Sanskrit Study Companion

Study · Reference · Reading

Choose a Subject

Move from the script and sound system to grammar, vocabulary, texts and source-aware reference tools. Choose a subject to begin.

Devanagari title with an arrangement of Sanskrit letterforms

Devanagari · Module 01

Simple Characters

Vowels and consonants arranged by sound. Select any row or column heading in the consonant table to trace its cross-reference.

Independent forms

Vowels

Each vowel can stand independently or appear as a sign attached to a consonant.

अन्नम् अश्नीमः — न्नTiro Devanagari Sanskrit shaping check

Phonetic arrangement

Consonants

Select several row and column headings at once; select a heading again to remove it.

Rows show place of articulation. Columns show voicing, aspiration and nasality.

The selected row and column are shaded; their intersection receives the strongest colour.

Marks and dividers

Common Devanāgarī Signs

These signs modify a sound, show a printed elision, or divide a text. The examples place combining signs on ka so that their position is visible.

Common Devanagari signs, names, transliteration and use
Form	Name	IAST value	What it marks
क्	विरामvirāma · halant	no following vowel	Suppresses a consonant’s inherent a; it can leave a bare final consonant or help form a written conjunct.
कं	अनुस्वारanusvāra	ṃ	Marks a nasal whose phonetic realization and grammatical replacement can depend on the following sound and textual convention.
कँ	चन्द्रबिन्दुcandrabindu	nasalized vowel	Marks vowel nasalization. Its printed height and shape can vary with the surrounding character and font.
कः	विसर्गvisarga	ḥ	Marks a post-vocalic voiceless breath whose surface form is especially sensitive to the following sound under visarga sandhi.
ऽ	अवग्रहavagraha	’	In printed Sanskrit, most commonly records the loss of initial a after final e or o; it is not a general apostrophe.
।	दण्डdaṇḍa	text divider	Divides a prose unit or a verse half according to the editorial and textual convention in use.
॥	द्विदण्डdouble daṇḍa	major text divider	Commonly closes a verse or a larger numbered unit; exact punctuation practice varies by edition and genre.

Descriptions are original project teaching prose checked against Whitney §§5, 9–16a and 70–73a; they are not Whitney’s wording.

Conjunct Tool 01

Find a Conjunct

Use exact IAST, plain Latin letters, or type directly in Devanagari.

On-screen Sanskrit keyboard

Tap characters, then press Search.

Latin / IAST vowels

Other IAST special characters

Devanagari vowels

Devanagari consonants

Examples

क्ष

Enter a conjunct above and press Search. Plain letters return every matching diacritic possibility.

Search results

Conjunct Tool 02

Complete Conjunct Table

First consonant down the side, second consonant across the top. The complete table fits the page and uses the page's normal scroll.

Table size 100%

Click a row or column heading to highlight it; click again to turn it off. The intersection has its own colour. Select a cell to open it in Conjunct Search.

A visible virāma (्) is correct where the font has no joined ligature for that consonant pair.

Devanagari · Module 03

Search the Script

Find an independent vowel, a simple syllable, a conjunct, or a conjunct carrying any vowel.

Examples

अ → क्ष

Search with precise IAST, plain Latin spelling, or Devanagari. Results distinguish vowels, simple syllables and conjunct syllables.

Script search results

Devanagari · Module 04

Word Converter

Transliterate a Sanskrit word in either direction. IAST diacritics give the most precise result.

Try an example:

Plain Latin uses common spellings: aa = ā, sh = ś, and ri = ṛ. Use diacritics when you need to distinguish sounds such as t/ṭ, d/ḍ, n/ṇ, or s/ś/ṣ.

Devanagari · Module 05

Pronunciation Laboratory

Connect script, IAST, IPA and articulation. Generated audio uses a Sanskrit-specific backend when one is connected.

Classical mode presents the conservative phonemic system used in Sanskrit teaching. It does not apply modern Hindi schwa deletion.

Select a sound

Phoneme chart

Tap a character to see its IPA value and how it is articulated.

Vowels

Consonants

शब्द

Words and short phrases

IAST or Devanāgarī

Checking the curated audio pack and pronunciation backend…

Licensed recordings play first and work offline. New or arbitrary text uses the connected Sanskrit synthesizer.

मात्रा

Mora timing

Short vowels normally count as one mora; long vowels and diphthongs count as two. Closed syllables are heavy.

Word or phrase in IAST

The pulses demonstrate duration only; they are not speech.

Tradition labels matter: Vedic recitation adds pitch accents and tradition-specific rules; modern regional readings may vary. The Classical chart is the default reference, while Vedic and regional material is explicitly labelled.

Sandhi title with two streams of Sanskrit sounds joining

Internal Sandhi · Vowels

Vowels Inside a Word

Use these charts where a root, stem, suffix or compound member meets the next element. They show the sound boundary, not a claim about how an unfamiliar word must be segmented.

Coverage: these charts make the five basic vowel-sandhi families explicit and add common boundary exceptions. They are not a claim to exhaust every optional, Vedic or morphology-specific operation in the Aṣṭādhyāyī.

Complete internal vowel sandhi: simple first vowel

Use this within a word—for example, where a root meets a suffix or two members of a compound meet.

Internal vowel sandhi when the first vowel is simple
First ↓ / second →	a	ā	i	ī	u	ū	ṛ	ṝ	ḷ	e	ai	o	au
a / ā	ā	ā	e	e	o	o	ar	ar	al	ai	ai	au	au
i / ī	ya	yā	ī	ī	yu	yū	yṛ	yṝ	yḷ	ye	yai	yo	yau
u / ū	va	vā	vi	vī	ū	ū	vṛ	vṝ	vḷ	ve	vai	vo	vau
ṛ / ṝ	ra	rā	ri	rī	ru	rū	ṝ	ṝ	rḷ	re	rai	ro	rau
ḷ	la	lā	li	lī	lu	lū	lṛ	lṝ	ḹ†	le	lai	lo	lau

† Long vocalic l (ḹ) is theoretical and does not ordinarily occur. In a few internal morphological environments, ī/ū + vowel may appear as iy/uv instead of y/v.

Complete internal vowel sandhi: complex first vowel

Internally, e, ai, o and au recover their glide element before another vowel.

Internal vowel sandhi when the first vowel is complex
First ↓ / second →	a	ā	i	ī	u	ū	ṛ	ṝ	ḷ	e	ai	o	au
e	aya	ayā	ayi	ayī	ayu	ayū	ayṛ	ayṝ	ayḷ	aye	ayai	ayo	ayau
ai	āya	āyā	āyi	āyī	āyu	āyū	āyṛ	āyṝ	āyḷ	āye	āyai	āyo	āyau
o	ava	avā	avi	avī	avu	avū	avṛ	avṝ	avḷ	ave	avai	avo	avau
au	āva	āvā	āvi	āvī	āvu	āvū	āvṛ	āvṝ	āvḷ	āve	āvai	āvo	āvau

सवर्णदीर्घसन्धि · Savarṇa-dīrgha

Pāṇini 6.1.101 · अकः सवर्णे दीर्घः · akaḥ savarṇe dīrghaḥ

Same-quality vowels become long

Two vowels of the same quality merge into the corresponding long vowel.

Similar vowel combinations
Combination	Result
a/ā + a/ā	ā
i/ī + i/ī	ī
u/ū + u/ū	ū
ṛ/ṝ + ṛ/ṝ	ṝ
ḷ + ḷ	ḹ†

गुणसन्धि · Guṇa

Pāṇini 6.1.87 · आद् गुणः · ād guṇaḥ

a or ā before i, u, ṛ or ḷ

The two boundary vowels are replaced by the corresponding guṇa result.

Guṇa vowel combinations
Combination	Result
a/ā + i/ī	e
a/ā + u/ū	o
a/ā + ṛ/ṝ	ar
a/ā + ḷ	al

A preverb ending in a/ā before a root beginning in short ṛ may exceptionally give ār.

वृद्धिसन्धि · Vṛddhi

Pāṇini 6.1.88 · वृद्धिर् एचि · vṛddhir eci

a or ā before e, ai, o or au

The two boundary vowels are replaced by the corresponding vṛddhi result.

Vṛddhi vowel combinations
Combination	Result
a/ā + e/ai	ai
a/ā + o/au	au

यण्सन्धि · Yaṇ

Pāṇini 6.1.77 · इको यण् अचि · iko yaṇ aci

i, u, ṛ or ḷ before a different vowel

The first vowel becomes its corresponding glide.

Vowel to glide changes
First vowel + V	Result	Example
i/ī + V	y + V	i + a → ya
u/ū + V	v + V	u + a → va
ṛ/ṝ + V	r + V	ṛ + a → ra
ḷ + V	l + V	ḷ + a → la

अयादिसन्धि · Ayādi

Pāṇini 6.1.78 · एचोऽयवायावः · eco ’yavāyāvaḥ

e, ai, o or au before a vowel

The first vowel opens into an ay-, āy-, av- or āv- sequence before the next vowel.

Complex vowels before another vowel
First vowel + V	Result	Example before a
e + V	ay + V	e + a → aya
ai + V	āy + V	ai + a → āya
o + V	av + V	o + a → ava
au + V	āv + V	au + a → āva

Context and exceptions · Whitney §§127–138

The chart gives a default—not one result for every context

The familiar contrast a + i → e versus augment a- + root-initial i → ai is one example of a wider principle: first identify the order of the sounds, the kind of boundary and the grammatical role of each element. A narrower rule does not create a freely interchangeable second answer; it applies only in its stated environment.

Ordinary vowel contacta/ā + i/ī → eWhitney §127

Augment before a roota- + root-initial i → aiWhitney §136a · a + ita → āita in Whitney’s notation

Order matters: a + i → e, but i + a → ya. These are different ordered boundaries, not exceptions to one another.
Internal and external rules can differ: internal e + a → aya, while external final e before initial a retains e and loses the following a, shown by avagraha: e + a → e ’ (§§131–135).
Specific grammatical roles can select a narrower rule: augment a- before root-initial i/u/ṛ gives ai/au/ār (§136a); externally, a preverb ending in a/ā before a root beginning in ṛ can give ār instead of ordinary ar (§137a).
Some vowels are retained: vowels classed as pragṛhya remain uncombined; Whitney includes dual endings in ī, ū and e (§138).
Period and rule ordering matter: Vedic usage has additional exceptions and optional patterns. Consonant rules also have exact environments, blockers and later operations, so no chart row should be applied as an unrestricted letter substitution.

Internal Sandhi · Consonants

Consonants in Close Contact

Consonant changes depend on the exact root, stem, suffix, word boundary and intervening sounds. Begin with the immediate contact, then test the wider formation for triggers, blockers and later rules.

Bounded lesson: this page teaches the two internal consonant families for which the site currently has complete checked sources and examples. Read the table from grammatical boundary to contact test; the named families at the bottom are signposts, not rules to apply from this page.

Complete inventory for this lesson

Two verified paths, plus one immediate blocker

First locate the grammatical boundary. Then inspect the sounds that meet. The longer cards below supply the derivation, limitations, translations, and exact source links.

Complete inventory of the source-checked internal consonant rules taught on this page
Boundary	Contact test	Verified rule	Checked result	English
Root + ending	Root-final gh, dh, or bh before ending-initial t or th	Bartholomae’s law	√budh + ta → buddha	awakened; understood
Immediate contact	s/dental class beside ś/palatal class, in either order	ścutva	√yaj + na → yajña	sacrifice
Immediate blocker	When ś precedes dental n, rule 8.4.44 blocks the dental-to-palatal change	ścutva limitation	praś + na → praśna	question

Whitney §§112 and 118 distinguish internal from external combination. “Complete” here means complete for the declared two-family lesson, not a claim to exhaust Sanskrit consonant sandhi.

