Documenting the 1984 twol rules

The file has the following sections

SURFACE ALPHABET
containing the surface Mari alphabet
SURFACE SETS
containing the linguistically motivated surface subsets of the alphabet
LEXICAL ALPHABET
containing the underlying Mari alphabet
LEXICAL SETS
containing the lexically motivated underlying subsets of the alphabet
DEFINED SUBEXPRESSIONS
defining the realisations of some underlying symbols
TWO-LEVEL RULES
the rule section
ALPHABET
the alphabet used in the automata
AUTOMATA
contains the hand-encoded finite-state automata of the rules of the rule section
ALIGNMENT
contains, hmm, more of the same

Comments to the initial sections

The surface alphabet contains the small letters of the Mari alphabet a-A, and in addition the symbols '-' and '0' (unclear what they mean).

The surface sets are: V vowels, Vo back, VO front and Ve neutral vowels. Cons consonants, Cst stops and fricatives. = is a symbol meaning 'any letter'.

The underlying lexical alphabet contains, in addition to the Mari alphabet 17 symbols. Of these, e1, a1, i1 u1, y1 are special vowel symbols (used for XXX), and the rest , i.e. ! + / { _ $ } @ ~ & - # are symbols used to identify certain morphophonological contexts.

The lexical sets parallel the surface ones. V, Cons and = are built like for the surface sets, but the rest are different: Here V1 contains the modified vowels and ? the identifier symbols.

In the defined subexpressions section, we see that all the ? symbols must be realised as 0. Then the sets C/V* and C/Y* are defined as any phoneme and any consonant and glide, respectively.

Comments to the rule section

Rules s-1 through s-5 delete suffix identifier symbols.

Rule s-6 reduces vowels to schwa (V1 to Y) in a given context.

Rule s-7a through s-7c is the vowel harmony rule for e1. It realises e1 as o, O or e, following a stem containing front, back and neutral vowels, respectively.

Rule s-8abcd is the rule for the other special vowels. They are realised as their general surface vowels to the left of #:0 and -:-.

Rule s-9 deletes the special vowels in front of vowel-initial suffixes.

Rule s-10 inserts a schwaa between consonant-final stems and consonant-initial suffixes.

Rule s-11 deletes a and e in front of a- and e-initial suffixes.

Rule s-12 devoices voced fricatives after certain stem codas.

Rule s-13 deletes the symbol 8 in front of e1:e initial suffixes.


Trond Trosterud
Last modified: Sun Jan 11 13:25:09 2004