% % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % %
%                                                                           %
% File:            dakota.dtr                                               %
% Purpose:         Dakota verbal morphology and phonology                   %
% Author:          Chris Kennedy, November 28, 1994                         %
% Email:           kennedy@ling.ucsc.edu                                    %
% Address:         Stevenson College, UCSC, Santa Cruz, CA 95064, USA       %
% Version:         2.00                                                     %
%                                                                           %
%        Copyright (c) UC Santa Cruz, 1994.  All rights reserved.           %
%                                                                           %
% % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % %

# atom A C Z S ( ) { } .

# atom ARRIVE BE_DRY BE_HOT BE_LOST BE_OLD KILL STAGGER WALK.

%   Dakota is a member of the Siouan family spoken in the Northern Great
%   Plains region of the United States and Canada.
%
%   The DATR theory developed below (version 2.00) represents the verbal
%   morphology paradigm for two specific classes of verbs: non-infixing
%   stative and active intransitives.  Subsequent versions of this theory
%   will be extended to represent the morphology of non-infixing
%   transitives, and eventually both non-infixing and infixing verbs.
%
%   Some basic characteristics of verbal morphology:
%
%   Dakota indicates both object and subject agreement.  The stative
%   intransitives are marked for person with the object agreement
%   affixes; the active intransitives are marked for person with the
%   subject agreement affixes.  Agreement affixes are prefixes that
%   adjoin to Stem constituents.  Plurality is indicated by a suffixal
%   clitic that adjoins to a Word constituent.  Future tense and negation
%   are also indicated by suffixal clitics.
%
%   The function of the Verb node is to state the agreement affixes and
%   to determine the morphological constituency of the various forms.  As
%   noted above, while agreement affixes attach to Stems, the suffixes that
%   indicate plurality, future tense, and negation attach to Word
%   constituents.

Verb:
    <> ==

    <mor form> ==  < "<syn pol>"
                     "<syn tense>"
                     "<syn num>"
                     "<syn per>" >

    <phn form> == Syllabify:<HeadProj1:<"<mor form>">>

%   The following equation is simply to allow compact theorem dumps.
%   It will need to be updated when active transitive verbs are added
%   to the fragment since they require two sets of per/num values.

    <syn form> ==  { "<syn pol>"
                     "<syn tense>"
                     "<syn num>"
                     "<syn per>" }
    <syn cat> == verb

    <one sub> == w a
    <one obj> == m a
    <two sub> == y a
    <two obj> == n i
    <three> ==
    <incl> == un k
    <excl> == <incl>

%   The square brackets enclose a word constituent.
%   Stem constituency is not explicitly marked.

    <sing> == [ <> "<root>" ]

%   Plural forms are constructed on the basis of singular forms:

    <plur> == [ <sing> p i ]
    <plur one> == [ <sing incl> p i ]

%   Dual forms correspond to singular or plural forms depending
%   on whether they are inclusive or exclusive:

    <dual incl> == <sing incl>
    <dual excl> == <plur one>

%   Negative and future forms are marked by the adjunction of a
%   suffixal clitic:

    <neg> == [ <> S n i ]
    <fut> == [ <> k t e ].

Vactive_intrans:
    <> == Verb
    <mor form> == Verb:<mor form sub>.

Vstative:
    <> == Verb
    <mor form> == Verb:<mor form obj>.

% Vactive_trans:
%   <> == Verb
%   <mor form> == ??
%   <plur three obj> == w i C a.

% % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % %

%   The following is a list of lexical items.  Lexical items inherit
%   their verbal characteristics from the relevant Vx (x in {stative,
%   active_intrans, active_trans}) node.

Ceka:
    <> == Vactive_intrans
    <sem> == STAGGER
    <root> == C e k.

Khata:
    <> == Vstative
    <sem> == BE_HOT
    <root> == kh a t.

Hi:
    <> == Vactive_intrans
    <sem> == ARRIVE
    <root> == h i.

Puza:
    <> == Vstative
    <sem> == BE_DRY
    <root> == p u z.

Mani:
    <> == Vactive_intrans
    <sem> == WALK
    <root> == m a n i.

Nuni:
    <> == Vstative
    <sem> == BE_LOST
    <root> == n u n i.

Thani:
    <> == Vstative
    <sem> == BE_OLD
    <root> == th a n i.

% Kte:
%   <> == Vactive_trans
%   <sem> == KILL
%   <root> == k t e.

% % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % %
%
%   Head projection in Dakota
%
%   The DATR theory developed here is designed to represent "head
%   projection" in Dakota.  Head projection describes the relation
%   between a lexical head--a morphological constituent of bar level zero
%   (which I will refer to as a "Word")--and the head of the prosodic
%   word: the most prominent syllable.  The relevant generalization in
%   Dakota can be stated as follows: every lexical head must project a
%   prosodic head.  This generalization describes the following facts:
%   
%     (1) in general, Dakota words bear a single stress on the
%         second syllable;
%     (2) monosyllabic forms and monosyllabic CVC roots that surface
%         with a final epenthetic vowel bear initial syllable stress;
%     (3) compounds consisting of concatenated Words bear two stresses,
%         one on each member of the compound;
%     (4) affixes adjoined to a Stem may bear stress but clitics
%         adjoined to Word (e.g., pi 'plural', Sni 'negation', 
%         kte 'future') may not; complex morphemes consisting of
%         monosyllabic Word plus clitic bear stress on the first syllable.  
%    
%   The theory described below accounts for these facts in the following
%   way: the HeadProj1 node takes as input a fully specified
%   morphological structure (see above for Dakota verb morphology) and
%   determines which [-cons] segment projects the prosodic head.  This
%   segment is marked with a diacritic '!'.  The theory is stated in the
%   form of a finite state transducer that is sensitive to the
%   morphological structure of its input.  The machine reads across the
%   input and, in the standard case, marks the second [-cons] segment as
%   the locus of head projection.  If the input contains a monosyllabic
%   Word constituent, however, the initial syllable is marked.  Head
%   projection is noniterative.
%    
%   This statement of stress-assignment differs from standard accounts,
%   where the location of the stressed vowel at the segmental level is
%   determined by foot structure.  The Dakota facts indicate that the
%   relation between the lexical head and the prosodic head must be more
%   directly stated: the theory of head projection exemplified here is
%   one way to do this.  

# vars $vowl: a e i o u an in un A.
# vars $con0: ph th kh p t k b d g c s z h y w q n m l x r C Z S.
% Following line a hack to cope with QDATR variable definition glitch
# vars $con1: ph th kh p t k b d g c s z h y w q n m l x r C Z S.

HeadProj1:
    <$char> == $char <>
    <$vowl $char> == $vowl HeadProj2:<$char>.

HeadProj2:
    <[> == [ <>
    <]> == ! ] HeadProj3:<>
    <$vowl> == $vowl ! HeadProj3:<>
    <$con0> == $con0 <>
    <$con0 ]> == ! $con0 ] HeadProj3:<>.

HeadProj3:
    <> ==
    <$char> == $char <>
    <[ $char> == [ HeadProj1:<$char>.

%   Note that the Node:path 'HeadProj3:<[ $char>' allows sequences of
%   Word constituents (as in compounds) to be submitted as input.  Each
%   Word constituent will be submitted to evaluation at HeadProj1.
%
%   Some sample theorems:
%   
%   HeadProj1:<[ m a n i ]> = [ m a n i ! ].
%   HeadProj1:<[ C e k ]> = [ C e ! k ].
%   HeadProj1:<[ [ C e k ] p i ]> = [ [ C e ! k ] p i ].
%   HeadProj1:<[ [ [ h i ] k t e ] S n i ]> = [ [ [ h i ! ] k t e ] S n i ].
%
% % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % %
%
%   A syllable parser for Dakota
%   
%   The Syllabify node presented below is a syllable parser for Dakota
%   which produces as output a syllabified morpheme that does not show
%   morphological constituency, but does indicate main stress (i.e., the
%   head of the Prosodic Word) if this was marked in the input (e.g., by
%   using the HeadProj1 node as the source of that input).  Syllabify is
%   stated as a finite state transducer and is sensitive to the
%   morphological structure of the input string.  Crucially, it
%   syllabifies a Word-final consonant as the onset of a syllable headed
%   by an epenthetic vowel, but it syllabifies a Stem-final vowel into
%   the onset of a following syllable (consisting of underlying segments)
%   if doing so would not create an illicit onset (more than two
%   consonants), otherwise the consonant is deleted (unparsed).  This
%   captures the following generalizations:
%   
%     (1) Dakota does not permit codas;
%     (2) word-final consonants trigger epenthesis to satisfy (1);
%     (3) Stem-final consonants are syllabified into a following
%         onset or deleted to satisfy (1).