Bartholomae’s law · often nicknamed “Buddha sandhi”

Aspiration moves to an ending beginning in t or th

This source-verified Sanskrit environment is internal and morphological: a root ending in the voiced aspirate gh, dh or bh is followed by t or th at the start of an ending. It is not a blanket rule for any aspirated stop before any voiceless consonant.

Exact environmentroot-final gh/dh/bh + ending-initial t/thWhitney §160

Structural outcomegh + t/th → gdh · dh + t/th → ddh · bh + t/th → bdhThe cluster is voiced; aspiration appears on dh.

1 · Following stopt/th → dhPāṇini 8.2.40 supplies the first step after a voiced aspirate.

2 · Preceding stopgh/dh/bh → g/d/bPāṇini 8.4.53 supplies the corresponding unaspirated voiced stop before the new dh.

Worked Bartholomae’s-law examples from Whitney section 160
Formation	After 8.2.40	After 8.4.53	English
√budh + ta	budh + dha	buddha	awakened; understood
√labh + ta	labh + dha	labdha	obtained; gained
√labh + tvā	labh + dhvā	labdhvā	having obtained

श्चुत्व · ścutva · source-checked contact rule

A dental shifts to the matching palatal in close contact

Pāṇini 8.4.40 covers s and the dental tu class in contact with ś or a member of the palatal cu class. The dental-side sound is replaced by ś or the corresponding palatal. The palatal trigger may stand before or after it; Whitney §118c independently summarizes the contiguous-palatal change.

Target and triggers / dental tu ↔ ś / palatal cuClose contact in either order; Pāṇini 8.4.40.

Corresponding replacementss → ś · t/th/d/dh/n → c/ch/j/jh/ñThe replacement preserves the corresponding class position.

Source-attested ścutva application and immediate limitation
Source formation	Contact test	Outcome	Supported meaning
√yaj + na	j + n → j + ñPalatal j precedes dental n; 8.4.40 changes n to corresponding palatal ñ.	yajña	sacrificeExact Vasu example; meaning checked at MW L=169220–169221.
praś + na	ś + n ↛ ś + ñPāṇini 8.4.44 blocks the dental-to-palatal change when ś precedes.	praśna	questionExact Vasu 8.4.44 result; meaning checked at MW L=137599.

Named for orientation only

Families outside this bounded lesson

These names help a learner recognize what still requires a future source-complete treatment. No example or automatic substitution is supplied here where the necessary conditions and limitations have not yet been verified together.

Internal consonant families deliberately excluded from the current teaching scope
Family	Current status	Why it is not an active rule here
ṣṭutva	Pāṇini 8.4.41 located; teaching row withheld	The worked examples and the immediately following limitations in 8.4.42–43 still require source-complete verification.
ṣatva	Future lesson	Its trigger may occur earlier in the word, and intervening sounds, lexical restrictions, and blockers must be checked together.
ṇatva	Future lesson	It is likewise non-local in many formations; a two-letter replacement row would conceal its domain and blockers.

External Sandhi · Vowels

Vowels at a Word Boundary

Compare the final vowel of the first word with the initial vowel of the next, then follow that row and column through the complete adopted lookup grid.

Open a rule: select a result cell to highlight its final-vowel column and following-vowel row, then see the rule name and a worked example. A space in the chart marks retained hiatus at a word boundary.

Sandhi Rules: Vowels

A space indicates a word break.

’ marks the loss of a following initial a at the word boundary; Devanāgarī shows this with avagraha ऽ.

The Little Red Book of Sanskrit Paradigms, Table 61: Sandhi rules for vowels
Final vowels								Initial
-a/-ā	-i/-ī	-u/-ū	-ṛ	-e	-ai	-o	-au	Initial
-ā-	-ya-	-va-	-ra-	-e ’-	-ā a-	-o ’-	-āva-	a-
-ā-	-yā-	-vā-	-rā-	-a ā-	-ā ā-	-a ā-	-āvā-	ā-
-e-	-ī-	-vi-	-ri-	-a i-	-ā i-	-a i-	-āvi-	i-
-e-	-ī-	-vī-	-rī-	-a ī-	-ā ī-	-a ī-	-āvī-	ī-
-o-	-yu-	-ū-	-ru-	-a u-	-ā u-	-a u-	-āvu-	u-
-o-	-yū-	-ū-	-rū-	-a ū-	-ā ū-	-a ū-	-āvū-	ū-
-ar-	-yṛ-	-vṛ-	-ṝ-	-a ṛ-	-ā ṛ-	-a ṛ-	-āvṛ-	ṛ-
-ai-	-ye-	-ve-	-re-	-a e-	-ā e-	-a e-	-āve-	e-
-ai-	-yai-	-vai-	-rai-	-a ai-	-ā ai-	-a ai-	-āvai-	ai-
-au-	-yo-	-vo-	-ro-	-a o-	-ā o-	-a o-	-āvo-	o-
-au-	-yau-	-vau-	-rau-	-a au-	-ā au-	-a au-	-āvau-	au-

External Sandhi · Consonants

Consonants at a Word Boundary

Start with the final sound of the first word, then identify the initial sound that follows. Work through stops, visarga and final m before consulting the complete source grid.

Highlight a path: select any body cell to highlight its logical row and column. A final stop normally follows the next sound’s voicing and may also change place; final s/r appears as visarga in pause and changes again before another sound.

1

Stops

Match voicing, then assimilate strongly before a stop, nasal or l.

2

Visarga

Choose the rule from the preceding vowel and the following sound.

3

Nasals

Final m becomes anusvāra before a consonant; internal nasals match the next stop.

4

Sibilants

Dental sounds become palatal or retroflex beside matching sounds.

Final stops: voicing and place assimilation

C represents any stop. The most useful first question is whether the following sound is voiced or voiceless.

Rules for final Sanskrit stops
Final sound	Following sound	Result	Worked example
voiceless stop	voiced sound	corresponding voiced stop	tat + bhavati→ tad bhavatiतत् + भवति → तद् भवति
voiced stop	voiceless sound	corresponding voiceless stop	sad + kṛ→ satkṛसद् + कृ → सत्कृ
t / d	c / ch	c	tat + ca→ taccaतत् + च → तच्च
t / d	j / jh	j	sat + jana→ sajjanaसत् + जन → सज्जन
t / d	l	l	tat + loka→ tallokaतत् + लोक → तल्लोक
t / d	n / m	matching nasal	tat + nāma→ tan nāmaतत् + नाम → तन्नाम
k / g	vowel or voiced sound	g	vāk + īśa→ vāgīśaवाक् + ईश → वागीश

Final jaśtva · project learner heading

Keep the corresponding voiced stop before a voiced sound

Pāṇini 8.2.39 first replaces an eligible pada-final jhal stop with the same-place member of the technical set jaś. In the external environment described by Whitney, that voiced result remains before a following vowel or other voiced sound. “Final jaśtva” is a convenient modern grammatical label used by this project; it is not presented as wording by Pāṇini, Vasu or Whitney.

Pāṇinian operationpada-final jhal stop → corresponding jaś stopk → g · ṭ → ḍ · t → d · p → b

Operational environmentThe next sound is voiced, including a vowelWhitney §§157c and 159

Worked final-jaśtva examples recorded by Whitney section 159
Formation	After 8.2.39	Combined form	English
ap + ja	ab + ja	abja	water-born; a lotus
dik + gaja	dig + gaja	diggaja	elephant of a direction
ṣaṭ + aha	ṣaḍ + aha	ṣaḍaha	a period of six days

Visarga after a or ā

These are the high-frequency external rules. The apostrophe-like avagraha ’ marks the loss of an initial a.

Visarga rules after a and long a
Ending	Next sound	Result	Example
aḥ	a	o ’	rāmaḥ + atra→ rāmo ’traरामः + अत्र → रामोऽत्र
aḥ	other vowel	a + vowel	rāmaḥ + iha→ rāma ihaरामः + इह → राम इह
aḥ	voiced consonant	o	rāmaḥ + gacchati→ rāmo gacchatiरामः + गच्छति → रामो गच्छति
āḥ	vowel or voiced consonant	ā	narāḥ + gacchanti→ narā gacchantiनराः + गच्छन्ति → नरा गच्छन्ति
aḥ / āḥ	k kh p ph	ḥ remains	rāmaḥ + paṭhati→ rāmaḥ paṭhatiरामः पठति

Visarga after i, u, ṛ—and before sibilants

Other visarga transformations
Environment	Result	Example	Devanāgarī
iḥ / īḥ + voiced sound	ir / īr	muniḥ + asti → munir asti	मुनिरस्ति
uḥ / ūḥ + voiced sound	ur / ūr	guruḥ + vadati → gurur vadati	गुरुर्वदति
ḥ + c / ch	ś	rāmaḥ + ca → rāmaś ca	रामश्च
ḥ + ṭ / ṭh	ṣ	dhanuḥ + ṭaṅkāraḥ → dhanuṣṭaṅkāraḥ	धनुष्टङ्कारः
ḥ + t / th	s	namaḥ + te → namas te	नमस्ते
ḥ + ś / ṣ / s	ḥ or matching sibilant	hariḥ śete → hariḥ śete / hariś śete	हरिः शेते

Final m and anusvāra

Environment	Rule	Example
m + vowel	m remains	phalam asti → phalam asti
m + consonant	m → ṃ	phalam khādati → phalaṃ khādati
ṃ + k/kh/g/gh	ṅ may be written	saṃ + kalpa → saṅkalpa
ṃ + c/ch/j/jh	ñ may be written	sam + cara → sañcara
ṃ + ṭ/ṭh/ḍ/ḍh	ṇ may be written	aṃ + ṭa → aṇṭa (homorganic pronunciation)
ṃ + t/th/d/dh	n may be written	sam + toṣa → santoṣa
ṃ + p/ph/b/bh	m may be written	sam + pada → sampada

Source-based reference chart

The Little Red Book Consonant Chart

Use the final sound across the top and the following initial sound at the right. Select a row and column to trace one boundary; the continuous chart retains all nineteen source rows.

Sandhi Rules: Consonants

After Coulson; a hyphen indicates a join. Taylor’s chart spans pp.62–63, with the four numbered notes printed on p.63 retained below.

The Little Red Book of Sanskrit Paradigms, pages 62–63: one consonant sandhi chart, after Coulson, with page-63 footnotes
Final letter										Initial
-k	-ṭ	-t	-p	-ṇ	-n	-m	-ḥ/-r	-āḥ	-aḥ	Initial
-k-	-ṭ-	-t-	-p-	-ṇ-	-n-	-ṃ	-ḥ	-āḥ	-aḥ	k-/kh-
-g-	-ḍ-	-d-	-b-	-ṇ-	-n-	-ṃ	-r-	-ā	-o	g-/gh-
-k-	-ṭ-	-c-	-p-	-ṇ-	-ṃś-	-ṃ	-ś-	-āś-	-aś-	c-/ch-
-g-	-ḍ-	-j-	-b-	-ṇ-	-ñ-	-ṃ	-r-	-ā	-o	j-/jh-
-k-	-ṭ-	-ṭ-	-p-	-ṇ-	-ṃṣ-	-ṃ	-ṣ-	-āṣ-	-aṣ-	ṭ-/ṭh-
-g-	-ḍ-	-ḍ-	-b-	-ṇ-	-ṇ-	-ṃ	-r-	-ā	-o	ḍ-/ḍh-
-k-	-ṭ-	-t-	-p-	-ṇ-	-ṃs-	-ṃ	-s-	-ās-	-as-	t-/th-
-g-	-ḍ-	-d-	-b-	-ṇ-	-n-	-ṃ	-r-	-ā	-o	d-/dh-
-k-	-ṭ-	-t-	-p-	-ṇ-	-n-	-ṃ	-ḥ	-āḥ	-aḥ	p-/ph-
-g-	-ḍ-	-d-	-b-	-ṇ-	-n-	-ṃ	-r-	-ā	-o	b-/bh-
-ṅ-	-ṇ-	-n-	-m-	-ṇ-	-n-	-ṃ	-r-	-ā	-o	n-/m-
-g-	-ḍ-	-d-	-b-	-ṇ-	-n-	-ṃ	-r-	-ā	-o	y-/v-
-g-	-ḍ-	-d-	-b-	-ṇ-	-n-	-ṃ	zero¹	-ā	-o	r-
-g-	-ḍ-	-l-	-b-	-ṇ-	-ṃl-²	-ṃ	-r-	-ā	-o	l-
-k-	-ṭ-	-c(ch)-	-p-	-ṇ-	-ñ(ś/ch)-	-ṃ	-ḥ	-āḥ	-aḥ	ś-
-k-	-ṭ-	-t-	-p-	-ṇ-	-n-	-ṃ	-ḥ	-āḥ	-aḥ	ṣ-/s-
-g(gh)-	-ḍ(ḍh)-	-d(dh)-	-b(bh)-	-ṇ-	-n-	-ṃ	-r-	-ā	-o	h-
-g-	-ḍ-	-d-	-b-	-ṇ/ṇṇ-³	-n/nn-³	-m-	-r-	-ā	-a⁴	vowels
-k	-ṭ	-t	-p	-ṇ	-n	-m	-ḥ	-āḥ	-aḥ	zero

ḥ or r disappears, and if a, i or u precedes, this lengthens to ā, ī or ū.
For example: tān + labhasva = tāṃl labhasva.
Doubling occurs when the preceding vowel is short.
Except that aḥ + a = o ’/ऽ (avagraha).

