Dear All, but particularly Richard, who understands these things, and Mark,
who gets lumbered with implementing them...
I'm attaching version 1 of the IP grammar, which I thought I'd sorted, but
which turned out to be slightly hairier than expected in places when I came
to flesh out the notes. I'd be grateful if Richard had a chance to look it
over before Wed, because we may be using similar attributes in different
ways at different levels of the hierarchy, and this may or may not be a
good thing. (For example, headedness: see explanation in doc). I also
still need to identify the places where indexing is crucial.
In the light of this latest version, I don't think there is anything in the
ICSLP paper which is actually wrong, though some statements therein do not
pick out what I'd now see as the best combinations of defining features, so
you'll all hate me for confusing things.
Now I'll try to retrieve Heid and Hawkins. For some reason when I click on
it in Eudora it thinks I want to reply to the message, so I'll find another
way of opening the document!
Jill
--=====================_906392115==_
Content-Type: application/rtf; charset="us-ascii"
Content-Disposition: attachment; filename="IPgrammrtf.rtf"
{\rtf1\ansi \deff4\deflang1033{\fonttbl{\f1\froman\fcharset2\fprq2 Symbol;}{\f4\froman\fcharset0\fprq2 Times New Roman;}{\f15\fnil\fcharset2\fprq2 Wingdings;}}{\colortbl;\red0\green0\blue0;\red0\green0\blue255;
\red0\green255\blue255;\red0\green255\blue0;\red255\green0\blue255;\red255\green0\blue0;\red255\green255\blue0;\red255\green255\blue255;\red0\green0\blue128;\red0\green128\blue128;\red0\green128\blue0;\red128\green0\blue128;\red128\green0\blue0;
\red128\green128\blue0;\red128\green128\blue128;\red192\green192\blue192;}{\stylesheet{\widctlpar \f4\lang2057 \snext0 Normal;}{\*\cs10 \additive Default Paragraph Font;}}{\info{\title DRAFT}{\author J.House}{\operator J.House}
{\creatim\yr1998\mo9\dy18\hr17\min38}{\revtim\yr1998\mo9\dy21\hr17\min34}{\printim\yr1998\mo9\dy21\hr16\min25}{\version4}{\edmins295}{\nofpages4}{\nofwords742}{\nofchars4233}{\*\company Phonetics,UCL}{\vern57443}}\paperw11906\paperh16838
\widowctrl\ftnbj\aenddoc\formshade \fet0\sectd \linex0\headery709\footery709\colsx709\endnhere {\*\pnseclvl1\pnucrm\pnstart1\pnindent720\pnhang{\pntxta .}}{\*\pnseclvl2\pnucltr\pnstart1\pnindent720\pnhang{\pntxta .}}{\*\pnseclvl3
\pndec\pnstart1\pnindent720\pnhang{\pntxta .}}{\*\pnseclvl4\pnlcltr\pnstart1\pnindent720\pnhang{\pntxta )}}{\*\pnseclvl5\pndec\pnstart1\pnindent720\pnhang{\pntxtb (}{\pntxta )}}{\*\pnseclvl6\pnlcltr\pnstart1\pnindent720\pnhang{\pntxtb (}{\pntxta )}}
{\*\pnseclvl7\pnlcrm\pnstart1\pnindent720\pnhang{\pntxtb (}{\pntxta )}}{\*\pnseclvl8\pnlcltr\pnstart1\pnindent720\pnhang{\pntxtb (}{\pntxta )}}{\*\pnseclvl9\pnlcrm\pnstart1\pnindent720\pnhang{\pntxtb (}{\pntxta )}}\pard\plain \widctlpar \f4\lang2057 {
\b\fs28 An IP (supra-syllabic) grammar for ProSynth
\par
\par Version 1
\par
\par Jill House, 21/09/98
\par }{\fs28
\par
\par }{\b\fs28 1. Structure:
\par }\pard \widctlpar
\par \pard \widctlpar IP {\lang1024 {\field{\*\fldinst SYMBOL 224 \\f "Wingdings" \\s 12}{\fldrslt\f15\fs24}}} (ag){\sub 0-1 }(AG) {\b AG
\par }\pard \fi720\li720\widctlpar {\b
\par }\pard \widctlpar AG{\lang1024 {\field{\*\fldinst SYMBOL 224 \\f "Wingdings" \\s 12}{\fldrslt\f15\fs24}}} {\b F} (F)
\par \pard \widctlpar
\par \pard \widctlpar F {\lang1024 {\field{\*\fldinst SYMBOL 224 \\f "Wingdings" \\s 12}{\fldrslt\f15\fs24}}} {\b S[strong] }(S)
\par \pard \widctlpar
\par
\par ag {\lang1024 {\field{\*\fldinst SYMBOL 224 \\f "Wingdings" \\s 12}{\fldrslt\f15\fs24}}} f
\par
\par f {\lang1024 {\field{\*\fldinst SYMBOL 224 \\f "Wingdings" \\s 12}{\fldrslt\f15\fs24}}} S[weak] (S[weak])
\par
\par
\par {\i Explanation:
\par
\par }\pard \widctlpar Constituents in parentheses ( ) are optional, and theoretically unlimited in number, unless constrained with an index number.
