Allen 1978 (p. 17) judges the evidence to favour /kʷ ɡʷ/, while Cser 2020 (§2.2.2) comes to the opposite conclusion. The relevant facts, per the latter, are as follows:
⟨qu⟩ enjoyed a wide lexical distribution, while ⟨gu(V)⟩ was limited to a dozen or so words, where it was always preceded by /n/. The grammarian Velius Longus indicated that the ⟨u⟩ of ⟨qu⟩ was in some way different from /w/ in general. No geminate *⟨ququ⟩ is attested, whereas all (other) Latin stops are also found as geminates. Sequences of obstruent + glide are rare in Classical Latin. In poetry, whenever sequences of stop + glide occur in medial position, the scansion reveals that can be split across syllables, but this is never the case for ⟨qu⟩. Neither ⟨qu⟩ nor ⟨gu⟩ are ever followed by a consonant, unlike any (other) Latin stop, nor can they occur word-finally. The voicing contrast between ⟨nqu⟩ and ⟨ngu⟩ is not found in any (other) sequence of three consonants. Assimilation of the prefix ad- to a following ⟨qu⟩ is relatively rare, which is also the case when ad- is followed by a consonant cluster. The Proto-Indo-European predecessor of Latin ⟨qu⟩ is, in many cases, reconstructed as a single consonant */kʷ/, notably distinct from sequences of */kw/. Occasionally Latin /w/ scans as a vowel in poetry, when preceded by /s/ or /l/, but this is never the case for the ⟨u⟩ of ⟨qu⟩. Allen, William Sidney (1978) [1965]. Vox Latina—a Guide to the Pronunciation of Classical Latin (2nd ed.). Cambridge University Press. ISBN0-521-37936-9. Cser, András (2020). "The phonology of Classical Latin". Transactions of the Philological Society. 118. Publications of the Philological Society: 1–218. doi:10.1111/1467-968X.12184. S2CID219404384.
Cser 2020, §4.9. In footnote 206, he adds: "The evidence has been thoroughly assessed in the diachronic literature; see Sen (2012: 472–3; 2015: 15 sqq.), Meiser (1998: 68–9), Leumann (1977: 85–7)." Cser, András (2020). "The phonology of Classical Latin". Transactions of the Philological Society. 118. Publications of the Philological Society: 1–218. doi:10.1111/1467-968X.12184. S2CID219404384.
Gilbert 1939 Gilbert, Allan H (June 1939). "Mock Accents in Renaissance and Modern Latin". Publications of the Modern Language Association of America. 54 (2): 608–610. doi:10.2307/458579. JSTOR458579. S2CID164184102.
Gilbert 1939 Gilbert, Allan H (June 1939). "Mock Accents in Renaissance and Modern Latin". Publications of the Modern Language Association of America. 54 (2): 608–610. doi:10.2307/458579. JSTOR458579. S2CID164184102.
Allen 1978 (p. 17) judges the evidence to favour /kʷ ɡʷ/, while Cser 2020 (§2.2.2) comes to the opposite conclusion. The relevant facts, per the latter, are as follows:
⟨qu⟩ enjoyed a wide lexical distribution, while ⟨gu(V)⟩ was limited to a dozen or so words, where it was always preceded by /n/. The grammarian Velius Longus indicated that the ⟨u⟩ of ⟨qu⟩ was in some way different from /w/ in general. No geminate *⟨ququ⟩ is attested, whereas all (other) Latin stops are also found as geminates. Sequences of obstruent + glide are rare in Classical Latin. In poetry, whenever sequences of stop + glide occur in medial position, the scansion reveals that can be split across syllables, but this is never the case for ⟨qu⟩. Neither ⟨qu⟩ nor ⟨gu⟩ are ever followed by a consonant, unlike any (other) Latin stop, nor can they occur word-finally. The voicing contrast between ⟨nqu⟩ and ⟨ngu⟩ is not found in any (other) sequence of three consonants. Assimilation of the prefix ad- to a following ⟨qu⟩ is relatively rare, which is also the case when ad- is followed by a consonant cluster. The Proto-Indo-European predecessor of Latin ⟨qu⟩ is, in many cases, reconstructed as a single consonant */kʷ/, notably distinct from sequences of */kw/. Occasionally Latin /w/ scans as a vowel in poetry, when preceded by /s/ or /l/, but this is never the case for the ⟨u⟩ of ⟨qu⟩. Allen, William Sidney (1978) [1965]. Vox Latina—a Guide to the Pronunciation of Classical Latin (2nd ed.). Cambridge University Press. ISBN0-521-37936-9. Cser, András (2020). "The phonology of Classical Latin". Transactions of the Philological Society. 118. Publications of the Philological Society: 1–218. doi:10.1111/1467-968X.12184. S2CID219404384.
Cser 2020, §4.9. In footnote 206, he adds: "The evidence has been thoroughly assessed in the diachronic literature; see Sen (2012: 472–3; 2015: 15 sqq.), Meiser (1998: 68–9), Leumann (1977: 85–7)." Cser, András (2020). "The phonology of Classical Latin". Transactions of the Philological Society. 118. Publications of the Philological Society: 1–218. doi:10.1111/1467-968X.12184. S2CID219404384.
Gilbert 1939 Gilbert, Allan H (June 1939). "Mock Accents in Renaissance and Modern Latin". Publications of the Modern Language Association of America. 54 (2): 608–610. doi:10.2307/458579. JSTOR458579. S2CID164184102.
uconn.edu
homepages.uconn.edu
There is, however, a fringe view that the short high vowels /iu/ were tense [iu] and that the long mid vowels /eːoː/ were lax [ɛːɔː], implying that none of the Latin short-long vowel pairs differed in quality (Calabrese 2005). Calabrese, Andrea (2005). "On the Feature [ATR] and the Evolution of the Short High Vowels of Latin into Romance"(PDF). University of Connecticut Working Papers in Linguistics. 13: 33–78. Archived from the original(PDF) on 13 November 2012.
web.archive.org
There is, however, a fringe view that the short high vowels /iu/ were tense [iu] and that the long mid vowels /eːoː/ were lax [ɛːɔː], implying that none of the Latin short-long vowel pairs differed in quality (Calabrese 2005). Calabrese, Andrea (2005). "On the Feature [ATR] and the Evolution of the Short High Vowels of Latin into Romance"(PDF). University of Connecticut Working Papers in Linguistics. 13: 33–78. Archived from the original(PDF) on 13 November 2012.