The statistics of the phonetic value of AB65 /ju/

As Palmer notes in A Companion to Linear B: Mycenaean Greek Texts and their World vol. 1, "The phonetic value of *65 is probably ju, but this reading has not yet been officially adopted; hence the transliteration ju? is adopted here."

We should instead adopt j? for now, based on current evidence. A statistical analysis shows that the /j*/ is solid, but the vowel /*u/ is not.

Here are some statistics to demonstrate *65's phonetic value. From what we can see in the numbers, it is the vowel rather than the consonant that remains uncertain. *65's membership in the /j*/ row is solid based on the numbers.

Based on 3,381 Linear B sign groups from my Linear B data project, we discover that the j* row is distinct from the other consonant rows by its overwhelming tendency to be preceded by an *i consonant-vowel (CV) syllable, cf. ko-no-si-ja, i-je-re-u, mi-jo-qa, etc. This also undoubtedly helped Michael Ventris, sometime after the 3rd state of his trial grids, to identify the j* row as the semi-vowel it is.

/ja/ *57 is preceded by an *i CV 63% of the time. Compare with only 9% *a.
/je/ *46 is preceded by an *i CV 75% of the time. Compare with only 13% *a.
/jo/ *36 is preceded by an *i CV 61% of the time. Compare with only 14% *a.
/j?/ *65 is preceded by an *i CV 55% of the time. Compare with only 18% *a.

ja je jo j?
*a- *a-ja
*e- *e-ja
*i- *i-ja
*o- *o-ja
*u- *u-ja

*a is greater than *i for almost all consonant rows except the j* row.1

Here, the numbers show a solid confirmation for including *65 in the j* row.

What about the vowel /*u/, though?

Looking at the other consonant rows, we find non-zero values for the incidence of a *u CV preceding the row's *u value - demonstrating the regressive and progressive alphasyllabary spelling rules of consonant clusters (see Woodard 1997). All numbers are low because the *u CV symbols are used less frequently in Linear B than other vowels.

*u-du 4%
*u-ju 0% of 11 sign groups that include /ju/
*u-ku 2%
*u-mu 0% of 22 sign groups that include /mu/; see non-zero value below
*u-nu 12%
*u-pu 2%
*u-pu2 8%
*u-ru 30%
*u-su 12%
*u-tu 5%

We also find non-zero values for the incidence of a *u CV following the row's *u value - including the nasal /m/ that fails above, which can begin consonant clusters like -μφ- and -μν-.

du-*u 22%
ju-*u 0% of 11 sign groups that include /ju/
ku-*u 17%
mu-*u 14%
nu-*u 2%
pu-*u 0% of 63 sign groups that include /pu/; see non-zero value above
pu2-*u 9%
ru-*u 2%
su-*u 6%
tu-*u 7%

The vowel for *65 cannot be confirmed by the expected regressive and progressive spelling rules we would hope to confirm the vowel with, whereas all other *u CV values show a non-zero value of *u CVs preceding and / or following the relevant CV value.

What vowels are regressive and progressive candidates for *65?

*a-ju 18%
*e-ju 18%
*i-ju 55% - the high number here coincides with the j* semi-vowel confirmation also seen in /ja/, /je/ and /jo/
*o-ju 9%

ju-*a 36%
ju-*e 27%
ju-*i 0% - confirming that the high number of *i above confirms the J* semi-vowel rather than any *i regressive or progressive spelling of a consonant cluster
ju-*o 27%

The numbers, alas, are entirely ambiguous. The vowel of AB65 must remain a mystery for us until further sign groups using the symbol are found to improve our statistical data.

AB65's consonant row is confirmed by the frequency of its preceding phoneme boundary with *i CV symbols. AB65's vowel is ambiguous. We should adopt j? as its phonetic transliteration until further evidence surfaces.

1 In no other consonant row is the preceding *i % significantly greater than *a as it is in the j* row.
d* row: *i-da 19% is slightly greater than *a-da at 17%.
n* row: *i-ni 32% is greater than *a-ni 24%, probably due to regressive and / or progressive spelling rules (see Woodard 1997). *i-nwa 45% is greater than *a-nwa 18%, but /nwa/ only occurs 11 times in the sign group sample data, and is skewed by its small sample size and a variety of declensions on the sign groups pe-ru-si-nwa and ti-nwa- where the i-nwa boundary doesn't vary.
q* row: *i-qa 8% is slightly greater than *a-qa at 5%. *i-qo 21% is greater than *a-qo 7% and is the largest discrepancy, which is still significantly smaller than the j* discrepancies.
s* row: *a-su and *i-su are equal at 14% each.
z* row: *i-za 21% is greater than *a-za 12%. *i-zo 21% is slightly greater than *a-zo 15%.

Palmer, Ruth, 2008 A Companion to Linear B: Mycenaean Greek Texts and their World (ed. Yves Duhoux and Anna Morpurgo Davies) "How to Begin?"

Woodard, Roger D., 1997 Greek Writing from Knossos to Homer: A Linguistic Interpretation of the Greek Alphabet and the Continuity of Ancient Greek Literacy

This entry was posted in Hypotheses & Theories, Inscriptions and tagged , , , , , , , , . Bookmark the permalink.

2 Responses to The statistics of the phonetic value of AB65 /ju/

  1. Achilleas Vortselas says:

    Very interesting, hadn’t read that.

    Have you formulated a test, to do Statistical Hypothesis Testing? Also a table with the percentages would help readers better visualise this.

    • I’ll read back over significance testing in Mandel’s The Statistical Analysis of Experimental Data and see if I can structure the data better. Thanks!

Leave a Reply

Your email address will not be published. Required fields are marked *