A reliable order for solving consonant sandhi

Restore the boundary: identify the likely end of word one and start of word two.
Check m/ḥ first: anusvāra and visarga have their own rule families.
For a stop, match voicing: voiced before voiced, voiceless before voiceless.
Then check place: palatal, retroflex, dental, nasal or l can trigger fuller assimilation.
Finally test internal rules: ṣatva, ṇatva and inherited lexical forms.

Sandhi · Joining Words

Which Words Join in Devanāgarī?

Sound contact and grammatical wordhood are different questions. First decide whether the pieces make one inflected word or lexical compound; only then apply the internal or external sandhi rule licensed at that boundary.

Wordhood before spelling

One written word or a word boundary?

Use the grammatical construction to locate the boundary; use the linked sandhi chart to work out its visible form.

Write one verbal wordव्यगच्छन्vyagacchan

vi- (preverb) + a- (past augment) + gaccha- (present stem) + -n (third-plural secondary ending) are internal pieces of one imperfect form. See the preverb overview.

Keep a grammatical boundaryनरा गच्छन्तिnarā gacchanti

नराः is the subject and गच्छन्ति is the verb. External visarga sandhi changes narāḥ to narā before voiced g; it does not make a noun–verb compound. See external consonant sandhi.

Join the pieces of one inflected word

Preverbs, augments, stems, suffixes and endings that belong to one inflected form are written together after internal changes: वि + अ + गच्छ + न् → व्यगच्छन्. Check internal vowel sandhi.

Join a lexical compound

Members that form one compound are written as one compound word; the final inflection belongs to the whole. This differs from two independent words merely touching in a sentence. See compounds.

Apply external sandhi between independent words

Subjects, objects, modifiers and verbs remain distinct grammatical words, but their boundary still takes external sandhi. In Devanāgarī, use the normal sandhied spelling; in IAST, retain an analytical space where it clarifies units such as the conjunction ca. Check external vowel and consonant sandhi.

Read continuous manuscripts cautiously

Manuscripts and editions differ in how visibly they mark word division. A continuous written string is not proof that every element is one grammatical word. Whitney §9 describes the editorial distinction.

Do vowel-final words stay separate?

Grammatically, yes; in the displayed sounds, follow the applicable sandhi. Before another vowel, external a + i can coalesce as e; i + a can show the glide ya; and final e or o before initial a can retain the first vowel while the lost a is marked by avagraha. Devanāgarī and IAST should show that surface result, while explanation or analytical spacing can identify the recovered word boundary. Use the external vowel grid, the internal vowel rules, and the explanations for the exact environment.

What if the next word begins with a consonant? A genuinely vowel-final word normally keeps its final vowel because there is no vowel–vowel contact, so an ordinary word space is reasonable: नरा गच्छन्ति, narā gacchanti. If the first word actually ends in visarga or a consonant, external consonant sandhi may join its written consonant to the following Devanāgarī syllable; IAST can still mark the analytical word division. Check external consonant sandhi.

Worked imperfect

“The people dispersed”

Build the subjectjana + jas → janāḥजन + जस् → जनाः
The noun is nominative plural: “the people.”
Build the imperfect verbvi + a + gaccha + nवि + अ + गच्छ + न्
vi- supplies the dispersive meaning, a- is the past augment, gaccha- is the present stem, and -n is the third-plural secondary ending.
Apply internal sandhivi + a → vya; vyagacchanवि + अ → व्य; व्यगच्छन्
Final i of the preverb becomes y before the vowel-initial augment. The separate change √gam → gaccha- belongs to present-stem formation, not this sandhi.
Apply external sandhijanāḥ + vyagacchan → janā vyagacchanजनाः + व्यगच्छन् → जना व्यगच्छन्
The visarga is lost before voiced v, but the subject and verb remain two grammatical words.

जना व्यगच्छन्janā vyagacchanThe people dispersed.

Tools · Combine

Put a Boundary Together

Enter IAST or Devanāgarī. A space fixes an external word boundary. A + marks the exact boundary without forcing it to be internal; use the scope menu to test internal rules, external rules, or both.

No combination yet

The tool changes one boundary at a time and reports only rules represented by the adopted charts or the explicitly listed internal-contact rules.

Tools · Resolve

Test Possible Underlying Boundaries

Enter a sandhied form in IAST or Devanāgarī. Resolution is inherently many-to-one, so the tool returns possible source pairs rather than one falsely certain answer.

No resolution yet

Because several inputs can produce the same surface, candidates remain labelled possibilities until vocabulary and sentence context confirm one.

Overview · Major Rule Map

Three Questions Organize Sandhi

Internal and external tell you where a boundary occurs. Vowel, consonant and visarga tell you what meets there. Guṇa, yaṇ, voicing, ścutva and the other family names tell you what kind of change happens.

1 · Where?

Boundary position

Internal: inside a word√budh + ta → buddha
बुध् + त → बुद्ध External: between wordstat + ca → tacca
तत् + च → तच्च

2 · What meets?

Sound group

Vowel + vowela + i → e · अ + इ → ए
Consonant contact√budh + ta → buddha · बुध् + त → बुद्ध
Final m / anusvāra (ṃ)phalam + khādati → phalaṃ khādati
फलम् + खादति → फलं खादति
Visarga (ḥ)rāmaḥ + atra → rāmo ’tra
रामः + अत्र → रामोऽत्र

3 · What happens?

Rule family

Coalescencea + a → ā
Glide formationi + a → ya
Strengtheninga + i → e
Voicingap + ja → abja
Place assimilationyaj + na → yajña
Nasalizationphalam + khādati → phalaṃ khādati
Retentionrāmaḥ + paṭhati → rāmaḥ paṭhati
Lossnarāḥ + gacchanti → narā gacchanti

स्वरसन्धि · svara-sandhi

1 · Vowels merge or become glides

सवर्णदीर्घ सन्धिsavarṇa-dīrgha sandhihomogeneous-vowel lengthening: Two vowels of the same quality become the corresponding long vowel.a + a → āअ + अ → आOpen the internal-vowel lesson →
गुण सन्धिguṇa sandhiguṇa replacement: a/ā before i/ī, u/ū, ṛ/ṝ or ḷ gives the corresponding guṇa result.a + i → eअ + इ → एOpen the internal-vowel lesson →
वृद्धि सन्धिvṛddhi sandhivṛddhi replacement: a/ā before e/ai or o/au gives ai or au.a + e → aiअ + ए → ऐOpen the internal-vowel lesson →
यण् सन्धिyaṇ sandhisimple-vowel glide formation: A non-a simple vowel becomes its corresponding glide before a different vowel.i + a → yaइ + अ → यOpen the internal-vowel lesson →
अयादि सन्धिayādi sandhicomplex-vowel opening: e, ai, o and au show their y or v element before another vowel.e + a → ayaए + अ → अयOpen the internal-vowel lesson →
पूर्वरूप सन्धिpūrvarūpa sandhifirst-vowel retention with avagraha: At the stated external boundary, final e/o remains before initial a, whose loss is marked by avagraha.rāme + atra → rāme ’traरामे + अत्र → रामेऽत्रOpen the external-vowel chart →

व्यञ्जनसन्धि · vyañjana-sandhi

2 · Consonants adjust in contact

जश्त्वjaśtvafinal-stop voicing: An eligible word-final stop takes its corresponding voiced stop before a voiced sound; later rules can alter that intermediate form.tat + bhavati → tad bhavatiतत् + भवति → तद् भवति“That happens.”Open the external-consonant chart →
बुद्ध-सन्धिBartholomae’s lawroot-final aspiration transfer; “buddha-sandhi” is a teaching nickname: Root-final gh/dh/bh before ending-initial t/th produces a voiced cluster whose second member carries the aspiration.√budh + ta → buddhaबुध् + त → बुद्ध“awakened”Open the source-checked internal lesson →
अनुस्वार सन्धिanusvāra sandhifinal-m nasal notation: Before a following consonant, final m can be written as anusvāra under the charted external rule.phalam + khādati → phalaṃ khādatiफलम् + खादति → फलं खादति“He or she eats fruit.”Open the external-consonant chart →
श्चुत्वścutvadental-to-palatal assimilation: A dental-class sound in immediate contact with a palatal-class sound takes the corresponding palatal form, subject to the stated direction and blocker.yaj + na → yajñaयज् + न → यज्ञ“sacrifice”Open the source-checked internal lesson →

विसर्गसन्धि · visarga-sandhi

3 · Visarga responds to the next sound

विसर्गस्य शकारःvisarga → śpalatal-sibilant assimilation: Before the stated palatal voiceless sound, visarga appears as the matching sibilant.rāmaḥ + ca → rāmaś caरामः + च → रामश्च“and Rāma”Open the visarga chart →
विसर्गस्य रेफःvisarga → rr-like replacement: In the stated voiced or vocalic environment after a non-a/ā vowel, visarga is represented by r.muniḥ + asti → munir astiमुनिः + अस्ति → मुनिरस्ति“The sage is.”Open the visarga chart →
अः → ओaḥ → oo-replacement before a voiced sound: In the charted environment, final aḥ becomes o; before initial a, avagraha marks that vowel’s loss.rāmaḥ + atra → rāmo ’traरामः + अत्र → रामोऽत्र“Rāma is here.”Open the visarga chart →
आः → आāḥ → āvisarga loss in the stated voiced environment: After ā, visarga can disappear before the voiced or vocalic environments listed in the chart while the word boundary remains.narāḥ + gacchanti → narā gacchantiनराः + गच्छन्ति → नरा गच्छन्ति“The men go.”Open the visarga chart →

Overview · Explain

How to Work Through Sandhi

Begin with the boundary, decide whether it is inside one word or between words, then decide whether vowels or consonants meet there. The lessons below explain each current teaching-rule row in simple English and reuse its existing Sanskrit example.

Internal · Vowels

Start with the sounds, boundary and grammatical job

Read the ordered vowels, then identify the morphemes and boundary type. The familiar a + i → e versus augment a- + i-root → ai contrast is only one context-sensitive branch. Internal and external rules can differ; preverbs and augments can trigger narrower rules; pragṛhya vowels can remain separate; and Vedic usage can preserve additional exceptions. These conditions select a rule—they do not make every listed result an interchangeable option.