\par \pard \widctlpar
\par Constituents in bold are the heads of the units of which they form part.
\par
\par \pard \widctlpar Lower case is used as shorthand for a \ldblquote degenerate\rdblquote unit which lacks a head constituent; we shall call such constituents \ldblquote light\rdblquote , or [-heavy]. Thus, ag = AG[-heavy], and is alway
s coterminous with f, which = F[-heavy]. Note that at syllable level, the notion of weight will have a different connotation.
\par \pard \widctlpar
\par \pard \widctlpar When used as an attribute below, [+/- head] relates to the unit\rquote s role at that level in the hierarchy, not to its internal constituency.
So an AG which is [+head] is the head of its IP, and a F which is [+head] is the head of its AG.
\par \pard \widctlpar
\par {\b
\par {\pntext\pard\plain\fs28 2. \tab}}\pard \fi-283\li283\widctlpar{\*\pn \pnlvlbody\pndec\pnb0\pni0\pnf4\pnfs28\pnstart2\pnindent283\pnhang{\pntxta . }}{\b\fs28 Attributes
\par }\pard \widctlpar {\fs28
\par }\pard \widctlpar IP
\par \pard \widctlpar Currently the largest domain in the hierarchy; may ultimately be assigned attributes relating to position in turn, or other discourse unit, and discourse properties (such as a properly phonological characterisation of pitch pattern).
\par \pard \widctlpar
\par \pard \widctlpar Needs attribute which indicates how many AGs it contains:
\par {\pntext\pard\plain\f1 \'b7\tab}\pard \fi-283\li1003\widctlpar{\*\pn \pnlvlblt\pnf1\pnstart1\pnindent283\pnhang{\pntxtb \'b7}}AG no: \tab \tab (n)
\par \pard \widctlpar
\par AG/ag
\par Attributes relate to position, strength, weight, tonal properties and number of feet.
\par {\pntext\pard\plain\f1 \'b7\tab}\pard \fi-283\li1003\widctlpar{\*\pn \pnlvlblt\pnf1\pnstart1\pnindent283\pnhang{\pntxtb \'b7}}strength\tab \tab [+/-strong]
\par {\pntext\pard\plain\f1 \'b7\tab}weight\tab \tab [+/-heavy]
\par {\pntext\pard\plain\f1 \'b7\tab}headedness\tab \tab [+/-head]
\par {\pntext\tab}\pard \li720\widctlpar{\*\pn \pnlvlcont\pnf1\pnstart1\pnindent283\pnhang{\pntxtb \'b7}}
\par {\pntext\pard\plain\f1 \'b7\tab}\pard \fi-283\li1003\widctlpar{\*\pn \pnlvlblt\pnf1\pnstart1\pnindent283\pnhang{\pntxtb \'b7}}nucleus \tab \tab [+/-nucleus]
\par {\pntext\pard\plain\f1 \'b7\tab}onset\tab \tab [+/-onset]
\par {\pntext\pard\plain\f1 \'b7\tab}position in IP:\tab pre-initial (=ag) (0)
\par \pard \fi720\li2160\widctlpar initial (=onset) (1)
\par medial (2 - n)
\par final (=nuclear)
\par {\pntext\pard\plain\f1 \'b7\tab}\pard \fi-283\li1003\widctlpar{\*\pn \pnlvlblt\pnf1\pnstart1\pnindent283\pnhang{\pntxtb \'b7}}number of feet\tab (n)
\par {\pntext\pard\plain\f1 \'b7\tab}pitch accent\tab \tab select from paradigm: (H, HL, L, LH, \endash interim set!)
