\_sh v3.0  563  Readme Notes

\id ADAPT3A
\s Generate sample from Yawelmani
\p This sample is based on one of the fullest examples in the phonology textbook "Generative
Phonology" by Michael Kenstowicz and Charles Kisseberth. The Yawelmani dialect of
Yokuts phonological sketch is pages 77-99 in the 1979 version of the book.
\s2 Introduction
\p This sample shows how the Shoebox Generate process can implement the complex set of
ordered rules required by the Yawelmani language. Such a Generate process could be the
final process of adaptation into Yawelmani. It could also be used to test a set of phonological
rules for a language, experiment with rule ordering, etc.
\s2 Example Text File
\p The file YAWDERIV.TXT shows all the Yawelmani underlying to surface form derivations
given in the book, with the reference numbers of the examples. The names of the rules that
applied are also given. By looking at the rules you can see the sequence in which they were
applied and how they interact.
\s2 Phonological Rules File
\p The file YAWPHON.RUL is the Phonological rule file that performs the derivations. The top
of the file contains the definitions of symbols used in the rules. Then come the rules. The rules
appear in the order in which they are applied. Each rule is preceded by a comment giving the
name of the rule and the reference number of the rule from the book. You do not need the
book to understand the example, but if you have access to the book, it will explain the
reasons why the sequence of rules are chosen the way they are.
\s3 Definitions of Symbols
\p Starting at the top of the rule file, the first symbol defined is "C" for Consonant. 
\p The next symbol defined is "Cl" for Consonant plus length mark. This is a convenient
combination for some of the Yawelmani rules. 
\p Similarly the next symbol "C#" for Consonant plus boundary mark is convenient for processes
that happen both before a consonant and word final. You can define any symbol to be any
convenient combination you need. Combinations like C# can be used to simplify rules.
\nt Note: A symbol name can be made up of nearly any characters you wish but it can NOT have a space in it,
e.g.  "Hi V" would be an invalid symbol.
\p The next symbol defined is "V" for vowel. The symbol "V" includes both long and short
vowels. After "V", "Vl" is defined as long vowels, and "Vs" is short vowels. You can use any
type of name or notation in the symbol that you want. If you want something that looks like
feature notation, you can define symbols that begin with plus and minus. For example, the
following two definitions are of "+high" and "-high". These are used as a pair in rules that
lower vowels in certain contexts. 
\nt Note: When a member of +high is changed to -high, it changes to the member at the corresponding
location. 
\p In this example, "i" in +high corresponds to "e" in -high, and "u" in +high corresponds to "o" in
-high. These relationships and feature references are taken from the terminology used in the
example in the book.
\s3 Phonological Rules
\p After the definitions comes the list of ordered rules. For example, the first rule is "Shortening
of vowels in verbal nouns (31)". The (31) is the reference number of the rule in the book. The
rule says that a long vowel becomes short between a consonant and one or more consonants
followed by word boundary.
\nt Note: The notation "C*" means zero or more instances of "C". So the combination "[C][C*]" means
one or more instances of "C".
\p The fourth rule is a special rule that removes the hyphens. It is not necessary to write a rule to
do this because they are removed automatically at the end of the derivation. But removing the
hyphens ahead of time allows the following rules to apply both across morpheme boundaries
and within morphemes. Note that the blank second line in the rule is important. It means
replace the matched thing with nothing.
\p The reason I removed hyphens in this example is that Yawelmani vowel harmony works
anywhere in a word. For example the suffix "iws" becomes "-iwis" or "-uwus" depending on
the preceding vowel. Vowel harmony says that the first vowel harmonizes with the last vowel
of the preceding morpheme and then the second vowel in the affix (which is an epenthesized
"i") harmonizes with the first vowel. So the rules applies once across a morpheme boundary
and once within the word.
\p Rules that are sensitive to morpheme boundaries should be performed before the hyphens are
removed. For example, the first rule does not apply across a morpheme boundary, and the
second and third rules require a morpheme boundary.
\p The vowel harmony rules make use of the symbol "Cl" which is defined to be the consonants
plus the length symbol. It is convenient to define such a symbol because it allows "Cl*" to
cross over the length symbol as well as over any number of consonants.
\p Vowel harmony is the only Yawelmani rule that cannot be expressed in a single rule in
Shoebox. It requires two rules because Shoebox does not have an alpha variable notation.
\p The vowel lowering rule changes a +high variable to a -high variable. The number of the
element matched in +high is used to choose the same numbered element of -high.
\p The truncation rule makes use of the fact that if a variable occurs twice in the match, but only
once in the output, then the output is the first instance in the match. The truncation rule can
match any two dissimilar vowels, and will always delete the second one.
\s2 Conclusion
\p As you can see from the example derivations given, the Generate process can handle
complicated rules and rule ordering. This can be very useful, but set up this way, the whole
process runs "behind the scenes", and if some form is not deriving properly, it's not easy to
see what rule is misapplying, or where two rules need to be in reverse order, etc. The next 
adaptation sample has this same set of rules but splits them up so that several Generate
processes are used to process the rules. The result is that each step of the derivation is clearly
shown.
\p
\s To go to the next Adaptation tutorial, open the project in the ADAPT3B folder.