[A2A] Fun exercise.
In biblical (Old Testament) times, the Canaanite alphabet was common to all Canaanite nations—from Moab and Edom in the southeast to Phoenicia in the northwest. The language was substantially the same, as well, throughout that period, with dialectal differences that widened over the centuries, so that by the time of the late First Temple period (750–580 BCE) mutual comprehension was only partial.
There are plenty of references to letter frequency in modern Hebrew (one good English-language one is Stefan Trost’s Character Frequency: Hebrew). However, although linguistically modern Hebrew is very similar to biblical Hebrew, it does contain many words of foreign origin, and tends to use ‘full spelling’ (i.e. use the letters vav and yod to indicate the vowels /o/u/ and /i/, respectively), which can skew the results somewhat.
So a better test would be a certified ancient text of sufficient length to be indicative.
The Mesha Stele—a commemoration by King Mesha of Moab of the liberation of his people from the “yoke” of Israelite rule—is a good candidate in that regard, as it dates to around 840 BCE, and is written in the Moabite language, in the Canaanite script (see right).
To analyse the letter frequency, I used dCode, one of several free online tools. Like most text analysis software, requires the text to be set in Roman characters. This is a classic application of SimHebrew (Hebrew simulated in Roman characters), so I converted the Canaanite/Hebrew text (courtesy of the Hebrew Wikipedia entry on the Mesha Stele) into SimHebrew using the Square Hebrew/SimHebrew Converter—then pasted the result into dCode, and ran the analysis.
The result: the least common consonant is f (which in SimHebrew represents péh sophit—the word-end form of the letter péh). Since word-end forms of characters did not exist in Canaanite/Old Hebrew (they were invented centuries later), this can be ignored. The next least frequent consonant is z (= zayin in SimHebrew), with the letter s (samekh) coming a close second:
So there you have it. This result, incidentally is broadly similar to the frequency analysis of modern Hebrew texts.