\par {\pntext\pard\plain\f1 \'b7\tab}boundary tone\tab H% or L%
\par \pard \widctlpar
\par Constraints and dependencies:
\par {\pntext\pard\plain\f1 \'b7\tab}\pard \fi-283\li1003\widctlpar{\*\pn \pnlvlblt\pnf1\pnstart1\pnindent283\pnhang{\pntxtb \'b7}}Only the ag is [-heavy]; all other AGs are [+heavy]
\par {\pntext\pard\plain\f1 \'b7\tab}Only the ag can be pre-initial (0)
\par {\pntext\pard\plain\f1 \'b7\tab}Only the final AG is [+head]
\par {\pntext\pard\plain\f1 \'b7\tab}Only the final AG is [+nucleus]
\par {\pntext\pard\plain\f1 \'b7\tab}\pard \fi-283\li1003\widctlpar{\*\pn \pnlvlblt\pnf1\pnstart1\pnindent283\pnhang{\pntxtb \'b7}}Only the [+nucleus] AG and the pre-initial ag normally have a boundary tone
\par {\pntext\pard\plain\f1 \'b7\tab}\pard \fi-283\li1003\widctlpar{\*\pn \pnlvlblt\pnf1\pnstart1\pnindent283\pnhang{\pntxtb \'b7}}Only the initial ([+strong, 1]) AG is [+onset]
\par {\pntext\pard\plain\f1 \'b7\tab}An AG may be simultaneously [+onset, +nucleus]
\par \pard \widctlpar
\par {\i Example:}
\par
\par \tab \tab \tab \tab \tab IP (4 x AGs)
\par
\par \pard \fi720\widctlpar \tab \tab [I\rquote m] [trying to] [find the pho] [netics department]
\par \pard \widctlpar \tab \tab \tab ag\tab AG\tab \tab AG\tab \tab AG
\par strength [+/-strong]\tab -\tab +\tab \tab +\tab \tab +
\par weight [+/- heavy] \tab -\tab +\tab \tab +\tab \tab +
\par headedness\tab \tab -\tab - \tab \tab -\tab \tab +
\par
\par initial\tab \tab \tab -\tab +\tab \tab -\tab \tab -
\par onset\tab \tab \tab -\tab +\tab \tab -\tab \tab -
\par
\par final\tab \tab \tab -\tab -\tab \tab -\tab \tab +
\par nucleus\tab \tab -\tab -\tab \tab -\tab \tab +
\par
\par position: no.\tab \tab 0\tab 1\tab \tab 2\tab \tab 3
\par no. of feet\tab \tab 1\tab 1\tab \tab 1\tab \tab 2
\par
\par pitch accent\tab \tab -\tab H\tab \tab H\tab \tab HL
\par boundary tone\tab \tab L%\tab -\tab \tab -\tab \tab L%\tab
\par
\par
\par {\b Notes:}
\par \pard \widctlpar There is some obvious redundancy here: initial/onset and final/nucleus. It
may not be necessary to retain both attributes for these positions, though one relates to structural position and the other to consequent properties of that position.
\par \pard \widctlpar
\par \pard \widctlpar Strength and weight are also apparently interchangeable at this level. In our ICSLP paper, I was using the feature [strong] to identify \ldblquote pre-head\rdblquote material, which is why I have left it in here.
However, the distinction between [strong] and [heavy] is needed at Foot level, with [-heavy] being needed for the degenerate foot (pre-head material). For consistency the AG attribute would be better analysed as [-heavy].