Syllabify:
    <> ==
    <[> == <>
    <]> == <>
    <!> == ! <>
    <$vowl> == ( $vowl ) <>
    <$con0 ]> == <$con0 A>
    <$con0 $vowl> == ( $con0 $vowl ) <>
    <$con0 $con1 $vowl> == ( $con0 $con1 $vowl ) <>.

%   Background facts:
%   
%     Cex     'kettle'
%     zi      'yellow'
%     Cexzi   'brass kettle'  (compound formed by Stem concatenation)
%     Cexazi  'yellow kettle' (compound formed by Word concatenation)
%   
%   Some sample theorems: 
%   
%     Syllabify:<[ C e x ]> = ( C e ) ( x A )
%     Syllabify:<[ C e x z i ]> = ( C e ) ( x z i )
%     Syllabify:<[ [ C e x ] [ z i ] ]> = ( C e ) ( x A ) ( z i )
%
% % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % %

%   The following is a list of example inflected word tokens:

Word1:
    <> == Ceka
    <syn pol> == neg
    <syn num> == sing
    <syn per> == two.

Word2:
    <> == Khata
    <syn num> ==sing
    <syn per> == one.

Word3:
    <> == Hi
    <syn num> == sing
    <syn per> == one.

Word4:
    <> == Puza
    <syn num> == sing
    <syn per> == two.

Word5:
    <> == Mani
    <syn num> == sing
    <syn per> == three.

Word6:
    <> == Nuni
    <syn num> == sing
    <syn per> == three.

Word7:
    <> == Thani
    <syn num> == plur
    <syn per> == one.

Word8:
    <> == Ceka
    <syn tense> == fut
    <syn num> == dual
    <syn per> == excl.

Word9:
    <> == Khata
    <syn num> == plur
    <syn per> == two.

Word10:
    <> == Hi
    <syn num> == plur
    <syn per> == one.

Word11:
    <> == Puza
    <syn num> == plur
    <syn per> == three.

Word12:
    <> == Mani
    <syn num> == plur
    <syn per> == two.

Word13:
    <> == Nuni
    <syn num> == dual
    <syn per> == incl.

Word14:
    <> == Thani
    <syn num> == dual
    <syn per> == excl.

Word15:
    <> == Ceka
    <syn pol> == neg
    <syn tense> == fut
    <syn num> == dual
    <syn per> == excl.

Word16:
    <> == Khata
    <syn pol> == neg
    <syn num> == sing
    <syn per> == two.

Word17:
    <> == Hi
    <syn num> == plur
    <syn per> == three.

Word18:
    <> == Puza
    <syn tense> == fut
    <syn num> == dual
    <syn per> == excl.

Word19:
    <> == Mani
    <syn num> == dual
    <syn per> == incl.

Word20:
    <> == Nuni
    <syn pol> == neg
    <syn tense> == fut
    <syn num> == dual
    <syn per> == incl.

Word21:
    <> == Thani
    <syn pol> == neg
    <syn tense> == fut
    <syn num> == sing
    <syn per> == one.

Word22:
    <> == Ceka
    <syn num> == dual
    <syn per> == excl.

% % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % %
%
%   Some example theorems:
%
% Word1:
%   <root> = C e k
%   <syn form> = { neg sing two }
%   <mor form> = [ [ y a C e k ] S n i ]
%   <phn form> = ( y a ) ( C e ) ! ( k A ) ( S n i ).
%
% Word2:
%   <root> = kh a t
%   <syn form> = { sing one }
%   <mor form> = [ m a kh a t ]
%   <phn form> = ( m a ) ( kh a ) ! ( t A ).
%
% Word3:
%   <root> = h i
%   <syn form> = { sing one }
%   <mor form> = [ w a h i ]
%   <phn form> = ( w a ) ( h i ) !.
%
% Word4:
%   <root> = p u z
%   <syn form> = { sing two }
%   <mor form> = [ n i p u z ]
%   <phn form> = ( n i ) ( p u ) ! ( z A ).
%
% Word5:
%   <root> = m a n i
%   <syn form> = { sing three }
%   <mor form> = [ m a n i ]
%   <phn form> = ( m a ) ( n i ) !.
%
% Word6:
%   <root> = n u n i
%   <syn form> = { sing three }
%   <mor form> = [ n u n i ]
%   <phn form> = ( n u ) ( n i ) !.
%
% Word7:
%   <root> = th a n i
%   <syn form> = { plur one }
%   <mor form> = [ [ un k th a n i ] p i ]
%   <phn form> = ( un ) ( k th a ) ! ( n i ) ( p i ).
%
% Word8:
%   <root> = C e k
%   <syn form> = { fut dual excl }
%   <mor form> = [ [ [ un k C e k ] p i ] k t e ]
%   <phn form> = ( un ) ( k C e ) ! ( k A ) ( p i ) ( k t e ) .
%
% Word9:
%   <root> = kh a t
%   <syn form> = { plur two }
%   <mor form> = [ [ n i kh a t ] p i ]
%   <phn form> = ( n i ) ( kh a ) ! ( t A ) ( p i ).
%
% Word10:
%   <root> = h i
%   <syn form> = { plur one }
%   <mor form> = [ [ un k h i ] p i ]
%   <phn form> = ( un ) ( k h i ) ! ( p i ).
%
% Word11:
%   <root> = p u z
%   <syn form> = { plur three }
%   <mor form> = [ [ p u z ] p i ]
%   <phn form> = ( p u ) ! ( z A ) ( p i ).
%
% Word12:
%   <root> = m a n i
%   <syn form> = { plur two }
%   <mor form> = [ [ y a m a n i ] p i ]
%   <phn form> = ( y a ) ( m a ) ! ( n i ) ( p i ).
%
% Word13:
%   <root> = n u n i
%   <syn form> = { dual incl }
%   <mor form> = [ un k n u n i ]
%   <phn form> = ( un ) ( k n u ) ! ( n i ).
%
% Word14:
%   <root> = th a n i
%   <syn form> = { dual excl }
%   <mor form> = [ [ un k th a n i ] p i ]
%   <phn form> = ( un ) ( k th a ) ! ( n i ) ( p i ).
%
% Word15:
%   <root> = C e k
%   <syn form> = { neg fut dual excl }
%   <mor form> = [ [ [ [ un k C e k ] p i ] k t e ] S n i ] 
%   <phn form> = ( un ) ( k C e ) ! ( k A ) ( p i ) ( k t e ) ( S n i ) .
%
% Word16:
%   <root> = kh a t
%   <syn form> = { neg sing two }
%   <mor form> = [ [ n i kh a t ] S n i ]
%   <phn form> = ( n i ) ( kh a ) ! ( t A ) ( S n i ).
%
% Word17:
%   <root> = h i
%   <syn form> = { plur three }
%   <mor form> = [ [ h i ] p i ]
%   <phn form> = ( h i ) ! ( p i ).
%
% Word18:
%   <root> = p u z
%   <syn form> = { fut dual excl }
%   <mor form> = [ [ [ un k p u z ] p i ] k t e ]
%   <phn form> = ( un ) ( k p u ) ! ( z A ) ( p i ) ( k t e ) .
%
% Word19:
%   <root> = m a n i
%   <syn form> = { dual incl }
%   <mor form> = [ un k m a n i ]
%   <phn form> = ( un ) ( k m a ) ! ( n i ).
%
% Word20:
%   <root> = n u n i
%   <syn form> = { neg fut dual incl }
%   <mor form> = [ [ [ un k n u n i ] k t e ] S n i ]
%   <phn form> = ( un ) ( k n u ) ! ( n i ) ( k t e ) ( S n i ) .
%
% Word21:
%   <root> = th a n i
%   <syn form> = { neg fut sing one }
%   <mor form> = [ [ [ m a th a n i ] k t e ] S n i ]
%   <phn form> = ( m a ) ( th a ) ! ( n i ) ( k t e ) ( S n i ) .
%
% Word22:
%   <root> = C e k
%   <syn form> = { dual excl }
%   <mor form> = [ [ un k C e k ] p i ]
%   <phn form> = ( un ) ( k C e ) ! ( k A ) ( p i ).
%
% % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % %

# show <root> <syn form> <mor form> <phn form>.

# hide Verb Vactive_intrans Vstative HeadProj1 HeadProj2 HeadProj3
       Syllabify Ceka Khata Hi Puza Mani Nuni Thani Kte.

% % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % % %

% The next line is the Revision Control System Archive Id: do not delete it.
% $Id: archive.dtr,v 1.1 1997/04/09 20:40:33 root Exp $