Internal · Consonants

Distinguish a root-ending rule from immediate palatal contact

The current bounded lesson teaches Bartholomae’s root + ending pattern and ścutva immediate contact, including the 8.4.44 praśna blocker. Ṣṭutva, ṣatva and ṇatva are future-topic signposts only; no worked rule is implied here.

External · Vowels

Keep the word boundary visible

Read the final vowel of the first word across the source chart and the initial vowel of the next word down its side. Simple vowels use the same core families; the focused chart then shows what changes when the first word ends in e, ai, o or au.

External · Consonants

Check the final sound first

Work through the existing rows in this order: final stops, visarga after a/ā, other visarga environments, then final m and homorganic nasals. Each explanation below is built from the current chart row and keeps its existing example.

Verbs title with Sanskrit root and present-tense imagery

Verbs · Module 01

Verb Roots

How the root vowel appears in the zero, guṇa and vṛddhi grades, with consonant and vowel environments shown separately.

Click any cell to see examples

Forming guṇa and vṛddhi from zero grade (inserting a/ā BEFORE root vowel)

V means that a vowel follows the graded element; C means that a consonant follows it. Each short and long zero-grade vowel has its own row.

Formation of guna and vrddhi from each zero-grade vowel before a following vowel or consonant
Zero grade	Guṇa (add -a-)		Vṛddhi (add -ā-)
Zero grade	Before V	Before C	Before V	Before C
∅ a-series	a_v + Vguṇa supplies a; apply the relevant vowel rule	a_v + C∅ + a + C → aC	ā_v + Vvṛddhi supplies ā; apply the relevant vowel rule	ā_v + C∅ + ā + C → āC
a + i	ayVa + i + V → ayV	eCa + i + C → eC	āyVā + i + V → āyV	aiCā + i + C → aiC
a + ī	ayVa + ī + V → ayV	eCa + ī + C → eC	āyVā + ī + V → āyV	aiCā + ī + C → aiC
a + u	avVa + u + V → avV	oCa + u + C → oC	āvVā + u + V → āvV	auCā + u + C → auC
a + ū	avVa + ū + V → avV	oCa + ū + C → oC	āvVā + ū + V → āvV	auCā + ū + C → auC
a + ṛ	arVa + ṛ + V → arV	arCa + ṛ + C → arC	ārVā + ṛ + V → ārV	ārCā + ṛ + C → ārC
a + ṝ	arVa + ṝ + V → arV	arCa + ṝ + C → arC	ārVā + ṝ + V → ārV	ārCā + ṝ + C → ārC
a + ḷ	alVa + ḷ + V → alV	alCa + ḷ + C → alC	ālVā + ḷ + V → ālV	ālCā + ḷ + C → ālC
a + ḹ†	alV†a + ḹ + V → alV	alC†a + ḹ + C → alC	ālV†ā + ḹ + V → ālV	ālC†ā + ḹ + C → ālC

† Long vocalic ḹ is theoretical and does not provide a normal attested root series. The regular table lists zero-grade absence ∅ rather than a: in the a-series, guṇa supplies a and vṛddhi supplies ā.

Read a + i + V → ayV as a grading formula: guṇa adds a before zero-grade i, which becomes y before the following vowel. Before a consonant, the same guṇa sequence appears as e: a + i + C → eC.Vṛddhi follows the parallel pattern with ā.

Forming guṇa and vṛddhi from zero grade (inserting a/ā AFTER root vowel)

In this exceptional pattern, the added a or ā follows the root vowel. The root vowel therefore becomes its corresponding glide before that added vowel.

Exceptional formation of guna and vrddhi by inserting a or long a after each zero-grade root vowel
Zero grade	Guṇa (add -a-)		Vṛddhi (add -ā-)
Zero grade	Before V	Before C	Before V	Before C
i + a	yaVi + a + V → yaV	yaCi + a + C → yaC	yāVi + ā + V → yāV	yāCi + ā + C → yāC
ī + a	yaVī + a + V → yaV	yaCī + a + C → yaC	yāVī + ā + V → yāV	yāCī + ā + C → yāC
u + a	vaVu + a + V → vaV	vaCu + a + C → vaC	vāVu + ā + V → vāV	vāCu + ā + C → vāC
ū + a	vaVū + a + V → vaV	vaCū + a + C → vaC	vāVū + ā + V → vāV	vāCū + ā + C → vāC
ṛ + a	raVṛ + a + V → raV	raCṛ + a + C → raC	rāVṛ + ā + V → rāV	rāCṛ + ā + C → rāC
ṝ + a	raVṝ + a + V → raV	raCṝ + a + C → raC	rāVṝ + ā + V → rāV	rāCṝ + ā + C → rāC
ḷ + a	laVḷ + a + V → laV	laCḷ + a + C → laC	lāVḷ + ā + V → lāV	lāCḷ + ā + C → lāC
ḹ + a†	laV†ḹ + a + V → laV	laC†ḹ + a + C → laC	lāV†ḹ + ā + V → lāV	lāC†ḹ + ā + C → lāC

This is the pattern behind exceptional series such as ij → yaj → yāj, uc → vac → vāc, sup → svap → svāp, and ṛkṣ → rakṣ → rākṣ. These roots are conventionally cited in guṇa because the position of the inserted vowel cannot be predicted from the zero grade alone.

Vocabulary examples: a/ā inserted BEFORE the root vowel

Each row applies the general vowel-grade model to a dictionary citation form. These are transparent workings, not claims that every displayed weak or strong form is lexically attested.

Vocabulary examples: a/ā inserted AFTER the root vowel

These exceptional roots are cited in guṇa because the position of the inserted vowel must be learned with the root.

Other lexical and irregular root-grade series

These roots complete the 501-root reference, but their cited forms cannot safely be derived by either simple insertion table. Learn them as lexical series and verify inflected forms in the dictionary.

Source, calibration and limits

The 501-entry reference retains 221 legacy course records, 51 earlier curated DCS additions and 229 roots selected reproducibly from the pinned kosha data-v0.1.0 release. The Core 1,000 review adds √rabh, consolidating prefixed ārabh and related occurrences under the exact MW root rather than treating them as separate roots. Every entry links internally to an exact Cologne Monier–Williams record; four citation conventions use a declared normalized lookup. Six denominative, derived or prefixed entries have no root-grade series, and seven MW records state no present class.

The zero/guṇa/vṛddhi columns apply Whitney’s general vowel correspondences (§§235–252) as a learner model. They are not independently verified lexical weak/strong series. The retained source PDFs for the legacy course seed are absent, so that selection is preserved but not described as fully reconstructible. Calibration is AI-assisted by OpenAI Codex, GPT-5 model family, dated 26 August 2026, schema 2.0.0; no human Sanskritist review is claimed.

Verbs · Module 03

Verbs and Their Grades

501 consolidated roots and retained verbal bases in standard Sanskrit dictionary order. Obvious duplicate citation variants and prefixed Core 1,000 occurrences are merged; six retained denominative, derived or prefixed entries are labelled as such rather than presented as independently graded roots.

Search by root, present form, grade form or English meaning.

How to read this reference: each row gives a source-linked citation entry, class status and concise lexical sense. The next three columns show a transparent vowel-grade model only; they do not assert that each form is attested. Derived entries show a dash, and an exact MW locator plus the row’s grade status appears under the meaning.

Verbs · Irregular reference

Irregular and Lexical Verb Roots

Roots whose grades, stems, or present systems cannot be read safely from one mechanical rule. They remain in the complete root table as well.

Search irregular roots

Verbs · Module 02

Verb Root Classes and Paradigms

Begin with the stem-building recipe for each of the ten present classes, then use the complete active and middle paradigms below.

Visual overview

How each class builds its present stem

The class number tells you the present-stem formation, not every form the root can make in other tense systems.

Class	Traditional group	Grouping	Present-stem operation	Worked third singular	Evidence
1	भ्वादिbhvādi · Whitney bhū-class	ThematicWhitney: a-conjugation	Add -a-; apply guṇa when the root permits it.Guṇa is not possible for every root.	भू → भव- → भवतिbhū → bhava- → bhavati“is; becomes”	Whitney §606 VI §735aMW L=151456AI-assisted source-checked · no human review
2	अदादिadādi · Whitney ad-class	AthematicWhitney: non-a-conjugation	The root itself is the present stem; endings join it directly.Singular active may use a strong form; consonant junctions vary.	अद् + ति → अत्तिad + ti → atti“eats”	Whitney §603 I §§611–612MW L=3575AI-assisted source-checked · no human review
3	जुहोत्यादिjuhotyādi · Whitney hu-class	AthematicWhitney: non-a-conjugation	Prefix reduplication; strong and weak stems can differ.For √hu, the strong stem is juho- and the weak stem juhu-.	हु → जुहो-/जुहु- → जुहोतिhu → juho-/juhu- → juhoti“sacrifices; offers”	Whitney §603 II §§642–647aMW L=263443AI-assisted source-checked · no human review
4	दिवादिdivādi · Whitney div-class	ThematicWhitney: a-conjugation	Add -ya- to the accented, unstrengthened root.The div/dīv example has a specifically prescribed vowel lengthening.	दिव् → दीव्य- → दीव्यतिdiv → dīvya- → dīvyati“plays; gambles”	Whitney §606 VIII §§759–765MW L=92229AI-assisted source-checked · no human review
5	स्वादिsvādi · Whitney su-class	AthematicWhitney: non-a-conjugation	Add -nu-; its strong form is -no-.The weak -nu- changes shape before some vowel-initial endings.	सु → सुनो-/सुनु- → सुनोतिsu → suno-/sunu- → sunoti“presses out (Soma)”	Whitney §603 IVa §§697–698MW L=245241AI-assisted source-checked · no human review
6	तुदादिtudādi · Whitney tud-class	ThematicWhitney: a-conjugation	Add accented -a-; normally leave the root unstrengthened.Some class-6 stems instead show nasal strengthening (§758).	तुद् → तुद- → तुदतिtud → tuda- → tudati“thrusts; strikes”	Whitney §606 VII §§750–758MW L=85879AI-assisted source-checked · no human review
7	रुधादिrudhādi · Whitney rudh-class	AthematicWhitney: non-a-conjugation	Insert a nasal before the final consonant; strong forms expand it to -na-.The nasal adapts to the following consonant; junction rules produce ruṇaddhi.	रुध् → रुणध्-/रुन्ध्- → रुणद्धिrudh → ruṇadh-/rundh- → ruṇaddhi“obstructs”	Whitney §603 III §§683–684bMW L=178755AI-assisted source-checked · no human review
8	तनादिtanādi · Whitney tan-class	AthematicWhitney: non-a-conjugation	Add -u-; its strong form is -o-.This is a very small class; √kṛ is a common irregular member.	तन् → तनो-/तनु- → तनोतिtan → tano-/tanu- → tanoti“stretches; extends”	Whitney §603 IVb §§697–714MW L=82226AI-assisted source-checked · no human review
9	क्र्यादिkryādi · Whitney krī-class	AthematicWhitney: non-a-conjugation	Add strong -nā- or weak -nī-.The ī of weak -nī- disappears before a vowel-initial ending.	क्री → क्रीणा-/क्रीणी- → क्रीणातिkrī → krīṇā-/krīṇī- → krīṇāti“buys”	Whitney §603 V §§717–718MW L=58088AI-assisted source-checked · no human review
10	चुरादिcurādi · Whitney cur-class	ThematicWhitney: derivative a-stem	Add -aya- to a generally strengthened root.Whitney treats this as derivative/causative; strengthening has exceptions (§1042).	चुर् → चोरय- → चोरयतिcur → coraya- → corayati“steals”	Whitney §§607, 775 §§1041–1043MW L=74646AI-assisted source-checked · no human review

Structural memory pattern: Whitney groups classes 1, 4 and 6 as a-stems and classes 2, 3, 5, 7, 8 and 9 as non-a-stems. The traditionally numbered tenth class also inflects as an a-stem, but Whitney treats it as a derivative/causative formation rather than a fully parallel present class.

