N.B. it is common practice in historical linguistics to use an asterisk to mark an unattested form. Because Proto-Indo-European (PIE) existed in prehistory only, it is not attested; thus PIE forms and phonemes are written here with a preceding asterisks. For Sanskrit, we use the Harvard-Kyoto system of transliteration, as devanAgarI would unfortunately be unreadable to a large class of readers.
Proto-Indo-European had a number of phonemes, the developments of which into Sanskrit (as well as other Indo-European languages) for the most part follow relatively regular patterns.
- Monophthongs: *e, *ee, *o, *oo, *a, *aa (short/long middle front, middle back, and low back)
- Semivowels: *i/*y, *u/*w, *R/*r, *L/*l, *M/*m, *N/*n (Each semivowel has a vocalic (vowel) form, a consonantal form, and a form consisting of the vocalic form plus the consonantal form; which form the phoneme took depended solely on the surrounding phonemic environment. Thus *i/*y was *i between consonants, *y between vowels, and so forth.)
- Diphthongs: any monophthong followed by *i or *u.
In Sanskrit, the PIE labial series developed unmodified, as did the dental series and the sibilant *s. However, in some contexts there were some modifications. For example, PIE palatovelar stops sometimes developed as cerebral (retroflex) stops in Sanskrit, especially before dentals. Similarly, by a rule of sandhi often known as the `RUKI' rule, Sanskrit s (and hence PIE *s) becomes s. (a retroflex sibilant) when directly preceded by r, R, RR, u, U, k, i, or I (including Sanskrit diphthongs such as o, au, e, and ai which end in those vowels). Finally, *bh and *dh developed as h in some Sanskrit dialects; some of these forms eventually gained general use.
The palatovelar system changed significantly more. By the first palatalization (one of the major early events in the developent of the satem languages in the IE family), this series lost most of its velar quality. *k' developed into z, a voiceless palatal sibilant; written ç in Whitney and ś in modern academic transliteration); *g' into j, originally a voiced palatal stop, but eventually becoming something of an affricate; and *g'h into h, a voiced guttural sibilant.
The changes occurring to the labiovelar series were even more complicated, as they were conditional developments. Before PIE front vowels and semivowels (*e, *ee, and *i/*y) they became palatal by what is known as the second palatalization, which is specific to the Indo-Iranian branch of IE. That is, before *e, *ee, or *i/*y, *kw became c (a voiceless palatal stop, becoming later a voiceless palatal affricate); *gw became j; and *gwh became h. However, before other sounds, the labiovelars lost their labial quality and became velar stops (k, g, and gh).
Other developments alter further still Sanskrit phonemics. For example, Grassman's Law (which also applies to other languages, such as Greek) states that a syllable-initial aspirate becomes unaspirated if followed in the same syllable or at the beginning of the next syllable by another aspirate. Thus PIE *bheudh- `to be awake or aware' developed into the Sanskrit root budh-, from which buddha `awakened'. In addition, the Sanskrit voiceless aspirates may come from PIE laryngeals, the existence of which is still under debate.
The resonants followed much simpler rules of developments. All the PIE monophthongs (*e, *a, *o, and the long versions of these) developed into Sanskrit a or aa, depending on their length. PIE semivowels were more varied: *i/*y and *u/*w developed more or less unchanged (into i, y, u, and v; later processes gave i and u long forms), *R and *L developed into R, *r and *l into r, *M and *N into a, *m into m, and *n into n.
Diphthongs developed as we would expect from the monophthong and semivowel developments and sandhi. *ai, *ei, and *oi all became e (ay before vowels); *aai, *eei, and *ooi became ai (Ay before vowels); *au, *eu, and *ou became o (av before vowels); and *aau, *eeu, and *oou became au (Av before vowels). By a development known as the Silver-Edgerton phenomenon, the vocalic+consonant of the semivowels *r and *l (that is, *Rr and *Ll) became ir or ur and il or ul.
Later developments within the Indo-Iranian language group present further evolution of Sanskit (and, later, Prakrits and the entire modern Indic language family). These developments, however, are outside the scope of this node. Also outside the scope of this node are the ways the evolution of Sanskrit's phonology affected its morphology---a topic for the as-yet-unwritten node `development of the Sanskrit morphological system'.