What are the historical justifications for first-syllable stress in the word “orthoepy”?

Looking at words beginning with ortho- there seem to be two possible pronunciations:

/ɔːrˈθɒ-/ as in orthogonal or orthography,
/ˈɔːrθə-/ as in orthodox or orthodontist,

where the second pronunciation has primary or secondary stress on the first syllable.

But in English phonology, the vowel /ɒ/ must be followed by a consonant, so the first pronunciation is ruled out for orthoepy. This is presumably what Walker meant when he said “Orthoëpy, having no consonant in the antepenultimate syllable, naturally throws its accent on the first” (1822).

You could also pronounce orthoepy /ɔːrˈθoʊ.iː.pi/. This seems to be what Robert Nares meant when he said that "the accented vowel is long in the antepenultima" (1782). However, there aren't any other words where ortho- is pronounced this way. I suspect this discouraged people from using it.

SPE offers a couple of analyses that might be relevant. (1) They treat Greek-like prefixes such as "photo-", "pseudo-", "amino-" as being something like separate words (if I recall correctly, which I'm not sure I do), and (2) they suggest the final "-y" in words like "presidency" is (as the spelling suggests) a phonological glide, not a vowel, which is why it doesn't cause a heavy penult to be stressed.

If the first morpheme of "orthoëpy" is being treated as though "ortho" is a separate word, the second "o" can be unstressed and diphthongized, like the final "o" of "motto" (and unlike the "o" of "Plato"). But a word ending in a consonant (after the antepenult of "Orthoëpy", i.e.) would not work this way.

I thought I’d update this as I decided to look at Elphinston and Nares, the sources Walker cites.

I was able to confirm that they both put the stress on the first syllable of “orthoepy”. Neither of them seems to give the same “no consonant in coda of antepenult” explanation that Walker does, so my best guess at the moment is that Walker came up with this explanation himself.

Robert Nares, in Elements of Orthoepy (1784), lists “orthoepy” as a word with “the accented vowel long in the antepenultima”. I don’t know what he means by saying the stressed vowel is long. Overall, Nares doesn’t seem incredibly concerned with explaining things; a lot of his book is just lists of words that belong to certain categories. On page 185, he also mentions potential pre-antepenult stress in the pronunciation of some other words such as academy, receptacle etc. (although “academy” at least has the justification that the “e” was long in Latin).

James Elphinston, in Propriety Ascertained in Her Picture, Vollume [sic] I (1786), spells the word “orthoeppy” and makes it clear that he stressed it on the first (pre-antepenult) syllable, with secondary stress on the third (penult). The passage where this is mentioned seemed to me informative enough to reproduce here:

Stres, like all elſe in our language, reduced now to’ ſcience; no word can hav doutfool emphatticizement. Dhe ſame ranks indeed dhat penultimately rely on a caracter of ov industry, ſo contrary to’ dhe ellegant antepenultimate enfoarcement, may be herd on the ſame penultimate key to’ tune dheir edducacion from an academy. Dhe Inglish clas juſt above muſt be bred at an accademy, az hoo ſhal arrain its accrimony? Yet elegant Nature in acaddemy with anattomy, az in anallogy with ſupremmacy ; in monnarky and theoccracy, geoggraphy and orthoggrapy, owns antepenultimate power. Such iz dhe unequal conteſt between dhe firſt and ſecond ov pollygamy and polyggamy. Manny tetraſyllabels, howevver, lay dheir primmary ſtres on dhe firſt, and conſequently, dheir seccondary on dhe third ſyllabel : az, pallinody, olligarky, hierarky, orthoeppy, ignominny, contumely, contumacy, cerremony, pattrimony, aggriculture, arbitrary, preffatory, and, in ſpite ov neceſſity, acces, direct, refreſh, conſiſt, all ſtrong on dhe ſeccond ; neſceſſary, acceſſary and acceſſory, dirrectory, reffectory, conſiſtory ; all ſtrong on dhe firſt ſyllabel. Stil dheze doo but ſpeak dactyllian regard to’ antepenultimate power, treating dhe laſt ſyllabel az a ſupernumerary.

So like Nares, he mentions “academy” used to have a variant pronunciation with initial stress, and he gives the additional similar example of “polygamy” (although he indicates that the pronunciation with antepenult stress was more common) and “ignominy” (which is quite interesting, since I can’t see any explanation for the initial stress in this word either).

Overall, there doesn’t seem to be a satisfactory explanation.