Source, checking and limits

These ten learner summaries are project-authored explanations checked row by row against William Dwight Whitney, Sanskrit Grammar, second edition (1889), Chapter IX §§601–775 and, for the traditional tenth class, Chapter XIV §§1041–1043. Each worked third-singular form and short translation is also checked against one exact entry in the bundled Cologne Monier–Williams XML edition. Follow each Whitney section link in the Evidence column to inspect the public-domain grammar and its scan-backed transcription.

The traditional Sanskrit -ādi labels, “thematic/athematic” labels, Devanāgarī display, prose summaries and derivational arrows are project teaching aids, not Whitney’s wording. Row limitations deliberately retain Whitney’s qualifications, including class-10 derivative status and exceptions to strengthening. AI-assisted calibration was produced by OpenAI Codex, GPT-5 model family, on 22 August 2026 under schema 2.0.0; no independent human Sanskritist review is claimed.

Source, transcription and limits

All 648 IAST cells in the 16 present-system tables and two perfect tables reproduce McComas Taylor, The Little Red Book of Sanskrit Paradigms, sixth edition (2009), book pp.43–60 / PDF pp.54–71; all 18 pages have been rendered and visually checked. Twelve source cells contain two alternatives, yielding 660 complete displayed forms. Whitney Chapter IX and Chapter X provide broader present-system and perfect-system context, not replacement transcription witnesses.

Devanāgarī, expanded context for the PDF’s inherited paired-table headings, English person/number labels and navigation are project display aids. This AI-assisted source-transcription calibration was produced by OpenAI Codex, GPT-5 model family, on 22 August 2026 under schema 2.0.0. It has not received independent full grammatical review. Taylor publicly declared the book to be in the public domain in 2013; attribution and the evidence link are retained in Resources.

Verbs · System reference

Finite and Non-Finite Paradigms

These belong here in Verbs: they organize whole verbal systems. Morphology still explains how individual endings and stems are built.

Complete example

√गम् · √gam

The selected finite or non-finite √gam paradigm appears here without leaving this page.

Verbs · Generated inflection

Complete Verb Charts

Choose a root to generate its available finite lakāras and non-finite kṛdanta forms from an indexed dhātupāṭha entry.

Find a root

Reliability, saved forms and provenance

Charts come from the University of Hyderabad’s Saṁsādhanī APIs. Every successful response is also saved in this browser, so an opened chart can be shown again if the service is temporarily unavailable; clearing this site’s browser data removes those saved responses, and a root not opened before still needs the service. Saṁsādhanī also publishes its GPL source code, which leaves a future independently hosted generator possible without treating generated output as literary evidence.

The site requests all ten finite systems in active, middle and passive, then suppresses an empty voice instead of presenting dashes as a paradigm. The non-finite view displays every suffix family returned by the current unprefixed kṛt API, including unavailable families. It does not claim perfect/future participles or the prefixed -ya absolutive when the API has not returned them. Generated forms are analytical aids, not evidence that every form is attested in literature.

Verbs · Preverbs · Overview

How Preverbs Build a Complex Verb

Preverb is the standard technical label used here. A preverb stands before a verbal form and helps build a prefixed verbal word, redirecting or reshaping its meaning. The finished meaning is lexical: it must be checked rather than guessed by translating each piece mechanically.

Order of pieces

One root, several verbal builds

vyagacchan is one imperfect form, not the generic form of vi-√gam.

preverbvi- augmenta- when required root / formed stemgam / gaccha- / gamaya- suffix / ending-ti / -n / -ta

Present Finite

Structure: preverb + present stem + personal ending.
vi-√gam: vi + gaccha + ti → vigacchati
वि + गच्छ + ति → विगच्छति

vi- is the preverb, gaccha- the present stem and -ti the third-singular present ending. There is no past augment.

Augmented Finite: Imperfect

Structure: preverb + augment + tense stem + secondary ending.
vi-√gam: vi + a + gaccha + n → vyagacchan
वि + अ + गच्छ + न् → व्यगच्छन्

a- is the past augment and -n is the third-plural secondary ending. This is the page's worked imperfect, not a citation form.

Derived Finite: Causative

Structure: preverb + causative stem + personal ending.
vi-√gam: vi + gamaya + ti → vigamayati
वि + गमय + ति → विगमयति

gamaya- is the causative stem and -ti the third-singular present ending. The causative stem is formed before the ending is added.

Participial Formation

Structure: preverb + zero-grade root form + participial suffix.
vi-√gam: vi + gam + ta → vi + ga + ta → vigata
वि + गम् + त → वि + ग + त → विगत

This -ta participle uses the root’s weak, zero-grade form ga-, not the present stem gaccha-. Before -ta, final m is lost after a, so gam + ta → gata-. vigata can mean “gone apart” or “dispersed.”

Internal Boundary Change

Only the augmented example begins with a vowel collision. Before augment a-, final i of vi- becomes the glide y: vi + a → vya. In vi + gaccha-, vi + gamaya- and vi + gata-, the following sound is a consonant, so that vowel-to-glide change is not triggered.

Stem Formation Is Not Sandhi

gam is the citation root; gaccha- is its present stem, gamaya- its causative stem and gata- its participial base. Those formation-specific shapes are not competing sandhi results caused by vi-. The external ending or participial suffix belongs after the relevant stem has been selected.

Scope: these are representative structures supported by the exact vi-√gam dictionary records and the bounded worked imperfect. They do not stand in for every Sanskrit tense, mood, voice or non-finite formation.

Sources, scope and verification

The bounded inventory follows Aṣṭādhyāyī 1.4.58, with verbal use distinguished under 1.4.59. Whitney §§1076–1081 supplies wider historical and syntactic context; §§1086–1087g discusses internal combination and the augment; §§585–586 describes the augment itself; and §954d records the weak participial treatment √gam + ta → gata.

The exact bundled vi-√gam record, MW L=194440, supplies -gacchati and causative -gamayati; adjacent MW L=194441 supplies vigata. The displayed imperfect derivation is an original constructed working, not a literary attestation. The heading, explanations, diagrams, translations and derivational steps are original project teaching prose, not Pāṇini's, Whitney's or Monier–Williams's wording. AI-assisted by OpenAI Codex, GPT-5 model family, 25 August 2026, schema 1.0.0; source-checked, with no independent human Sanskritist review.

Verbs · Preverbs · List

Preverb and Root Reference

Search all twenty-two traditional prādi forms or the exact dictionary-backed root combinations selected for this learner reference. Sort by Sanskrit preverb or by the main root; the direction button reverses either order.

Search preverb, root or meaning Alphabetical order

Preverb	Core direction	Main root	Representative combined meaning	Evidence

What this list does and does not claim

The twenty-two rows come from the traditional prādi inventory, but surface variants such as nis/nir and ud/ut belong to conditioned families rather than independent English meanings. The 38 combinations are a bounded selection of exact bundled Monier–Williams verb="pre" records. They are useful dictionary examples, not a corpus-frequency ranking, and the list is not exhaustive.

No finite-verb pairing has been supplied for dus, dur or su because the adopted tagged records do not support a secure representative set. The omission is explicit rather than conjectural. Explanations and concise meanings are project-authored teaching prose. AI-assisted by OpenAI Codex, GPT-5 model family, 24 August 2026, schema 1.0.0; source-checked, with no independent human Sanskritist review.

Morphology title with a Sanskrit stem and ending joining to form a word

Morphology · Module 01

Ending Analyzer

Enter an inflected Sanskrit form. The analyzer identifies plausible endings and shows what grammatical information each ending can carry.

Try

प्रकृति + प्रत्यय

The analyzer is intentionally transparent: it lists possible analyses rather than pretending that an ending always has only one meaning.

1

Find the ending

Work from the right edge. Test the longest likely ending first, since Sanskrit endings often contain several letters.

2

Recover the stem

Remove the ending, then reverse predictable joining changes. A visible surface form may not reveal the dictionary stem directly.

3

Use syntax

Case, number, gender, person and voice must agree with the rest of the sentence. Context chooses among formal possibilities.

Morphology · Form Generator

Generate a Complete Paradigm

Enter a dictionary stem or choose a known paradigm. The result keeps the lemma, stem, case, number and surface form visibly separate.

Morphology · Module 02

Complete Nominal Paradigms

Twenty-nine source-transcribed paradigms from Taylor tables 1–29. Stems and stem alternants are black; inflectional endings are red.

Source, transcription and limits

Every IAST cell is reproduced from McComas Taylor, The Little Red Book of Sanskrit Paradigms, sixth edition (2009), book pp.1–29 / PDF pp.12–40, and all 29 pages have been visually checked. Slash alternatives remain source-visible; where Taylor abbreviates the second alternative, the complete display form appears beside the exact compact notation. Whitney Chapter V supplies broader grammatical context, not a silent replacement witness.

Devanāgarī, family-label normalization and stem/ending colour segmentation are project display aids, not a second textual witness. This is an AI-assisted source-transcription calibration by OpenAI Codex, GPT-5 model family, dated 22 August 2026, schema 2.0.0. It has not received an independent full grammatical review. Taylor publicly declared the book to be in the public domain in 2013; attribution and the evidence link are retained in Resources.

Quick reference

Generic nominal endings

These are abstract ending patterns. The full paradigms below show how each ending joins to a real stem.

Case	Singular		Dual		Plural
Case	Masculine / feminine	Neuter	Masculine / feminine	Neuter	Masculine / feminine	Neuter
Nominative	ः · -ḥ	—	औ · -au	ई · -ī	अः · -aḥ	इ · -i
Accusative	अम् · -am	—	औ · -au	ई · -ī	अः · -aḥ	इ · -i
Instrumental	आ · -ā	आ · -ā	भ्याम् · -bhyām	भ्याम् · -bhyām	भिः · -bhiḥ	भिः · -bhiḥ
Dative	ए · -e	ए · -e	भ्याम् · -bhyām	भ्याम् · -bhyām	भ्यः · -bhyaḥ	भ्यः · -bhyaḥ
Ablative	अः · -aḥ	अः · -aḥ	भ्याम् · -bhyām	भ्याम् · -bhyām	भ्यः · -bhyaḥ	भ्यः · -bhyaḥ
Genitive	अः · -aḥ	अः · -aḥ	ओः · -oḥ	ओः · -oḥ	आम् · -ām	आम् · -ām
Locative	इ · -i	इ · -i	ओः · -oḥ	ओः · -oḥ	सु · -su	सु · -su
Vocative	—	—	औ · -au	ई · -ī	अः · -aḥ	इ · -i

Morphology · Module 03

Complete Pronoun Paradigms

Personal pronouns and the full masculine, neuter, and feminine paradigms of tad, ayam, and asau.

Source, transcription and limits

All 264 IAST cells in these 11 paradigms reproduce McComas Taylor, The Little Red Book of Sanskrit Paradigms, sixth edition (2009), tables/book pp.30–40 / PDF pp.41–51; all 11 pages were rendered and visually checked. The 22 two-form cells retain Taylor’s notation and explicit complete readings; 33 printed vocative dashes remain absence marks. Whitney Chapter VII provides broader pronoun context, not a replacement transcription witness.

Devanāgarī, English case/number labels, category labels and navigation are project display aids. This AI-assisted source-transcription calibration was produced by OpenAI Codex, GPT-5 model family, on 22 August 2026 under schema 2.0.0. It has not received independent full grammatical review. Taylor publicly declared the book to be in the public domain in 2013; attribution and the evidence link are retained in Resources.

य

Relative yad

“who, which” follows the tad pattern with y-. It commonly pairs with tad: “the one who …, that one …”

क

Interrogative kim

“who? what? which?” follows the same pattern with k-: कः kaḥ, का kā, किम् kim.

ए

Near demonstrative etad

“this”: एषः eṣaḥ, एषा eṣā, एतत् etat; most oblique forms use eta-.

