this text's formatting needs clearing up. by Noel Mann
The following was written by Noel Mann in an unpublished manuscript written February 2, 2005. It is reproduced here unedited.
Phonostatistics applies to analytical techniques to quantify the differences between speech varieties. Syllostatistics applies to analytical techniques to quantify the syllable differences between speech varieties. Lexicostatistics covers analytical techniques to quantify the lexical differences between speech varieties17. The method described in Blair (1990:31-33) is a hybrid of phonostatistics and lexicostatistics. This can be called phonolexicostatistics. The method described below is a hybrid of syllostatistics and lexicostatistics and as such may be called syllolexicostatistics. Syllostatistics and syllolexicostatistics generally fit Asian languages better than more traditional phonostatistic methods since the syllable is the primary domain of historical sound changes. (more preamble and examples to follow) Mechanics (Basic steps in the analysis) Step 1: Determine basic syllable structure of the languages Does the language have initial clusters? How do initials map across languages? Does the language have final consonants? How do final consonants and nuclear elements map across languages? Does the language have medial consonants? How do the medials interact with initial and rhyme elements? (If rhymes or final consonants do not seem to interact with other syllable elements, it may be necessary to create additional categories) How do the medials map across languages? Step 2: Group words to be compared into sets with the same gloss Step 3: Determine root syllables Step 4: Determine apparent cognicity of syllables (using criteria) Step 5: Determine lexicostatistic correlates (using lexicostatistic correlates) Criteria (Basic categories and element matches) Category 1 Probable match of cognate syllables Onset and rhyme elements match (map in predictable or recurrent manner) Category 2 Possible match of cognate syllables Either onset or rhyme elements match (either maps in predictable or recurrent manner) Category 3 Unlikely match of cognate syllables Onset and rhyme do not match (do not match in predictable or recurrent manner) Ignore Non-root syllables (reduplication, derivational or supplemental meanings, and grammatical particles) Tone or other suprasegmental elements which are considered necessary to establish cognicity.
Lexicostatistic correlates (Based on syllable category matches) Syllables Category 1 Category 2 Category 3 1 1 0 0 2 1 1 0 3 1 1 1 4 2 1 1 5 2 2 1 6 2 2 2 7 3 2 2 8 3 3 2 9 3 3 3 10 4 3 3 One issue that may need to be resolved is how to deal with split cognates. The above methodology will count these as different. For example: Mountain Var 1 tSHum Var 2 tSHum tla? Var 3 tla? Figure 1.
In a case like this, there are different possibilities. Using the above criteria, comparing variety 1 to 2, we have a category 1 for the syllable comparison between tSHum and tSHum, while there is no correlation between the second syllable tla? in Variety 2 leaving this a category 3. Using a phonostatistic method for this would achieve the same results. Thus, in this case, all three Varieties are not cognate. If however, the investigator suspects that Variety 2 has preserved polysyllabic form of the parent language, while Varieties 1 and 3 have lost different syllables from the parent language. In this case two sub-analytical methods could ensue. The first would consider only the syllables which are cognate and discard those that are not. This would lead to Varieties 1 and 2 as well as varieties 2 and 3 as cognate while varieties 1 and 3 are not cognate. The second would consider all of the forms as descending from a common ancestor and would count the loss of a syllable as a common process. This may arise from different stress patterns from one language to another which results in one syllable being preferentially retained rather than another – there may also be more idiosyncratic reasons one syllable is retained over another. In this second analysis method all of the varieties are cognate including the comparison between Variety 1 and Variety 3. Note that examples like this also point out inconsistencies in our procedures as if our data appears like that shown following, we would probably retain all of the forms as part of the root, but if we were missing one variety, the analysis would be different, such as shown in the data in Figures 2 and 3. Mountain Mountain Mountain Var 1 tSHum Var 1 tSHum Var 1 tla? Var 2 tSHum tla? Var 2 tSHum tla? Var 2 tSHum tla? Var 3 tla? Var 3 tSHu Var 3 la Figure 1. Figure 2. Figure 3. 76 Step 3: Determine root syllables After applying this step, the following data is the basis of comparison. Mountain Mountain Mountain Var 1 tSHum Var 1 tSHum Var 1 tla? Var 2 tSHum tla? Var 2 tSHum Var 2 tla? Var 3 tla? Var 3 tSHu Var 3 la Figure 1a. Figure 2a. Figure 3a. This means that the presence or absence of a single word form in one of the varieties may require a reanalysis or reinterpretation of the root syllable or syllables.