.M:1 .L:59 .X:4 .R:Q READ.ME .H:READ.ME .H: .H: .H: THE PYE ANALYSIS OF LANGUAGE by Clifton Pye 200 Arrowhead Lawrence, KS 66044 (913) 749-2680 This distribution disk contains the three PAL programs (formix.exe, phonix.exe and syntax.exe), a description of the programs and their use (pal.doc), a sample transcript (daniel.dat), and its associated analyses (daniel.chk, etc.). I would suggest making a working copy of the PAL routines by copying the PAL programs and the sample transcript to another disk. Then try experimenting with the routines by analyzing the sample transcript. I have added several new features to the routines that are not documented in pal.doc. Formix.exe now contains a routine for converting Wordstar files typed in the document mode to ASCII text. This routine is preliminary and may not convert all versions of Wordstar files. The checking routine has also been changed to accommodate texts with any number of tiers. There is a new routine which allows the user to change the order of tiers in the child's lines. This allows the syntactic analysis routine in syntax.exe to produce a concordance of whatever information occurs in the two final tiers of the child's lines. For example, suppose the child's lines contained the following tiers: 1. Phonetic 2. Orthographic 3. Syntactic 4. Semantic 5. Discourse You could produce a list of every discourse element for every word in the orthographic tier by switching tiers 2 and 4. You could produce a list of every semantic role for every word in the text by switching the tier order to 1 3 5 2 4. Experiment to see what effects you can produce by using this option in the program. I have also added a merge routine. This option allows you to merge outputs of the concordance routines produced by the syntax.exe program. This is explained below. The phonological analysis routines in phonix.exe now support an additional number of characters (consonants G, K, L, q and vowels &, @, ^, iy, ey, ay, oy, Iy, Ey, Vy, 6y, uy, aw, ow, uw, i~, e~, a~, u~). These can be used to supplement the characters in the ASCII alphabet or defined to suit the user's own needs. Phonix also now contains a routine which looks up a target pronounciation for the child's words in the file phonetic.dic. This file has to be on the same drive as the phonix.exe program in order to be accessed. It is a standard ASCII file, so that as long as you are using a word processor which produces ASCII text, you may edit this file to add new words to it or change the phonetic transcription of the existing words to reflect local or individual dialects. Phonix includes a new routine for converting the PAL phonetic characters to the IPA character assignments in the SIL font sets. These font sets are available through the SIL as shareware fonts for use in Word for Windows. The SIL IPA character assignments are similar to those in PAL except for the following: 2 PAL is converted to SIL F => voiced velar fricative G => palatal fricative K => barred /l/ @ => low, back vowel ^ => high, back, unrounded vowel ! => high, front, rounded vowel I recommend using PAL to carry out the phonological analyses and then using the conversion routine to produce a final report with the SIL IPA font sets. The lexical and syntactic analysis routines will only handle children's files up to 20K bytes in length. If you wish to analyze a longer file, you will need to divide the file into two or more smaller files and analyze each independently. The phonological analysis routine will handle up to 175 words which share the same segment in the same position. I have used it successfully to analyze the words in the file phonetic.dic so this limitation should not cause any practical difficulties. The syntax.exe program contains routines for calculating standard descriptive statistics as well as routines for producing lexical and syntactic concordances. The concordance routines provide a fast, efficient analysis of the syntactic productivity of a subject's words and morphemes. The concordance routines are limited to processing files up to 30K bytes in length or containing approximately 750 child utterances (on computers with 640K bytes of RAM memory). This limit can be overcome by dividing long files into shorter files and using the concordance routine separately on each of these files. The concordance routine is also affected by items in the config.sys file in your computer's root directory. If you have problems getting the program to work, try altering your config.sys file so that it has the line: files=20 buffers=9 The lexical concordance routine prints the child's utterances to a scratch file labelled 'scratch'. It will also tell you how far it got in its analysis of this file before it ran out of memory. Use a word processor to edit the scratch file and delete all of the utterances up to, but not including, the utterance where the computer stopped in its previous analysis. Then rename the scratch file something else (so the concordance routine does not write over it) with the DOS renaming routine, e.g. ren a:scratch a:child.new. Then you can simply repeat the concordance analysis on this new file. Once completed, use the merge routine in the formix.exe program to combine the separate concordances into a complete concordance for the entire sample. Above all - Experiment! If your computer stops before the analysis is done, or prints an error message it is most likely due to a sample file that is too long. Try breaking the file into separate files and repeating the analysis. The most common formatting mistake is to put a period after the name line at the beginning of the data file. The period causes the checking routine to cycle indefinitely. DO NOT USE A PERIOD AFTER THE NAME LINE! I would appreciate being informed about any glitches which you discover in the routines. I will send along free updates of the PAL programs to anyone who suggests a practical change or addition to the programs.