Morphology · Module 04

Participles and Non-Finite Forms

These forms bridge verbs and nominals: identify their verbal stem first, then parse their nominal ending.

Click any example or paradigm cell for its complete formation

Source, transcription and limits

The two declension tables below reproduce all 48 IAST cells in Taylor tables/book pp.41–42 / PDF pp.52–53, and both pages were rendered and visually checked. Taylor’s table 42 note about feminine nayantī following nadī is retained as a source note. Whitney Chapter XIII supplies broader participial context, not a replacement transcription witness.

The separate clickable formation overview and workups, Devanāgarī, English labels and navigation are project-authored aids outside the exact-cell claim. This AI-assisted calibration was produced by OpenAI Codex, GPT-5 model family, on 22 August 2026 under schema 2.0.0. It has not received independent full grammatical review. Taylor publicly declared the book to be in the public domain in 2013; attribution and the evidence link are retained in Resources.

Formation	Typical shape	Example	Core meaning
Present active	present stem + ant	गच्छन् · gacchan (stem gacchant-)	going; one who is going
Present middle	present stem + māna	लभमानः · labhamānaḥ	obtaining for oneself
Past passive	root + ta / na	कृतः · kṛtaḥ; भग्नः · bhagnaḥ	done; broken
Perfect active	perfect stem + vas	विद्वान् · vidvān (stem vidvas-)	one who has known
Future active	future stem + ant	करिष्यन् · kariṣyan	about to do; going to do
Gerundive	root + tavya / anīya / ya	कर्तव्यम् · kartavyam	to be done; must be done
Absolutive	root + tvā; prefixed root + ya	कृत्वा · kṛtvā; आगत्य · āgatya	having done; after doing
Infinitive	guṇa root + tum	कर्तुम् · kartum	to do; in order to do

Fast reading test: if the form agrees with a noun in gender, case and number, treat it as a participle. If an indeclinable form ends in -tvā, -ya or -tum, it does not decline.

Analyze · Generate

Sanskrit Parser

Analyze what the adopted sources support. Generate only what an exact model supplies.

Parser · Analyze & Generate

One Workspace, Two Directions

Analyze a written form without suppressing ambiguity, or request one precisely specified form. Results identify whether evidence comes from the site’s references, a generator or an external service.

Written form → supported possibilities

Analyze

Accepts Devanāgarī or IAST. The built-in references compare endings, exact Core 1,000 lemma candidates and source-held verb citation forms, then suggest matching entries from the bundled Monier–Williams dictionary. A configured hosted edition may also append an explicitly labelled Sanskrit Heritage analysis.

Try

Supported possibilities

Enter one form. The Parser lists supportable possibilities; it does not choose a context-dependent answer for you.

Scope and rights boundary

This is a deliberately bounded study aid, not unrestricted automatic parsing or sentence translation. Suggested definitions come from the bundled Cologne Monier–Williams edition and remain lexical evidence, not a context-sensitive translation. Core Vocabulary ranking and gloss fields retain their documented DCS/kosha/Cologne provenance and licences. Ending analyses are reference rules, and regular nominal output adapts the three displayed Taylor-derived models. Finite verbs are returned live by the University of Hyderabad’s Saṁsādhanī service for an exact indexed reading; its output is generated, not quoted from literature.

1,000Core Vocabulary lemmas

501verb-root families

446 / 501entries with retained exact generator mappings

964selectable dhātupāṭha readings

व्याकरणम् · vyākaraṇam

Grammar

See how endings turn into relationships.

Grammar · Module 01

The Eight Cases

Read the relationship expressed in its sentence. An ending identifies a case, but context selects that case’s function.

Case	Primary relation	How it works	Source example and working	Evidence and status
Nominativefirst case · prathamā	Who or what is presented as the subject?Clause roles · subject, attribute, apposition, predicate nominative	It marks the grammatical subject and words that qualify or rename that subject.Limit: grammatical subject is not the same as semantic doer; a passive patient can be nominative.	इन्द्रो ब्राह्मणो ब्रुवाणःindro brāhmaṇo bruvāṇaḥIndra, calling himself a BrahmanBoth indraḥ and predicate brāhmaṇaḥ are nominative; external sandhi gives -o before the following voiced consonants.	Whitney §267 example §268aSyntax · source-attested exampleAI-assisted source-checked · no human review
Accusativesecond case · dvitīyā	Whom or what does the action directly reach?Argument structure · direct object, goal, extent, adverbial use	It most often marks a transitive verb’s object; it can also express a goal of motion or an extent of place, time or manner.Limit: it is not a universal English “to” case; predicate choice distinguishes it from dative uses.	अग्निम् ईडेagnim īḍeI praise Agniagnim is accusative singular, the direct object of first-person īḍe.	Whitney §269 example §270Syntax · source-attested exampleAI-assisted source-checked · no human review
Instrumentalthird case · tṛtīyā	With whom, or by what means?Participant roles · accompaniment, means, instrument, agent	Its core “with” relation extends from accompaniment to the means, instrument or agent by which something happens.Limit: the example shows means; an instrumental is read as a passive agent only when the construction supports that role.	शस्त्रेण निधनम्śastreṇa nidhanamdeath by the swordśastreṇa is instrumental singular and supplies the means associated with nidhanam.	Whitney §278 example §280Syntax · source-attested exampleAI-assisted source-checked · no human review
Dativefourth case · caturthī	To or for whom or what?Argument structure · indirect object, recipient, beneficiary, purpose	It points toward a recipient, beneficiary, reference or intended end; purpose is an especially common extension.Limit: not every English “to” or “for” maps to a dative, and Whitney notes that Sanskrit prepositions do not govern it.	धनुर् दर्शय रामायdhanur darśaya rāmāyaShow the bow to Ramarāmāya is dative singular, the recipient of the showing; dhanuḥ surfaces as dhanur before voiced d-.	Whitney §§285–285a example §286b limit §288AI-assisted source-checked · no human review
Ablativefifth case · pañcamī	From whom or what?Semantic relations · source, separation, removal, distinction, cause	It presents a starting point: physical departure, separation, origin, or the source from which a result arises.Limit: “from” is the organizing relation, not a fixed translation; the local predicate and context determine the English wording.	लोभात् क्रोधः प्रभवतिlobhāt krodhaḥ prabhavatiAnger arises from greedlobhāt is ablative singular and identifies the source or cause from which krodhaḥ arises.	Whitney §289 example §291Syntax · source-attested exampleAI-assisted source-checked · no human review
Genitivesixth case · ṣaṣṭhī	Whose, or related to what?Noun modification · possession, appurtenance, partitive, subjective/objective relation	It qualifies a noun by placing it in a relation that the two nouns and their context define more precisely.Limit: possession is common but not automatic; English “of” can conceal several distinct relationships.	इन्द्रस्य वज्रःindrasya vajraḥIndra’s thunderboltindrasya is genitive singular and modifies nominative vajraḥ with a possessive/appurtenant relation.	Whitney §§294a–b example §295Syntax · source-attested exampleAI-assisted source-checked · no human review
Locativeseventh case · saptamī	In, on or at whom or what?Setting · spatial location, point in time, sphere, condition, circumstance	It locates something in a place, time, state or sphere; contextual extensions can overlap other case relations.Limit: its default is situation, not destination, although a “pregnant” construction can express a goal or place of rest.	पर्वतस्य पृष्ठेparvatasya pṛṣṭheon the ridge of the mountainpṛṣṭhe is locative singular; genitive parvatasya specifies the mountain whose ridge is the location.	Whitney §§301a–c example §302a limit §304AI-assisted source-checked · no human review
Vocativeaddress form · saṃbodhana	Who is being addressed?Discourse function · direct address	It calls to or names the addressee rather than filling an ordinary clause role.Limit: Whitney notes that native grammarians treat it specially; apart from accent, it differs from the nominative only in the singular and not always there.	अग्नेagneO Agni!agne is the vocative singular of the masculine i-stem agni, attested at the start of Whitney’s cited verse.	Whitney §266a form §307k example §314cAI-assisted source-checked · no human review

Dual syncretism

The dual has three case forms: one for nominative–accusative–vocative, one for instrumental–dative–ablative, and one for genitive–locative. Context supplies the distinction.

गजौ · गजाभ्याम् · गजयोःgajau · gajābhyām · gajayoḥ

gajau serves as nominative, accusative or vocative dual; gajābhyām as instrumental, dative or ablative dual; and gajayoḥ as genitive or locative dual.

Whitney §308a

Neuter identity

Across the numbers, the neuter accusative is like the nominative; in the dual the shared form also serves the vocative. Vocative details still depend on stem class and accent.

फलम् · फले · फलानिphalam · phale · phalāni

These neuter singular, dual and plural forms can each be nominative or accusative. The dual phale and plural phalāni also serve as vocatives; the vocative singular is phala.

Whitney §307b · §308a · §309c

Grammar · Module 02

Compounds · samāsa

Expand the relation between the members, then ask whether the last member, all members together, or an external referent supplies the compound’s semantic head.

Type	Core relation	How it works	Source example and working	Evidence and status
Tatpuruṣadependent determinative	What case-relation connects the first member to the final member?Nominal composition · case relation · internal head	The final member names the whole; the first is understood as its genitive, dative, instrumental, ablative, locative or accusative dependent.Limit: Whitney uses tatpuruṣa for the wider determinative class; this learner row uses the common narrower label for his dependent division. The hidden case must be read from lexical and contextual evidence.	देवसेनाdevasenāarmy of godsdeva + senā → devasenā. Project expansion: देवानां सेना devānāṃ senā; the first member has a genitive relation to the internal head senā.	Whitney §1263a rule §1264 example §1264aSource-attested compoundAI-assisted source-checked · no human review
Karmadhārayadescriptive determinative	How does the first member qualify the final member?Nominal composition · description/apposition · internal head	The final noun names the whole while the first member describes or appositionally identifies it, rather than standing in a distinct case-relation.Limit: Whitney says the descriptive/dependent boundary is not absolute, and a noun-final form can also receive a possessive Bahuvrīhi reading in context.	नीलोत्पलnīlotpalablue lotusnīla + utpala → nīlotpala. Project expansion: नीलम् उत्पलम् nīlam utpalam; “blue” directly qualifies the internal head “lotus.”	Whitney §1263a rule §1279 example §1279a limits §§1279c–dSource-attested compoundAI-assisted source-checked · no human review
Dvandvacopulative / coordinative	Are the members coordinated as if by “and”?Nominal composition · coordination · multiple-member head	Two or more members are syntactically coordinate. The ordinary form follows the final member’s declension and uses dual or plural number according to the members counted.Limit: Whitney also records alternative “or” relations and neuter singular collectives, so neither “and” nor dual inflection is universal.	विद्याकर्मणीvidyākarmaṇīknowledge and actionvidyā + karman → vidyākarmaṇī. Project expansion: विद्या च कर्म च vidyā ca karma ca; two coordinated items license the attested dual.	Whitney §§1252–1252a inflection §1253a example §1253b collective §§1253c–dSource-attested compoundAI-assisted source-checked · no human review
Bahuvrīhipossessive adjective	Who or what possesses the relation expressed inside the compound?Adjectival composition · possession · external head	A noun-final compound acquires adjectival value: the whole qualifies an external noun and adds a “having” or “possessing” relation to its internal members.Limit: the same internal members can form a non-possessive determinative; context, agreement and sometimes accent distinguish the readings. Whitney rejects “relative” as the defining label.	बृहद्रथbṛhadrathapossessing great chariotsbṛhat + ratha → bṛhadratha. Project expansion: बृहन्तो रथा यस्य bṛhanto rathā yasya; the person described, not ratha, is the external head.	Whitney §§1292a–b rule §1293 example §1293a name/limit §§1293c–dSource-attested compound stemAI-assisted source-checked · no human review
Avyayībhāvaadverbial / indeclinable use	Does an indeclinable first member make the whole function adverbially?Adverbial composition · indeclinable/particle · clause modifier	Whitney’s central type is an adverbially used accusative of a secondary adjective compound whose first member is an indeclinable or particle.Limit: other case-forms occur, and a few yathā- compounds are adjectival. “Indeclinable first member” alone therefore does not settle the compound’s use.	यथावशम्yathāvaśamas one choosesyathā + vaśam → yathāvaśam. The relative adverb sets the manner and the accusative-form whole modifies the clause adverbially.	Whitney §§1313–1313a example §1313c limits §§1313d–eSource-attested compoundAI-assisted source-checked · no human review
Dvigunumeral-first substantive	Does a numeral-first compound name a collective, measure or related substantive?Numeral composition · substantivization · collective/abstract value	Whitney reports the native class for substantively used numeral-first compounds, together with some related adjective compounds; number and gender follow the lexicalized use.Limit: not every numeral-first expression is Dvigu: compound cardinals and numeral possessives have separate analyses, and Whitney’s Dvigu examples vary in accent and semantic subtype.	त्रिलोकीtrilokīthe three worldstri + loka → trilokī. Project expansion: त्रयो लोकाः trayo lokāḥ; the attested feminine form names the collection as a substantive.	Whitney §§1312–1312a example §1312b compound numerals §1261 numeral possessives §1300Source-attested later compoundAI-assisted source-checked · no human review

