Chapter 4

Lesson 1—What Is Sandhi?

When Sanskrit is chanted, sounds do not always remain separate from one another. As one sound flows into the next, it may gently change to make the pronunciation smoother and more natural. This joining and adapting of sounds is called sandhi, “joining.”

In this guide, we are mainly concerned with sandhi as a chanting principle, not as a full grammar subject. In the wider grammar, sandhi can even fuse two forms into one—nara (“man”) and indraḥ (“lord”) combine into narendraḥ, “lord of men.”

Common Word-Joining Patterns

The full rules of sandhi are extensive, but the chart below gives only a few common patterns that you may meet in chants, compounds and mantra texts. Some sandhi changes appear in writing; at other times the change is heard only in pronunciation. You do not need to memorize the chart now. Use it to see how sounds may change when words are joined.

Anusvāra and Visarga in This Guide

The examples above show only a small part of sandhi. This chapter focuses on anusvāra and visarga: how this guide writes them, how their pronunciation changes in continuous chanting, and how each is handled at a marked pause. Below is an overview of how these two conditional signs behave; the next two lessons examine each one separately.

In Chapter 6, the syllable-break and tonal lines use brackets to show how a written anusvāra or visarga is pronounced in context.

At Line and Verse Endings

A single daṇḍa (|) marks the end of a chant line, and a double daṇḍa (||) marks the end of a verse. In this guide, both mark a pause in chanting, so the following sound does not affect an anusvāra or visarga before the boundary.

• Anusvāra at a real pause. Anusvāra normally does not occur before such a pause in this guide. A word-final m is written as m, while fixed bīja forms such as oṁ retain their nasal resonance.
• Visarga at a real pause. Visarga is retained as ḥ / ः and pronounced in its pause form, as explained in Lesson 3.
• A typographic line break. Chants are often printed with additional line breaks that are not chant-line or verse endings and therefore do not carry a single or double daṇḍa. Such breaks are matters of page layout. Unless your group customarily pauses there, continue chanting across the break and pronounce any retained anusvāra or visarga according to the next sound. This remains true whether reading from a text or chanting from memory.

Lesson 2—The Anusvāra (ṁ)

Lesson 1 gave the writing convention: where this guide keeps anusvāra and where it writes the matching nasal directly. This lesson now turns to pronunciation. When anusvāra is written, read it as a conditional nasal sound: look ahead to the next sounded consonant and let that sound determine how the nasal is formed. The section “A Closer Look at How Anusvāra Is Pronounced” goes further into the mechanics. You may study it now or return to it later.

The anusvāra (“after sound”) is a special nasal sound in Sanskrit. In Devanāgarī it is written as a dot above the vowel-carrying syllable, whether the vowel stands alone or is combined with a consonant (e.g. गं, अं); in Roman transliteration it is written ṁ (or ṃ in the pure IAST system). It represents a nasal whose pronunciation depends on the consonant that follows: one of the five nasal consonants—ṅ, ñ, ṇ, n or m—before a grouped consonant, or a continuous nasal resonance before semivowels, sibilants and h.

Why Anusvāra Is Used

Anusvāra marks a nasal sound whose exact pronunciation depends on what follows. At the end of a word, final m / म् before a consonant becomes anusvāra. In recitation, before a grouped consonant, it is sounded as the nasal of that same varga; before semivowels, sibilants and h, it becomes a soft nasalized resonance.

In Devanāgarī, the dot shows that a nasal sound is present.

In this guide we write the matching nasal directly before a grouped consonant—saṅkalpa, saṅgha—rather than anusvāra. Anusvāra is retained within a word before y, r, l, v, ś, ṣ, s and h. It is also retained at a word boundary before a following consonant. Other publications may retain anusvāra before grouped consonants as well.

A Closer Look at How Anusvāra Is Pronounced

Anusvāra has two main pronunciations, depending on what follows: one of the five nasal consonants before a grouped consonant, and a softer nasal resonance before semivowels, sibilants and h. Both are anunāsika, or nasal sounds. With the five nasal consonants, the flow of air through the mouth is blocked and the air passes through the nose. With the softer resonance, air passes through both the mouth and nose.