\par \pard \widctlpar
\par Tonal specifications will be adjusted to conform with empirical evidence from data.
\par
\par
\par F/f
\par Attributes differentiate strength, headedness, weight, position, no. of syllables
\par {\pntext\pard\plain\f1 \'b7\tab}\pard \fi-283\li1003\widctlpar{\*\pn \pnlvlblt\pnf1\pnstart1\pnindent283\pnhang{\pntxtb \'b7}}strength\tab \tab +/-
\par {\pntext\pard\plain\f1 \'b7\tab}headedness\tab \tab +/-
\par {\pntext\pard\plain\f1 \'b7\tab}weight\tab \tab +/-
\par {\pntext\pard\plain\f1 \'b7\tab}position in AG:\tab (1 - n)
\par {\pntext\pard\plain\f1 \'b7\tab}no of syllables\tab (1 - n)
\par \pard \widctlpar
\par Constraints and dependencies
\par {\pntext\pard\plain\f1 \'b7\tab}\pard \fi-283\li1003\widctlpar{\*\pn \pnlvlblt\pnf1\pnstart1\pnindent283\pnhang{\pntxtb \'b7}}Only the f is [-heavy]; all Fs are by definition [+heavy]
\par {\pntext\pard\plain\f1 \'b7\tab}Feet which are [+strong] are normally rhythmically stressed
\par {\pntext\pard\plain\f1 \'b7\tab}The leftmost F in each AG is [+head]
\par {\pntext\pard\plain\f1 \'b7\tab}The [+head] F carries the pitch accent information assigned at AG level
\par \pard \widctlpar
\par {\i Example:
\par }\tab \tab \tab ag\tab \tab AG\tab \tab AG
\par
\par \pard \fi720\widctlpar \tab \tab [in the] [well] [-known] [maths de] [partment]
\par \pard \widctlpar
\par \tab \tab \tab f\tab F F\tab \tab F\tab F
\par
\par strength [+/-strong] -\tab +\tab -\tab \tab +\tab +
\par weight\tab [+/-heavy] \tab -\tab +\tab +\tab \tab +\tab +
\par headedness [+/-head] \tab -\tab +\tab -\tab \tab +\tab -
\par
\par position in AG\tab \tab 1\tab 1\tab 2\tab \tab 1\tab 2
\par no of syllables\tab \tab 2\tab 1\tab 1\tab \tab 2\tab 2
\par
\par
\par {\b Notes:
\par }\pard \widctlpar The use of both strength and weight has been necessary to distinguish between cases like \ldblquote -known\rdblquote and \ldblquote -part-\rdblquote above. Rhythmic beats would be expected on \ldblquote well\rdblquote , \ldblquote
maths\rdblquote and \ldblquote part\rdblquote , but not \ldblquote known\rdblquote . Under some analyses, \ldblquote well-known\rdblquote would constitute a single foot (=timing unit), whereas in others the vowel quality and syllable weight would e
nsure an analysis into two feet , which may provide more reliable domains for f0 realisation. It is hoped that with the current set of attributes, we can identify both types of foot as necessary.
\par \pard \widctlpar
\par \pard \widctlpar Feet that have been hitherto parsed as being of the type \ldblquote s, s\rdblquote (containing two strong syllables), will need to be reanalysed as two F, the first F strong and the second F weak, each containing one strong syllable.
In this system, internal foot structure can only be S[+strong], (S[-strong]), i.e. \ldblquote s, (w)\rdblquote , using the shorthand where s = strong and w = weak.
\par \pard \widctlpar
\par \tab \tab \tab AG\tab \tab AG\tab
\par \tab \tab \tab | \\\tab \tab | \\
\par \tab \tab \tab F(s)\tab F(w)\tab F(s)\tab F(s)
\par \tab \tab \tab |\tab |\tab | \\\tab | \\\tab
\par \tab \tab \tab S(s)\tab S(s)\tab S(s)S(w) S(s)S(w)\tab
\par
\par \tab \tab \tab well - known maths de part ment
\par {\b
\par }}
--=====================_906392115==_
Content-Type: text/plain; charset="us-ascii"
--=====================_906392115==_--