Devanagari Sha श

Sha is the thirtieth consonant of the varnamala and the first of the three sibilants, which the grammarians group together with h in a class they call ushman, a word for heat or steam that describes the continuous breath these sounds require rather than the closure the stops make. It is the palatal sibilant, made with the tongue against the hard palate, and it is the sound an English speaker hears as sh. It opens the names of the deity and of the power that stand at the centre of the Shaiva and Shakta traditions, both of which have entries elsewhere in this atlas. Sanskrit distinguishes three sibilants where English has two, and keeping them apart is why the scholarly transliteration systems mark this one with an acute accent and the retroflex one with a dot beneath.

Popular spellings that avoid diacritics write both as sh and merge the distinction, which is why the same Sanskrit word appears in English books in several forms. That merging also happens in speech: across much of northern India this sibilant and the retroflex one are pronounced alike. The order of the three sibilants is worth noticing for what it says about the table as a whole. They are arranged palatal, retroflex, dental, which is the same front-to-back progression the five stop classes follow, so the ushman class repeats the ordering principle of the vargas on a shorter series rather than introducing a second principle at the end of the alphabet.

The three sibilants sit at three of the five places of articulation, palatal, retroflex and dental, and in classical Sanskrit they are fully distinct: words differ by which one appears, and the grammar states rules that convert one into another under defined conditions. In the speech of most present-day users of the Devanagari-writing languages the palatal and the retroflex have merged, so the script preserves a distinction that its readers no longer make. That situation is common in long-lived writing systems and it has a specific consequence for this atlas: a description of these letters that reported only modern pronunciation would misdescribe the classical values the letters were designed for, and one that reported only the classical values would leave a reader puzzled by what they actually hear. Both are recorded.

The absence of a labial sibilant is worth noting too, since it shows that the ushman class is not simply the sibilant row of a five-place grid: Sanskrit has sibilants at three places only, and the class is completed by h, which is not a sibilant at all but is grouped with them because it shares the continuous airflow. The table follows the sound system rather than imposing a shape on it.