ac sandhi is the general name for sandhi changes that involve two vowels. For example, the rule we studied in the previous lesson is an example of ac sandhi:
इको यणचि। ६.१.७७
iko yaṇaci (6.1.77)
ikaḥ yaṇ aci
An ik vowel is replaced by its respective yaṇ sound when a vowel follows [in saṃhitā].
ac sandhi has many rules of varying importance. Some are truly minor, and others are important general patterns. Here, we will focus on the important general patterns and complete our basic picture of ac sandhi.
But there is also an important issue we should address: rule 6.1.77 has a serious flaw. In Sanskrit, two similar vowels should combine and become long:
द्रौपदी इन्द्रम् अपश्यत् → द्रौपदीन्द्रम् अपश्यत्
draupadī indram apaśyat → draupadīndram apaśyat
Draupadi saw Indra.
But rule 6.1.77 will produce an error:
* द्रौपदी इन्द्रम् अपश्यत् → द्रौपद्यिन्द्रम् अपश्यत्
* draupadī indram apaśyat → draupadyindram apaśyat
Draupadi saw Indra.
So in addition to completing our basic picture of ac sandhi, we will also ensure that our system handles the example above correctly.
As in the previous lesson, the rules below will borrow context from the adhikāra rule 6.1.72:
In saṃhitā, …
ec as first vowel
Once we understand rule 6.1.77 (iko yaṇaci), we can easily understand 6.1.78:
एचो ऽयवायावः। ६.१.७८
eco 'yavāyāvaḥ (6.1.78)
An ec vowel becomes ay, av, āy, or āv, respectively [when a vowel follows in saṃhitā].
So we get sandhi changes like this:
ने + अ → नय
ne + a → naya
भो + अ → भव
bho + a → bhava
Rule 6.1.78 is nice and simple. Now let's turn to the rules where a is first:
a as first vowel
In Sanskrit, a will combine with most vowels to form a compound vowel:
सीता इन्द्रम् अपश्यत् → सीतेन्द्रम् अपश्यत्
sītā indram apaśyat → sītendram apaśyat
Sita saw Indra.
Here, one vowel (e) replaces two vowels (ā and i). How might we model this behavior with our rules?
Pāṇini approaches this problem by creating a new adhikāra rule:
एकः पूर्वपरयोः। ६.१.८४
ekaḥ pūrvaparayoḥ (6.1.84)
A single [term] is substituted for the previous and following.
What does this rule mean? Recall that we represent “previous” terms with the fifth case and “following” terms with the seventh. So, this rule means that in the scope of this adhikāra rule, terms in the fifth and seventh case are both replaced by a single term.
With this context in place, we can model what happens when a is the first vowel. In general, the change is simple. With the help of a new term:
The vowels a, e, and o are called guṇa.
we can define our rule:
a [and the following vowel] become [a single] guṇa [in saṃhitā].
But if the second vowel is a compound vowel, we use a slightly different rule. Again, we define a new term:
The vowels ā, ai, and au are called vṛddhi.
And use it in our rule:
[a and the following] ec vowel become [a single] vṛddhi [in saṃhitā].
The terms guṇa and vṛddhi are important and will be used throughout the grammar. We will return to them later on.
Substitution with the closest option
Unfortunately, rules 6.1.87 and 6.1.88 have a problem that we've seen before: these rules don't tell us which specific vowel to use. We know what the correct result should be, but the rule allows some clearly incorrect results:
सीता इन्द्रम् अपश्यत् → सीतोन्द्रम् अपश्यत्
sītā indram apaśyat → sītondram apaśyat
Sita saw Indra.
So Pāṇini offers this rule to help us perform the correct substitution:
स्थाने ऽन्तरतमः। १.१.५०
sthāne 'ntaratamaḥ (1.1.50)
In substitution, the closest [is preferred].
Roughly, “closeness” refers to properties like points of articulation, semantics, and so on. Since these rules are about sounds, the closest replacement is the one that matches the points of articulation of the sounds being replaced.
If we return to our example above, ā is pronounced at the soft palate and i is pronounced at the hard palate. e uses both of these points of articulation, so it is a good match. o uses the soft palate, but it does not use the hard palate; it uses the lips instead. So o is not a good match. e is a closer substitute than o, so we should choose e:
अ + इ → ए
a + i → e
And likewise, o is better if the combination is a and u:
अ + उ → ओ
a + u → o
Addition of r
The rules above seem to work as intended. But if we test this rule against our Sanskrit knowledge, we find another problem. In Sanskrit, ṛ has no compound vowel. Instead, it combines with a with some help from the semivowel r:
सीता ऋच्छति → सीतर्च्छति
sītā ṛcchati → sītarcchati
But with our current system, rule 6.1.87 (ādguṇaḥ) can hardly function. at seems like the closest guṇa vowel, but this produces a bad result:
* सीता ऋच्छति → सीतच्छति
* sītā ṛcchati → sītacchati
The fix is another paribhāṣā about how to perform a substitution:
uḥ aṇ ra-paraḥ
[In substitution,] an aṇ vowel that replaces an ṛ is followed by r.
And with this rule in hand, we can perform the substitution correctly and get the desired result.
Two similar vowels
Finally, we can return to the example from the start of this lesson and complete our basic picture of vowel sandhi. Recall the example we wish to model:
द्रौपदी इन्द्रम् → द्रौपदीन्द्रम्
draupadī indram → draupadīndram
To handle this special case, we just need a new rule:
अकः सवर्णे दीर्घः। ६.१.१०१
akaḥ savarṇe dīrghaḥ (6.1.101)
akaḥ savarṇe dīrghaḥ
ak and a following savarṇa [vowel] become a dīrgha (long) [in saṃhitā].
The meaning of this rule is clear. And with rule 1.1.50 (sthāne'ntaratamaḥ), it is also clear what the result should be for each vowel:
अ + अ → आ
a + a → ā
इ + इ → ई
i + i → ī
उ + उ → ऊ
u + u → ū
ऋ + ऋ → ॠ
ṛ + ṛ → ṝ
With just a few short rules, we have fully characterized the basic patterns of ac sandhi. This is the power the Pāṇinian system gives us.
In the next lesson, we explore a critical problem with our current system and learn how Pāṇini decides to solve it.