savarṇa sounds

The previous lesson described how the Shiva Sutras let us refer to different sound groups concisely. But we still have some important open questions:

  • In the Shiva Sutras, why does the sound a refer to both short a and long ā?

  • What is an it letter, really?

This lesson will answer the first question, and the next lesson will answer the second question. As we answer these questions, we will also see some actual rules from the Aṣṭādhyāyī and enter the system more deeply.

A new problem

To start the discussion, here is a small sandhi change:

  • सीता अश्वम् इच्छति → सीताश्वम् इच्छति
    sītā aśvam icchati → sītāśvam icchati
    Sita wants a horse.

The vowels ā and a combine to a single shared vowel ā. And there are other combinations possible, too:

  • अ + अ → आ
    a + a → ā

  • अ + आ → आ
    a + ā → ā

  • आ + आ → आ
    ā + ā → ā

All four of these combinations are part of the same general idea: if any two “a” vowels combine, the result is ā:

  • अ/आ + अ/आ → आ
    a/ā + a/ā → ā

How can we refer to the category of “a” vowels concisely? More generally, some sounds are similar to each other in an important way. How can we concisely refer to similar sounds?

Our first rule

To address the question above, we need several rules. Pāṇini starts by introducing this rule:

  • तुल्यास्यप्रयत्नं सवर्णम्। १.१.९
    tulyāsyaprayatnaṃ savarṇam (1.1.9)
    tulya-āsya-prayatnam savarṇam
    [Sounds with] the same āsya (place of articulation) and prayatna (articulatory effort) are called savarṇa (similar).

This is the first rule we've seen from the Aṣṭādhyāyī, so let's dwell on it for a moment:

  • The numbers 1.1.9 mean that this is chapter 1, part 1, rule 9. The Aṣṭādhyāyī has eight chapters, and each chapter has four parts. You can click on these numbers to see the rule's traditional interpretation and commentary, as collected on

  • First, we show the original rule in Devanagari and Roman script.

  • Next, we show the rule with its sandhi changes removed and with its compound separated, so that it is easier to understand. To save space on your screen, we've written this version in just Roman script.

  • In the translation, the words in (parentheses) are short translations of the Sanskrit terms they follow.

  • In the translation, the words in [brackets] are not explicitly in the rule and must be provided from context. Context may come either from prior rules or from our prior knowledge of the system.

Now, what does this rule actually mean?

  • āsya refers to one of the places of articulation in the mouth: the soft palate (where we pronounce ka), the hard palate (ca), the alveolar ridge (ṭa), the teeth (ta), or the lips (pa).

  • prayatna refers to how these sounds are pronounced: with full contact between places of articulation (as with ka), with partial contact (ya), or with no contact (a).

So, sounds with the same āsya and prayatna are called savarṇa, which means “similar.” This rule defines the term savarṇa, which can then be used in the rest of the system. Rules that define a term are called saṃjñā (“designation”) rules.

However, this rule is too general. By 1.1.9, the sounds i and ś could be counted as savarṇa with each other. This definition will cause many problems later. Has Pāṇini made a mistake?

Counteracting a rule

Pāṇini has not made a mistake. In the Aṣṭādhyāyī, it is common for one rule to state a general principle and for another to counteract it. To counteract the overapplication of 1.1.9, we have rule 1.1.10:

  • नाज्झलौ। १.१.१०
    nājjhalau (1.1.10)
    na ac-halau
    But vowels and consonants are not [savarṇa with each other].

Notice that the phrase “savarṇa with each other” is inferred from the context of rule 1.1.9 above. This extension of context from one rule to another is called anuvṛtti. There are specific principles that we can use to define anuvṛtti. But for now, let's just focus on understanding this rule.

Rule 1.1.10 refers to ac and hal, which we learned about in the previous lesson: ac refers to all vowels, and hal refers to all consonants.

With the extra context and these two definitions in mind, the meaning of the rule is clear. Rule 1.1.10 prevents sounds like i and śa from being savarṇa with each other. Together, 1.1.9 and 1.1.10 give us a complete definition of the term savarṇa.

Defining groups of sounds

Now that we have a complete definition of savarṇa, we can return to our original problem: how can we concisely refer to similar sounds?

Pāṇini's solution is to provide these two rules:

  • स्वं रूपं शब्दस्याशब्दसंज्ञा। १.१.६८
    svaṃ rūpaṃ śabdasyāśabdasaṃjñā (1.1.68)
    svam rūpam śabdasya a-śabda-saṃjñā
    A word [denotes] its own form if it is not a definition (saṃjñā).

  • अणुदित् सवर्णस्य चाप्रत्ययः। १.१.६९
    aṇudit savarṇasya cāpratyayaḥ (1.1.69)
    aṇ-udit savarṇasya ca a-pratyayaḥ
    The aṇ sounds and udit [sounds] also [denote] their savarṇa [sounds], if they are not pratyaya (suffixes).

