This is from the Wikipedia article on retroflex consonant, but isn't this wrong? I assume that Mandarin zh, ch, sh, and r sound should be pronounced with your tongue curled up, rather than "flat?"

Laminal post-alveolar, with a flat tongue. These occur, for example, in Polish cz, sz, ż (rz), dż and Mandarin zh, ch, sh, r.


These consonants are traditionally described as "retroflex" ("bent backwards"), meaning the tongue tip curls back towards the throat. However, Ladefoged & Wu (1984) (also discussed on Ladefoged & Maddieson (1996)) have carefully studied X-ray photographs of native Mandarin speakers, and showed conclusively that they're produced with a retracted tongue blade, either flat or a little bit convex, with no curling back of the tongue tip:

Speakers A, B, C pronounced the consonants represented in pinyin as 's', 'sh', 'x' (Top to bottom rows are equivalent to pinyin: s, sh, x).

Discussion (emphases mine):

The Standard Chinese so-called retroflex ṣ [=pinyin 'sh'] is shown in the middle row […] This gesture is plainly very different from that in the retroflex stops discussed in chapter 2. It does not involve the tip of the tongue being curled up and backwards into the palatal region, as in the Dravidian sub-apical retroflex stops, nor does it have the apical post-alveolar shape that occurs in the Hindi retroflex stops shown in figure 2.11. In our Standard Chinese data, all three speakers produce the constriction for this sound with the upper surface of the tip of the tongue, making it a laminal rather than an apical post-alveolar. The constriction is at about the same place for all three speakers, further back than in s, so that it is nearer to the center of the alveolar ridge. Both the height and the width of the channel are greater than in s, but the width varies considerably, from 18.5 mm for speaker A to 5 mm for speaker C. The location and width of the constriction are thus very comparable with those for English ʃ. The front of the tongue is fairly flat for speakers A and C, and slightly hollowed for speaker B, rather than being slightly raised towards the hard palate as it is in ʃ. Because the part of the tongue immediately behind the constriction is not domed as it is for ʃ, we have termed this sound a flat post-alveolar sibilant. A further point to note about this gesture is that no part of the tongue is touching the lower teeth, as occurs in the articulation of s. Instead the tongue is drawn slightly back, so that there is a sublingual cavity. Perkell, Boyce and Stevens (1979) have shown that a cavity of this type has a significant acoustic effect, producing a comparatively low frequency spectral peak. Additional x-ray data in other publications (Zhou and Wu 1963, Ohnesorg and Svarny 1955) all show substantially the same gesture, confirming the notion that Standard Chinese ṣ is a (laminal) flat post-alveolar sibilant. The traditional description of this sound as a retroflex is inappropriate as a description of its articulation.

Perhaps the “retroflex” misnomer comes from early Orientalists who thought these consonants sounded similar to the Sanskrit retroflex series.


  • Ladefoged, Peter and Maddieson, Ian. The Sounds of the World's Languages. 1996.
  • Ladefoged, Peter, and Wu, Zongji. Places of articulation: An investigation of Pekingese fricatives and affricates. Journal of Phonetics 12.3, 1984.
  • 1
    My understanding is that the misnomer actually comes from ancient Indian grammarians themselves who analysed this way the stops as well as the fricatives (because let's admit it, the system is more elegant this way and the acoustic effect change is not dissimilar). Sanskrit /š/ sound is basically non-voiced version of the Chinese sound recorded as <zh>/<r>, so it is natural the traditional designation was extended to it. For Chinese though, it will depend on dialect - let's not forget phonologically this is a rhotic.
    – Eleshar
    Sep 1 '16 at 12:40
  • Oh, that's interesting. I had assumed Sanskrit had actual retroflex sibilants, given that Dravidian and Hindi have true retroflex stops (well, Hindi less flexed than Dravidian). But I checked Ladefoged & Maddieson again and turns out that only Toda (Dravidian) has a true retroflex sibilant. Sep 1 '16 at 12:57
  • 1
    It is traditionally labeled as retroflex and thought to correspond to [ʂ] since the contrast was three-way like today's Polish s, sz, ś or today's Chinese and there were certainly true retroflex stops in sanskrit. Whether it was articulated in truly retroflex manner or it was just darker [ʃ] articulated e.g. as the Ladefoged's column C, row 2, is unclear.
    – Eleshar
    Sep 1 '16 at 15:30

In terms on laminal consonants, the "flat" part specifically refers to the part of the tongue which contacts the roof of the mouth. So the body of the tongue is curled up, and the blade of the tongue is flat against the roof of the mouth. In practice, there is very little surface area actually in contact, but the point of contact sets the position of the whole tongue.

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.