How were vowels u and i discerned from consonants v and j?

Question

If in original texts there were no means of distinguishing whether u and i were consonants or vowels, how then do we now know which ones were which?

The easy ruleset I learned in high school is that it is pronounced as a consonant if:

1. It succeeds the letters q, g, or s and precedes another vowel
2. It lies between two other vowels
3. It lies at the beginning of a word before another vowel, except where this breaks rule 2

However, in my further research, which largely involved manually checking all combinations in which u or v was written in a word list extracted from Whitaker's Words, I found that it is recognized as a vowel:

• When u succeeds a consonant and precedes a vowel, except for the consonants q, g, s, and usually l or r:
• when l or r succeeds another consonant,
• when l or r lies at the beginning of the word,
• when l or r ends the 3rd principal part of a verb,
• and in the words miluus and beluus.

Otherwise, it is probably a consonant.

There are likely a few holes in this description, but I think this should account for at least around 90% of cases. Back to the question, these patterns of consonants and vowels were deduced solely from the already completed work of others, so how did they know which were which?

Answer 1

The book Vox Latina by W. Sidney Allen has a section on semivowels which compares the i and u consonants and vowels. Among the sources of information mentioned about the way these were pronounced are

• What the Romans wrote about pronunciation
• The way that Latin words were transcribed in Greek
• Poetry scansion - Scansion in particular should be helpful for deducing whether i and u are consonants or vowels, because they would produce different patterns of long and short syllables. For example, poetry scansion shows that in cui and huic the u is a vowel. According to Sidney Allen

Finally, it should be noted that in cui, huic, and the interjection hui, the second letter is not the consonant but the vowel u, which forms a diphthong with the following i.1 It is true that Quintilian finds cui and qui somewhat similar (t i, 7, 27), but his reference to the' pinguem sonum' of the former suggests a back as opposed to a front vowel (cf. p. 34) as the more prominent element-and there is other evidence besides. The clearest proof is provided by the fact that elision is permitted before huic (but not, for example, before uis), and that in -alicui the cu does not' make position' for the preceding syllable, which remains light; both of these pieces of evidence indicate that the u must here be a vowel; similarly huic does not' make position' with a preceding final consonant.

As he points out, there is occasional poetic interchange between the u and i vowels and consonants

The close connexion between the vowel and the consonant in Latin is seen in occasional poetic interchange of function, as, for example, trisyllabic silva and disyllabic genua (with consonantal u 'making position')

but as long as this was only occasional, poetry scansion could be used to distinguish them.