Rule 1.1.68 does not define a term or counteract a rule. Instead, it is an instruction for us as we read the grammar. Such rules are called paribhāṣā (“explanation”) rules. paribhāṣā rules usually apply throughout the entire system, so it's important to understand them well.

What does rule 1.1.68 actually mean? We can make the rule clear with an example. Suppose that we see some rule about the word agni, which means “fire.” The point of rule 1.1.68 is that such a rule is about the specific form agni, and not about any other word that means “fire.”

Rule 1.1.69 then borrows this context to define another paribhāṣā. (That is, it inherits some context by anuvṛtti from rule 1.1.68). Rule 1.1.69 also uses two interesting terms:

  • aṇ has two interpretations, as we learned in the previous lesson. Here, it is the aṇ that includes all vowels and semivowels.

  • udit will be explained in the next lesson. For now, treat “udit sounds” as meaning “sounds followed by u.”

With these terms defined, we can see what 1.1.69 does for us:

  • In the grammar, a will refer to both itself and ā, which is savarṇa to it. Likewise for i, u, and so on.

  • In the grammar, ku̐ will refer to both ka and the four sounds kha, ga, gha, and ṅa, all of which are savarṇa to ka. And likewise for cu̐, ṭu̐, tu̐, and pu̐. ( is a nasal u. Why is this vowel nasal? We'll explain in the next lesson.)

So with these four rules, we can now refer to similar sounds simply and concisely. But this system also creates a new problem. What if we want to refer to short a but not long ā? It seems that we can't do that anymore. Has Pāṇini made a mistake?

Referring to short and long vowels

Pāṇini has not made a mistake. We have one more rule to consider:

  • तपरस्तत्कालस्य। १.१.७०
    taparastatkālasya (1.1.70)
    ta-paraḥ tat-kālasya
    [A sound] bordered by t [refers to the sound] with that duration.

Rule 1.1.70 follows right after rule 1.1.69, which we saw above. Note that it continues to use context provided from 1.1.69.

What does this rule actually mean? It means that at refers to the short vowel a but not to ā. Similarly, it means that āt refers to ā but not to the short vowel a. With this new rule, we can always tell these vowels apart.

This rule also explains part of the term udit, which we saw in 1.1.69 above. udit is ut-it: a term that has the vowel ut (short u) as an it letter. (But what is an it letter, really? We will answer that question soon.)

Different kinds of vowels

As a closing thought, perhaps you are wondering if rule 1.1.70 is worth the extra effort. Is this rule really necessary?

Yes. Sanskrit vowels make many important distinctions. They can differ in length:

  • ऊकालो ऽज्झ्रस्वदीर्घप्लुतः। १.२.२७
    ūkālo 'jjhrasvadīrghaplutaḥ (1.2.27)
    ū-kālaḥ ac hrasva-dīrgha-plutaḥ
    The three lengths u, ū, and ū3 [are called] hrasva (short), dīrgha (long), and pluta (prolated, overlong).

  • अचश्च। १.२.२८
    acaśca (1.2.28)
    acaḥ ca
    And they occur in the context of vowels.


  • उच्चैरुदात्तः। १.२.२९
    uccairudāttaḥ (1.2.29)
    uccaiḥ udāttaḥ
    [In the context of vowels], a high [tone is called] udātta (acute accent);

  • नीचैरनुदात्तः। १.२.३०
    nīcairanudāttaḥ (1.2.30)
    nīcaiḥ anudāttaḥ
    a low [tone is called] anudātta (grave accent);

  • समाहारः स्वरितः। १.२.३१
    samāhāraḥ svaritaḥ (1.2.31)
    samāhāraḥ svaritaḥ
    and a mix [of the two is called] svarita (circumflex),

  • तस्यादित उदात्तमर्धह्रस्वम्। १.२.३२
    tasyādita udāttamardhahrasvam (1.2.32)
    tasya āditaḥ udāttam ardha-hrasvam
    of which the beginning is udātta for half the length of a short [vowel].

And nasality:

  • मुखनासिकावचनो ऽनुनासिकः। १.१.८
    mukhanāsikāvacano 'nunāsikaḥ (1.1.8)
    mukha-nāsikā-vacanaḥ anunāsikaḥ
    An utterance [made with] the mouth and nose is called anunāsika (nasal).

So in the context of grammar, a refers to eighteen variants (three lengths × three accents × two options for nasality). And even at refers to six variants (3 accents × 2 nasality options).


With the Shiva Sutras and the rules above, we now have a powerful framework for referring to different Sanskrit sounds.

But there is still an important open question: what is an it letter, really? The next lesson answers this question and starts to explain the core of the Pāṇinian system.