Source, terminology and limits

These six explanations, technical groupings, Sanskrit expansions and worked readings are project-authored teaching prose checked against William Dwight Whitney, Sanskrit Grammar, second edition (1889), Chapter XVIII §§1247–1313. The displayed compounds and short meanings are exact witnesses in the linked sections; pitch accents are removed, Whitney’s historical ç is normalized to current ś, and matching accent-free Devanāgarī is supplied as a project reading aid.

Whitney’s descriptive/dependent distinction does not map without residue onto every later classroom label, and the same member sequence can receive different compound readings. Nothing here is represented as Whitney’s, Pāṇini’s or Vasu’s wording. AI-assisted calibration was produced by OpenAI Codex, GPT-5 model family, on 22 August 2026; no independent human Sanskritist review is claimed.

Grammar · Module 03

Particles and Indeclinables

Use these forms as signals, not one-word translation switches: position, construction and period can change the job they perform.

Form	Core relation	How it works	Source example and working	Evidence and status
चca	and; also; bothCoordination · enclitic/postpositive particle	It follows a word or group and links it with what precedes; a repeated ca … ca can mark both members.Limit: single and repeated patterns vary by period; later prose often omits the first ca.	अहं च त्वं चahaṃ ca tvaṃ caI and youEach ca follows one coordinated pronoun; the example is cited by MW from ṚV 8.62.11.	Whitney §1133MW L=70283Attested lexical exampleAI-assisted source-checked · no human review
वाvā	or; eitherCoordination · disjunction · postpositive particle	It follows the alternative it marks. Repetition can make an explicit “either … or” frame.Limit: it need not appear after every member; lists and negative disjunctions use several source-recorded patterns.	वेदं वेदौ वेदान् वाvedaṃ vedau vedān vāone Veda, or two, or more than twoThe final vā follows the last alternative while the number forms supply the earlier alternatives.	Whitney §1133 example §265aMW L=189435AI-assisted source-checked · no human review
तुtu	but; howeverDiscourse relation · adversative/postpositive particle	It normally follows some opening material and sets the following thought in contrast with what is active in the discourse.Limit: older tu can be asseverative or inciting, and verse may use it as an expletive; “but” is not automatic.	नीरुजस् तु किम् औषधैःnīrujas tu kim auṣadhaiḥBut what use are medicines to a healthy person?tu follows nīrujas and marks the contrast presupposed by the rhetorical question.	Whitney §1133b limits §1122b example §281fMW L=85706–85706.1AI-assisted source-checked · no human review
हिhi	for; because; indeedDiscourse relation · illative or asseverative particle	It generally follows the first word and can present a reason, supporting explanation or emphatic assertion.Limit: Whitney traces the causal use to an asseverative particle and also records metrically expletive uses.	सर्वो हि पृतना जिगीषतिsarvo hi pṛtanā jigīṣatiFor everyone wishes to win battleshi follows first-position sarvaḥ; external sandhi gives displayed sarvo hi.	Whitney §1133c history/limit §1122bMW L=262729 · exact exampleAI-assisted source-checked · no human review
एवeva	just; only; preciselyInformation structure · focus/emphasis particle	In later Sanskrit it strengthens the word immediately before it; English chooses “only,” “very,” “exactly,” or another contextual focus rendering.Limit: MW records uses with no exact separate translation, so “only” must not be inserted mechanically.	त्वम् एव यन्ता नान्योऽस्ति पृथिव्याम्tvam eva yantā nānyo 'sti pṛthivyāmYou alone are a charioteer; there is no other on eartheva follows tvam and gives it exclusive focus; the second clause makes that reading explicit.	Whitney §1102e context §1122bMW L=40181.2 · exact exampleAI-assisted source-checked · no human review
अपिapi	also; even; moreoverAdditive/focus particle · indefinite-forming use	It can add or emphasize an item; after an interrogative form it can create an indefinite expression.Limit: “although” belongs to combinations such as yady api; it is not the fixed meaning of api alone.	अन्यद् अपि · कोऽपिanyad api · ko 'pisomething more · someone or anyoneapi adds to anyat; after interrogative kaḥ, sandhi produces the indefinite ko 'pi.	Whitney §1077a indefinites §507MW L=10132.15, 10133, 10134AI-assisted source-checked · no human review
नna	not; no; norPolarity · simple negative particle	It negates the predicate or constituent within its scope and can be repeated across coordinated clauses.Limit: for a direct prohibition, first test mā; na has broader simple-negation work.	न ता नशन्ति न दभाति तस्करःna tā naśanti na dabhāti taskaraḥThey do not become lost; no thief harms themThe first na negates naśanti; the second separately negates dabhāti.	Whitney §1122c example §576MW L=102621AI-assisted source-checked · no human review
माmā	do not; let notModality/polarity · prohibitive particle	It forms a prohibition. The older regular pattern uses an augmentless past form, especially an aorist; later Sanskrit also uses an imperative.Limit: the familiar mā gaccha type is ordinary later, but Whitney does not treat it as the older default.	समाश्वसिहि मा शुचःsamāśvasihi mā śucaḥBe comforted; do not grievemā scopes over the augmentless prohibitive form śucaḥ; the preceding imperative is positive.	Whitney §§1122c–d construction §579 example §579a later use §579cMW L=161686AI-assisted source-checked · no human review
इतिiti	thus; quotation/citation boundaryMetalinguistic particle · speech, thought, name or reason	It follows and closes the expression being quoted, thought, named or otherwise treated as a unit.Limit: it is not an opening quotation mark and often receives no separate English word.	ब्रह्मजायेऽयम् इति चेद् अवोचन्brahmajāye 'yam iti ced avocanIf they have said, “This is a Brahman’s wife”iti follows the quoted words brahmajāye 'yam; ced avocan resumes the reporting clause.	Whitney §§1102a–b clause context §1131aMW L=28747AI-assisted source-checked · no human review
यदा … तदाyadā … tadā	when/whenever … thenClause linkage · temporal relative–correlative pair	yadā opens the relative time frame; tadā can point back to that time in the main clause.Limit: tadā is not obligatory and is not the only possible correlative; either adverb can also occur independently.	यदा क्षयं गतं सर्वम्yadā kṣayaṃ gataṃ sarvamwhen everything had gone to ruinThis MW-cited example attests the yadā clause without an expressed tadā, illustrating the stated limitation.	Whitney §1103b conjunction context §1132MW L=170138, 82170, 82174AI-assisted source-checked · no human review
यदि … तर्हिyadi … tarhi	if … then/in that caseClause linkage · condition and optional correlative	yadi introduces a condition; tarhi can mark the consequence as applying “then” or “in that case.”Limit: tarhi is optional, and the tense or mood in either clause depends on period and intended condition.	यदि जीवति भद्राणि पश्यतिyadi jīvati bhadrāṇi paśyatiif he lives, he sees good fortuneThe cited later-language example uses present forms in both clauses and no explicit tarhi.	Whitney §§1103d–e conjunction context §1132MW L=170219, 83501AI-assisted source-checked · no human review
यतः … ततःyataḥ … tataḥ	whence/since … thence/thereforeClause linkage · source-relative and correlative adverbs	The pair first expresses source: relative “from where/which” and correlative “from there/that.” Context can extend that relation to reason and result.Limit: “because … therefore” is a contextual causal reading, not the pair’s only or automatic meaning.	यतस् ततःyatas tataḥfrom any quarter whateverThe exact MW formula is spatial. Before t-, citation-form yataḥ surfaces as yatas.	Whitney §1098a conjunction context §1132MW L=169627/169627.4/81788.5AI-assisted source-checked · no human review

Source, normalization and limits

These twelve explanations and worked readings are project teaching prose checked against Whitney’s discussion of indeclinables and conjunctions, with exact form, sense, placement and example evidence from the bundled Cologne Monier–Williams edition. The eight former classroom sentences were removed; the remaining examples are source-attested sentences, cited formulas or explicitly identified lexical examples.

The display removes Vedic pitch accents, modernizes Whitney’s historical ç and āi/āu conventions to the site’s IAST, converts cited MW SLP1 mechanically to IAST and supplies synchronized accent-free Devanāgarī. Meanings remain contextual, especially for focus particles and relative–correlative pairs. AI-assisted calibration was produced by OpenAI Codex, GPT-5 model family, on 22 August 2026; no independent human Sanskritist review is claimed.

Grammar · Module 04

Sentence Patterns

Sanskrit word order is flexible because endings mark relationships. Select a pattern for a full explanation and worked examples.

Syntax · Agreement

Agreement Identifies Which Words Belong Together

A finite verb marks person and number; an adjective or participle qualifying a noun matches that noun in gender, case and number. Those signals help connect words even when order alone does not.

Technical terms · domainfinite verb · person · number · nominal concord · gender · case · morphosyntax

Reading testIdentify the finite ending, then test which subject has compatible person and number; read nominal modifiers through their shared gender, case and number.

Condition / limit: agreement is a constraint, not proof of a unique parse. Syncretic endings and coordinated subjects can leave more than one analysis, and an unexpressed subject can still be encoded by the verb.

शिवे ते द्यावापृथिवी उभे स्ताम्śive te dyāvāpṛthivī ubhe stāmMay Heaven and Earth, both, be gracious to you.The coordinated pair, adjective śive, resumptive ubhe and verb stām are dual. Whitney §265Source-attested witness · normalized display

यूयं किम् इति सीदथyūyaṃ kim iti sīdathaWhy do you sit?yūyam is second-person plural and sīdatha is its second-person plural finite verb. Whitney §1102bSource-attested witness · normalized display

इन्द्रो ब्राह्मणो ब्रुवाणःindro brāhmaṇo bruvāṇaḥIndra, calling himself a Brahman.indraḥ, predicate brāhmaṇaḥ and participle bruvāṇaḥ are nominative masculine singular; sandhi gives final -o. Whitney §268aSource-attested witness · normalized display

Whitney §§527, 536, 541 · §§267–268aAI-assisted source-checked · no human review

Syntax · Predication and ellipsis

A Copula or Repeated Verb Can Be Left Unexpressed

Sanskrit can express a predicate without an overt form of “be,” and a verb recoverable from a parallel clause can also be omitted. English supplies only the relation licensed by the local construction.

Technical terms · domaincopula · predication · ellipsis · predicate dative/genitive · antithetical clause · syntax

Reading testWhen no finite verb appears, identify the case and discourse relation first; then test a predicative or parallel-clause reading.

Condition / limit: adjacent nominatives do not automatically make a present-tense “X is Y” sentence. Whitney's witnesses also include non-nominative predicates and omission of a repeated past verb.