• Before grouped consonants (vargas): The tongue or lips block the flow of air through the mouth at the same place where the following consonant is made. The air then passes through the nose, producing ṅ, ñ, ṇ, n or m.
• Before semivowels, sibilants and h: Air passes through both the mouth and nose, creating a softer nasal resonance rather than one of the five nasal consonants. The tongue or lips may already be moving into place for the following sound. The exact sound can vary by recitation tradition; the descriptions below show how the following sound may color it:
  • Guttural h: The resonance remains open and breathy.
  • Palatal y, ś: The resonance is colored toward the hard palate.
  • Retroflex r, ṣ: The upward tongue position of the following sound gives the resonance a retroflex coloring.
  • Dental l, s: The resonance is colored forward toward the teeth. In saṁskāra, for example, the following dental s gives the preceding resonance a clear, forward, n-like color.
  • Labial v: The resonance is colored toward the lips.

Anusvāra in Chanting

In a chant, apply this look-ahead principle wherever anusvāra is written. Within a word, anusvāra may stand before semivowels, sibilants and h; at a word boundary, any consonant may follow. When the sounds are chanted continuously, the following consonant determines its pronunciation. This blending becomes more natural as the pace grows more fluent. 

In practice, individuals and chanting groups may handle anusvāra differently, often reflecting local custom or the way a chant was learned.

When chanting with others, it is generally best to follow the group’s practice for unity of sound.

Variation may also occur within a single chant when a pause is introduced at a place not marked by a single or double daṇḍa. Such a pause may likewise affect visarga.

When Anusvāra Is Not Used

Anusvāra normally appears when a nasal sound is influenced by a following consonant. When no consonant follows, Sanskrit normally writes m rather than anusvāra. This occurs before a pause, such as a daṇḍa (|) or double daṇḍa (||), and before a following vowel.

In Chapter 5 we will see that the anusvāra also plays a structural role: it closes the syllable it follows, making that syllable automatically heavy (guru).

Lesson 3—The Visarga (ḥ)

What Is the Visarga?

The visarga (विसर्ग, “sending forth” or “release”) is one of Sanskrit’s conditional signs: it is written in one stable way, but its pronunciation changes according to its position and the sound that follows. In Devanāgarī it is written as two stacked dots after a letter (ः), as in अः or गः. In Roman transliteration it is written as ḥ. The visarga is most common at the end of words, as in Rāmaḥ, namaḥ, śāntiḥ and agniḥ. It also occurs within compound words, as in duḥkha, niḥśreyase, prātaḥsmaraṇa and manaḥprayāsa. At a pause, it may be heard as a brief, soft h-like sound after the vowel before it. When chanting continues without a pause, its sound changes according to what follows. The visarga goes with the vowel before it and closes that syllable, making it heavy (guru). The whole syllable therefore counts as two beats.

In this guide, when a visarga appears in the written chant, we keep it as ḥ in Roman transliteration and ः in Devanāgarī. But some chant forms already include sandhi and have no visarga to keep. We leave those forms as they are. For example, we keep bhāvato nityam rather than changing it to bhāvataḥ nityam. When chanting, read the sign by context: at a pause it has its pause form; in a continuing line it takes its sound from what follows. The illustration and tables below show how to do this.

Visarga Inside a Word

Visarga often appears inside a single written word, especially in compounds. Its sound still adapts to the sound that follows. The ḥ is a soft breath before kh, as in duḥkha, and before p, as in manaḥprayāsa. It is a faint sibilant before ś, as in niḥśreyase, and before s, as in prātaḥsmaraṇa.

Visarga at the End of a Chant Line—the Echo Sound

At the end of a chant line, marked by a single daṇḍa (|) or double daṇḍa (||), the visarga is usually pronounced as a soft h-like release after the preceding vowel, often with an echo of that vowel. Thus namaḥ at the end of a chant line sounds like “nama-ha”, and agniḥ like “agni-hi”.

At a pause, some teachers use only the soft h-like release, while others add an echo of the preceding vowel. Both pronunciations are traditional, and practice varies by lineage and teacher. The table gives teaching approximations of the echoed forms; it is not intended as a universal rule for all recitation traditions.

Visarga in a Continuing Line—Adapting to What Follows

These tables show the principal visarga patterns when the chant moves directly into the following sound, with no single or double daṇḍa marking a pause. The following sound determines the main pattern; before a voiced consonant or vowel, the vowel before the visarga also matters. The forms shown after the arrows are not instructions to rewrite the chants in Chapter 6. They make these changes in pronunciation visible so that you can understand the mechanics of visarga. In the examples, the visarga being explained is highlighted in red.

Before an unvoiced consonant

Kṣ is a special case in some chant recitations. When a comma or teaching pause is marked before kṣ, as in mahadbhyaḥ, kṣullakebhyaśca, give the preceding visarga its full pause-like release.

Before a voiced consonant or a vowel

After -aḥ, -āḥ, saḥ or eṣaḥ

After other vowel endings