE WORD {
    "IPA Actual" CONTAINS PHONEX /\c\v/
}
WHERE SYLLABLE {
        HAS PRIMARY STRESS
}
FROM 1, 2, 3..9
INCLUDE ALIGNED PHONES
INCLUDE "IPA Actual (Word)", "Alignment (Word)"

// project context selecting all sessions in a corpus
FIND * IN
SELECT "IPA Target"
FROM "Corpus.*"
WHERE SPEAKER {
        IS FEMALE
        AND ROLE EQUALS "Target Child"
}
INCLUDE ALIGNED PHONES
INCLUDE "IPA Actual (Word)", "Alignment (Word)"

// free context selecting from files
FIND * IN
SELECT "IPA Target"
FROM "path/to/session.xml", "path/to/chatfile.cha"
INCLUDE ALIGNED PHONES
INCLUDE "IPA Actual (Word)", "Alignment (Word)"