\_sh v3.0  563  Readme Notes

\id ADAPT3B
\s Derivation sample from Yawelmani
\p This is the same Yawelmani data and rules as in the previous sample, but modified so that
derivations show the steps in the derivation, as they are shown in phonological papers. This is
done by spreading the rules out into multiple files, and using a separate Generate process for
each rule file.
\p The file YAWDERIV.TXT shows the derivations. The names of the rules being applied are
shown on the left. You can see the effect of each rule by looking at the changes in the derived
form. If the rule did not apply, then the form is the same as in the line above.
\s2 Interlinear Setup
\p Look at the interlinear setup to see the list of processes. You will see a list of seven Generate
processes. Each generate process applies a rule file that contains only one or two rules, and it
generates an intermediate line that shows the results of the rule or rules.
\s3 Tip
\p The rule names shown in the derivation file come from giving each intermediate marker the
name of the rule that was applied to produce the marker. (This is done through the Marker
Properties dialog box -- right-click on the name in the marker pane to see this.) To show
these names on the screen, choose View-Field Names.
\s2 Comments
\p One interesting thing about this step-by-step derivation display is that you can see when the
morpheme boundaries are removed. From that point on the hyphens are gone. The first
Generate process is set to keep the morpheme boundaries, so that they are visible to the rules
in the second process, which are sensitive to boundaries. The second process is allowed to
remove the boundaries because the rest of the rules are not sensitive to them.
\p This derivation layout cannot be made to directly produce derivations the way they are
normally published in books and papers because published derivations usually leave out the
form where rules have not applied: 
\s3      ?ili:-al 
\s3      ?ili:+al 
\s3      ?ili:al 
\s3      ---
\s3      ----
\s3      ?ile:al 
\s3      ?ile:l 
\s3      ?ilel
\p But it would be easy to delete from one of these derivations all of the lines that did not
change, leaving just a derivation showing the rules that applied. 
\nt Note:
\p If a derivation is cut from Shoebox and pasted into a document in a word processor, the
paste will show the markers rather than the field names, so the rule names will not appear in
the pasted material. You will have to change the markers to rule names by hand, or by using
search and replace.
\s3 Inconvenient
\p This detailed approach, while very useful for displaying what is happening in the derivation, is
not very useful for experimentation or for initial rule development. This is because it is not
easy to move rules around between the numerous rule files, and it takes a fair bit of work to
add a new rule (e.g., to keep the pattern of one rule per process, inserting a new rule in the
middle of the existing rule set means either adding a new Generate process in the middle of
the Interlinear setup--which messes up your numbering scheme, or moving all the rules down
one process and adding another Generate process at the end for the last rule). 
\s3 Tip
\p Instead, divide you list of rules into three or four sets that do not interact much with each
other and put each rule set in a separate rule file. After that, create a Generate process for
each rule file. Now as you test your rules, you can add new rules, rearrange the order of your
rules and delete rules easily within a given set without having to add new Generate processes
or create new rule files. This is a good way of experimenting with rules and seeing the results
of changes in the derivations. Of course, if a file contains a number of rules, then the field
name for the marker it produces can only be a general reference to the rule set, not the name
of the specific rule that has applied.
\p
\s To go to the next Adaptation tutorial, open the project in the ADAPT3C folder.