Retroflex sounds occur in most
Dravidian languages of
India, and it is quite likely that this is the source of the
Sanskrit series, which does not occur in any closely related language; and in most
Australian Aboriginal languages. At one time this, and a superficial similarity in appearance, was held to constitute a link between them: it is no longer.
The place of articulation of these sounds is technically postalveolar, with retroflexion as an additional articulation of the tongue; but usually the retroflex series are treated as a distinct place of articulation between dental and palatal. This applies to both Indian and Australian languages. Another similarity is that in both groups there are usually sounds at a third place of articulation, alveolar, between dental and retroflex.
t d n l r s z can be retroflex. In the IPA phonetic symbols they are indicated by continuing the letter down below the line and curling it to the right. Another common sign, such as in more traditional descriptions of Indian languages, is to put a dot under the letter. Some Australian Aboriginal communities use an orthography that represents retroflexes by a line under the letter. (In IPA this means alveolar instead.) Examples are in Uluru and Kata Tjuta, famous sacred places of the Anangu people; others use the common convention in Australian linguistics of writing rt rn rl.
Similarly, in Swedish, Norwegian, and Faroese the letter r before one of these is silent, but makes the next consonant retroflex, as in Lars or Björn. The result is that Lars sounds somewhat like "lahsh". But this does not happen in Danish or Icelandic. (montecarlo tells me Swedish rs is postalveolar these days; and on the radio I noticed so is the sound in a strong Orkney accent.)