As stated in Xie (2016):
The pronunciation of /ing/ becomes /iəŋ/ in typical Mandarin speech, with a transitional schwa between oral vowel and nasalized vowel.
This excrescent schwa is considered to help perceptually distinguish the velar nasal ending /-ŋ/ from the alveolar nasal /-n/. As Chang & Hsieh (2018) states:
Coronization, at first blush, may be treated as a “conflict” between an advanced tongue body target for the nonlow vowels (/i, ɤ/) and a retracted target for the following tongue dorsum constriction for /ŋ/. The specific conflict is resolved by yet another possible strategy: an “excrescent schwa,” i.e., [iəŋ] in Standard Chinese.
They found that this excrescent schwa was likely present for both Standard (Beijing accent) Chinese and for their Taiwanese Mandarin speakers, inferred from substantial back-to-front movement of tongue blade sensors detected in both groups.
However, to many speakers, the presence of the [iəŋ] variant is marked for a 'Northern' quality. This may even be considered non-standard, as Li (2004) explains:
The problem with ‘ing’ is that in most dialects of Mandarin, Beijing included, the rhyme is not pronounced as it is spelled, but rather with a schwa offglide — [iᵊŋ]. It is for this reason that in traditional verse and nursery rhymes from the last century ‘ing’ is always allowed to rhyme with ‘eng’ — a violation of rhyme conventions if ‘ing’ is not represented as phonological /iᵊŋ/.
However, since the arrival of Modern Standard Chinese in the 1920s,
instructors of the standard language have taught learners to pronounce ‘ing’ as [iŋ], in line with official orthography, and the push has been so successful that [iŋ] is how one must pronounce the final today to be taken as educated — [iᵊŋ] is now considered non-standard or uneducated.
Li also points out how the Zhuyin spelling equivalent of Pinyin -ing and -in, ㄧㄥ and ㄧㄣ respectively, is actually equivalent to -i + -eng and -i + -en, but is counter-intuitively pronounced with a monophthong for both in the standard Taiwanese Mandarin pronunciation.
This is similar to how -ing is taught to those who are learning Mandarin Chinese as a second language. As Squizzero (2022) points out:
The common production of Mandarin /iŋ/ as [iᵊŋ] is also often omitted from L2 Mandarin curricula, despite its possible perceptual relevance for L2 listeners as a way to distinguish [iŋ] from [in] in cases of nasal coda consonant deletion. [...] Deletion is considered
standard, and occurs even in the speech of news and radio broadcasters. The contrast between the two nasals, however, is preserved based on the quality of the previous vowel, as all vowels have separate allophones when preceding [n] vs. [ŋ], with the exception of speakers who produce [iŋ] instead of [iᵊŋ], discussed above.
This also leads to a rather particular situation in Taiwan, where Taiwanese Hokkien does have /iᵊŋ/, but those same speakers associate the Taiwanese Mandarin -ing /iŋ/ with /in/ instead of /iᵊŋ/ because of the difference in the vowel quality, thus leading to a -in/-ing merger. This in turn has prompted urban Taiwanese Mandarin to undergo hypercorrection, using a sort of in-between nasal that has been transcribed as /iɲ/, and generally perceived closer to /iŋ/, while still having merged the two.
Is this not ironically the opposite from what we see from Forvo? Yang (2010) finds that Taiwan Mandarin is merging /iŋ/ to /in/, southern Mainland Chinese Mandarin is doing the same thing, and Beijing Mandarin is inconsistently doing the opposite, merging /in/ to /iŋ/.
However, Chiu & Lu (2020) state that for some speakers of Taiwan Mandarin:
Although nasals were merged in terms of tongue posture, the degrees of nasalization of the preceding vowel were contrastive, suggesting that the merging process may be incomplete.
Note the emphasis on the vowel before. On the other hand, for Shanghai Mandarin, Faytak, Liu & Sundara (2020) have found:
The highest vowel /i/ conditioned a consistent, non-structure-preserving neutralization to a fronted velar or palatal nasal. [...] In contrast, Standard Mandarin controls produced the contrast in all vowel contexts. A follow-up perception experiment confirmed that, consistent with complete neutralization, Shanghai Mandarin listeners failed to distinguish alveolar and velar nasals following /i/. More surprisingly, so did Standard Mandarin listeners, suggesting a near merger of the nasal codas following /i/ in Standard Mandarin.
Thus, we have substantial variation in production. Could one attribute this to different "standardisations" of Taiwanese Mandarin and Mainland Mandarin post-2010, especially in pedagogy? Is ㄧㄥ '-ieng' in Taiwan vs -ing on the Mainland having such a big effect on people's 'standard' pronunciations? Maybe something for the research literature to work on!