Here are a few "rules of thumb" I use. I can't guarantee these will work in all cases.
- If you're an English speaker, look at a related English word from Latin. If it's spelled with "j," it was probably consonantal "j" in Latin as well. For example, we can compare Latin MAIVS to English "major," Latin OBIECTVM to English "object," and so on.
- Generally, derivatives of IACIO ("throw") are pronounced with /j/. There's a lot of them, so this is somewhat useful to know.
Another note: unlike most long consonants, phonetically long /jj/ was normally not written double in Latin. In addition, it seems that written I may have corresponded to /ji/ or /jji/ in some cases in words like TRAICIO "throw over" (composed of TRANS + IACIO). This means that you actually have to choose between five options for written i: /i/ /j/, /jj/ /ji/ and /jji/. (Or at least six if the source doesn't use macrons!)
Here is some of what W. Sidney Allen wrote about consonantal I in Vox Latina:
In the interior of a word, this sound rarely occurred singly between
vowels. Where once it had been present, it was lost prehistorically
(thus Latin trēs beside Sanskrit trayas). With a few exceptions
noted below, whenever a single, intervocalic i-consonant is written,
it stands for a double consonant, i.e. [yy]. Thus aio, maior,
peior, Troia stand for aiio, maiior, etc. This is quite clear from various types of evidence. It is specifically mentioned by Quintilian and other grammarians, who also tell us that Cicero and Caesar used in fact to spell such words with ii (†Qu. i, 4, 11; †Priscian, K. ii, 14) and it is supported by frequent inscriptional spellings (e.g.
PompeIius, cuiIus, eiius, maiIorem). In Italian a double consonant has been maintained in, for example, peggio from peius. Moreover, the consonant must be double in order to account for the fact that the preceding syllable is always metrically heavy; for the actual vowel is short—this is specifically mentioned by Ter. Maurus (†K. vi, 343), and is further evident from other considerations: e.g. maior is connected with măgis, being derived from măg-iōs.
So it appears sources that use a macron in "māius" (greater) are assigning false quantity to the vowel; according to Allen, the actual pronunciation would have been /majjus/ with a short "a" and a long "j."
It appears that "Maius" was probably pronounced the same as "maius": with two syllables, the first long because of a following double /j.j/ sound. I'm not sure why you learned that they were different. In Italian at least, the modern pronunciation is indistinguishable from that of words that we know had intervocalic consonantal /jj/ (the Italian descendant of "Maius" is maggio). The etymology of "Maius" seems to be a bit unclear, but one hypotheses seems to be that it ultimately comes from the same root as maius.
Interchange between /j/ and /i/ in poetry
In poetry, words that usually have syllabic /i/ sometimes have /j/ instead. (Similarly, /u/ and /w/ are sometimes substituted for one another in poetry.) I don't know the details about when this can occur.