Take our survey

Ancient Sanskrit Online

Appendix 1

Karen Thomson and Jonathan Slocum


I have followed Michael Coulson's useful example in Teach Yourself Sanskrit in giving the changes that occur in the lessons in the form of tables, to which the reader can refer when necessary. This section is designed for reference; a detailed account of Sanskrit sandhi can be found in the grammars listed in section 9 of the Series Introduction.

1. Sandhi of vowels.

The systematic change of final -i/-ī and -u/-ū to the semivowels y and v, made by the ancient editors, is not included here, as it is nearly always restored by the metrical text (one of three of these changes is however retained in the first line of the lesson 4 text, and exceptions will also be found in examples 216, 233 and 250). The sandhi of final -au is however given below, as it remains in the metrical text (there is no change of quantity involved to show that it was incorrect).

The table gives the vowel combinations that regularly occur in addition to the merging of similar vowels described in section 7.1 of the Series Introduction. The end-vowel is listed at the top of the column, and the following initial vowel down the left-hand side. Final does not occur, and original final o is rare, as are initial ai and au. These are not included.

    -a/-ā   -e   -ai   -au
a     no change   ā a   āv a
    a ā   ā ā   āv ā
i/   e   a i/   ā i/   āv i/
u/   o   a u/   ā u/   ā u/
  ar   a r̥   ā r̥   āv r̥
e   ai   a e   ā e   āv e
o   au   a o   ā o   āv o

Note: when a long vowel is followed by a short vowel and the metre indicates that no contraction takes place, the first vowel is generally rendered prosodically short.

2. Sandhi of consonants.

Only certain non-vocalic sounds are permissible at the end of words. The 'permitted finals' are k, , , t, n, p, m (), . The unvoiced breathing sound always appears at the end of words for original s, or r (final s is very common, final r uncommon). Of these permitted finals, , , and p are rare, and have not been included in the table. The labial nasal m occurs regularly, and the way it changes is straightforward: it always becomes the pure nasal when followed by a consonant of another class. Other changes are given in tabular form below.

Where the table gives forms such as 'c+ch', 'd+dh', this represents a double change: for instance, yád dha in example 134 in the lessons is the result of the juxtaposing of yát and ha: both words are altered.

The sandhi of sás and its derivatives eṣás and syás is exceptional: the final s () is dropped before all consonants.

    -k   -t   -n   -ḥ (except after a or )   -aḥ   -āḥ
k/kh   k   t   n   [1]   aḥ[8]   āḥ
g/gh   g   d   n   r   o  
c/ch   k   c   /ṃś     aś   āś
j   g   j     r   o  
t   k   t   ṃs   s[2]   as   ās
d/dh   g   d   n   r   o  
p/ph   k   t   n   [1]   aḥ[8]   āḥ
b/bh   g   d   n   r   o  
n/m     n   n   r   o  
y/v   g   d   n[3]   r   o  
r   g   d   n[3]   zero[4]   o  
  k   c+ch   ñ+ch     aḥ   āḥ
/s   k   t   n   [5]   aḥ[5]   āḥ[5]
h   g+gh   d+dh   n[3]   r   o  
vowels   g   d   n/nn//m̐r[6]   r   a/o[7]  
  • [1] But is/us becomes iṣ/uṣ (see examples 297 and 212)
  • [2] If preceded by i/, u/ occasionally becomes , and the following t then becomes (see examples 196 and 203)
  • [3] But occasionally becomes or m̐r, as before vowels (see footnote 6; an example occurs in the third verse of the Lesson 8 text)
  • [4] If a/i or u precedes it is lengthened (see example 18)
  • [5] But if the sibilant of the following word is immediately followed by voiceless consonant (like th) the is dropped (see example 239, jágata sthātúḥ)
  • [6] Final n is doubled following short vowels if an original final consonant has been lost, otherwise it remains unchanged. Following long vowels it behaves differently: ān becomes ām̐, and īn/ūn become īm̐r or ūm̐r (see the third verse of the Lesson 8 text, and example 237). The ān of the 3rd person plural subjunctive however remains (as originally *-ānt; see Lesson 3 text, final verse)
  • [7] Becomes o before a, otherwise a
  • [8] But the original ending -as remains in compounds, and some collocations tending to become compounds, as in bŕ̥has-páti, bráhmaṇas páti (see section 35.3)

  •   Map
  • Linguistics Research Center

    University of Texas at Austin
    PCL 5.556
    Mailcode S5490
    Austin, Texas 78712

  • For comments and inquiries, or to report issues, please contact the Web Master at UTLRC@utexas.edu