उपदेशो मूर्खाणां प्रकोपाय न शान्तयेupadeśo mūrkhāṇāṃ prakopāya na śāntayeGood counsel leads fools to anger, not calm.The predicate datives prakopāya and śāntaye express the result or tendency; the copular relation is unexpressed. Whitney §287aSource-attested witness · normalized display

सर्वाः सम्पत्तयस् तस्य संतुष्टं यस्य मानसम्sarvāḥ sampattayas tasya saṃtuṣṭaṃ yasya mānasamEvery good fortune belongs to one whose mind is content.Genitive tasya is predicative; the translation supplies “belongs to,” while saṃtuṣṭaṃ … mānasam agrees in the relative clause. Whitney §298Source-attested witness · normalized display

अग्निर् अमुष्मिं लोक आसीद् यमोऽस्मिन्agnir amuṣmiṃ loka āsīd yamo 'sminAgni was in yonder world; Yama in this one.Past āsīt is stated in the first antithetical clause and understood, not repeated, with yamaḥ. Whitney §597Source-attested witness · normalized display

Whitney §287a · §298 · §597AI-assisted source-checked · no human review

Syntax · Passive viewpoint

The Patient Is Nominative; an Expressed Agent Can Be Instrumental

A passive construction presents the affected participant as the grammatical subject. Whitney records the same instrumental-agent alignment with finite passive verbs, passive participles and gerundives.

Technical terms · domainvoice · patient · agent · nominative · instrumental · finite -yá- passive · participle · gerundive

Reading testBeside passive morphology, test whether an instrumental names the agent rather than an implement; then identify the nominative patient or impersonal predicate.

Condition / limit: the three source witnesses below contain a passive participle or gerundive, not a finite present passive. Finite passive stems in accented -yá- are separately described at Whitney §§768–771; not every instrumental is an agent.

यमेन दत्तःyamena dattaḥgiven by Yamayamena is the instrumental agent; passive participle dattaḥ agrees with a masculine nominative patient supplied by context. Whitney §282aSource-attested witness · normalized display

ऋषिभिर् ईड्यःṛṣibhir īḍyaḥto be praised by sagesṛṣibhiḥ is the plural instrumental agent; gerundive īḍyaḥ is nominative masculine singular. Whitney §282aSource-attested witness · normalized display

व्याधेन जालं विस्तीर्णम्vyādhena jālaṃ vistīrṇamA net was spread by the hunter.Neuter nominative jālam is the patient and agrees with vistīrṇam; vyādhena supplies the agent. Whitney §282aSource-attested witness · normalized display

Whitney §282a · §§768–771 · §999AI-assisted source-checked · no human review

Syntax · Genitive absolute

A Genitive Actor and Participle Frame a Separate Circumstance

In later Sanskrit, a genitive living actor with an agreeing genitive participle can stand apart from the main clause and supply a circumstance such as simultaneous seeing, hearing or speaking.

Technical terms · domainabsolute construction · genitive actor · agreeing participle · circumstantial clause · later-language syntax

Reading testFind the genitive participle and its genitive actor, bracket that pair, then read the remaining finite clause independently.

Condition / limit: Whitney calls the construction unknown in the earlier language, says its genitive actor is living, and notes a strong tendency toward verbs of seeing, hearing or uttering. A nuance of disregard can occur but is not automatic.

पश्यतो बकमूर्खस्य नकुलैर् भक्षिताः सुताःpaśyato bakamūrkhasya nakulair bhakṣitāḥ sutāḥWhile the foolish heron looked on, its young were eaten by ichneumons.paśyataḥ … bakamūrkhasya is the agreeing genitive pair; nominative sutāḥ belongs to the main passive clause. Whitney §300bSource-attested witness · normalized display

दिवं जगाम मुनीनां पश्यतां तदाdivaṃ jagāma munīnāṃ paśyatāṃ tadāHe went then to heaven while the ascetics watched.Plural genitives munīnām and paśyatām form the circumstance; jagāma is the finite verb of the main clause. Whitney §300bSource-attested witness · normalized display

इति वादिन एवास्य धेनुर् आववृते वनात्iti vādina evāsya dhenur āvavṛte vanātWhile he was still speaking, the cow returned from the forest.vādinaḥ … asya surfaces through sandhi as vādina evāsya; dhenuḥ is the main-clause subject. Whitney §300bSource-attested witness · normalized display

Whitney §300bAI-assisted source-checked · no human review

Syntax · Relative and correlative

A Ya- Clause Commonly Points to a Ta- Expression

The relative pronoun or adverb identifies an open participant or setting; a demonstrative expression in the related clause can point back to it. Each form receives the case required by its own clause.

Technical terms · domainrelative pronoun · demonstrative correlative · antecedent · relative clause · case assignment · clause linkage

Reading testTranslate the relative clause first, identify the job of its ya- form, then find the corresponding ta- expression and analyze its separate case.

Condition / limit: the two forms need not share a case, and the correlative can be omitted or replaced by an adverb. Whitney notes a preference—not an absolute rule—for the relative clause to come first.

यः सुन्वतः सखा तस्मा इन्द्राय गायतyaḥ sunvataḥ sakhā tasmā indrāya gāyataSing to that Indra who is the soma-presser's friend.Relative yaḥ is nominative inside its clause; correlative tasmā is dative with indrāya in the command. Whitney §512aSource-attested Vedic witness · normalized display

यं यज्ञम् परिभूर् असि स इद् देवेषु गच्छतिyaṃ yajñam paribhūr asi sa id deveṣu gacchatiThe offering that you protect goes among the gods.Accusative yam yajñam is the object of paribhūḥ asi; nominative saḥ becomes the subject of gacchati. Whitney §512aSource-attested Vedic witness · normalized display

हंसानां वचनं यत् तु तन् मां दहतिhaṃsānāṃ vacanaṃ yat tu tan māṃ dahatiBut what the swans said burns me.Neuter relative yat identifies the saying; correlative tat surfaces as tan before mām and is subject of dahati. Whitney §512aSource-attested witness · normalized display

Whitney §§508–510 · §512aAI-assisted source-checked · no human review

Syntax · Quotation and content marker

Iti Follows the Expression It Closes

इति iti is an indeclinable, postpositive boundary: it can close direct words or the content of a belief, perception or label, often without a separate English equivalent.

Technical terms · domainindeclinable · postpositive particle · direct quotation · content clause · metalinguistic boundary · sandhi

Reading testAt iti, look backward for the expression being enclosed, then identify how the surrounding verb treats that expression.

Condition / limit: iti marks the right boundary, not necessarily the left one; sandhi may fuse it with the preceding word, and the enclosed material need not be a verbatim spoken quotation.

ब्रह्मजायेऽयम् इति चेद् अवोचन्brahmajāye 'yam iti ced avocanIf they have said, “This is a Brahman's wife.”iti follows and closes brahmajāye 'yam; ced avocan belongs to the reporting frame. Whitney §1102bSource-attested Vedic witness · normalized display

तं देवा अब्रुवन् व्रात्य किं नु तिष्ठसीतिtaṃ devā abruvan vrātya kiṃ nu tiṣṭhasītiThe gods said to him, “Vrātya, why do you stand?”The quoted question ends with fused tiṣṭhasi + iti → tiṣṭhasīti; the reporting verb is abruvan. Whitney §1102bSource-attested Vedic witness · normalized display

यः श्रद्दधाति सन्ति देवा इतिyaḥ śraddadhāti santi devā itiWhoever believes that the gods exist.iti closes the believed content santi devāḥ; this is content rather than a separately reported speech event. Whitney §1102bSource-attested Vedic witness · normalized display

Whitney §§1102a–b · §1131aAI-assisted source-checked · no human review

Grammar · Searchable reference

Sanskrit Grammar Terminology

Search 122 core terms by Sanskrit, Devanāgarī, English meaning or category. Spellings with and without diacritics both work.

Source Reference · Vyākaraṇa

Pāṇini’s Aṣṭādhyāyī

Search all 3,983 numbered sūtras with the complete proofread Vasu historical layer, kept distinct from modern project-authored teaching notes.

सूत्रम्

Loading the complete source-text index…

Edition · Rights · Limits

Three Clearly Labelled Layers

The licensed source index, proofread historical interpretation and modern project teaching notes remain distinct.

Complete Source Edition

Scharf · Sanskrit Library · v0.1

Pāṇini, Aṣṭādhyāyī as in the Kāśikā: First XML edition, edited by Peter M. Scharf, version 0.1. Providence, RI: The Sanskrit Library, 2010; based on Scharf’s 1991–1992 first digital edition.

Complete Text Coverage

3,983 numbered sūtras

The searchable index follows The Sanskrit Library’s complete grammatical-data edition and preserves its traditional adhyāya and pāda order.

Index Rights

CC BY-NC-SA 3.0

Copyright © 2010 The Sanskrit Library. This adapted data layer is available only for noncommercial reuse with attribution and share-alike distribution. The licence does not relicense unrelated site code or original writing.

Readable Display

IAST and Devanāgarī

Both scripts are provided as reading aids, while source spacing and additional readings remain available with the indexed record.

Complete Historical Layer

3,983 Vasu Records · Books I–VIII

The pinned Sanskrit Programmers proofreading snapshot supplies an English rule rendering and Sanskrit vṛtti for every numbered record, plus a longer Vasu commentary body for 3,982. The source project expressly permits copying and use of its corrected text.

Modern Calibration Batch

Complete pāda 1.1 · 75 records

Each record has original plain-English learner prose, technical terms, domains, conditions, source-bounded working and examples, limitations, links, sources, verification status and per-record generation provenance. This project prose is never presented as Pāṇini’s or Vasu’s wording.

Known Limits

Source-checked · not human-reviewed

The learner layer was drafted with OpenAI Codex in the GPT-5 model family and checked against the named local witnesses. It is not a modern critical edition, expert certification or complete prakriyā, and it remains bounded to 1.1.1–1.1.75.

Sanskrit Library complete index ↗Edition and licence record ↗CC BY-NC-SA 3.0 ↗Vasu proofreading and reuse notice ↗Pinned proofread source snapshot ↗Vasu Book I ↗Vasu Book II ↗Vasu Book III ↗Vasu Book IV ↗Vasu Book V ↗Vasu Book VI ↗Vasu Book VII ↗Vasu Book VIII ↗Scan-formatting notice ↗

William Dwight Whitney seated with an open book in a parchment-toned scholarly portrait

Historical Reference · Complete Reading Edition

Whitney’s Sanskrit Grammar

Read every chapter and numbered section, search the whole transcription, and verify the wording against the scan.

Whitney · Complete Scan-Backed Transcription

William Dwight Whitney – Sanskrit Grammar

Loading the complete grammar text…

Bibliography · Digitization · Transcription · Rights

Source Record

The public-domain work, scanned witness and Wikisource transcription are distinct source layers.

01 · Bibliographic witness

Second Edition, 1889

William Dwight Whitney, A Sanskrit Grammar: Including Both the Classical Language, and the Older Dialects, of Veda and Brahmana, second revised and extended edition, Leipzig: Breitkopf & Härtel, 1889. The scanned witness is the Harvard University Press 1950 republication, with Geoffrey Cumberlege / Oxford University Press named on its title page.

02 · Digitization chain

Scan and Page Proofreading

The 1950 witness was scanned for Internet Archive item 1941sanskritgram00whituoft, uploaded to Wikimedia Commons as a 584-page DjVu, and proofread through the scan-backed English Wikisource index. Book-page markers open the corresponding scan pages.

03 · Reading coverage

Complete Grammar and Indexes

The reading edition includes all chapters, both indexes and the errata. Whole-book search returns each result to its chapter or numbered section.

04 · Rights

Public Domain + CC BY-SA 4.0

Whitney’s work and the source scan are public domain. Wikisource’s transcription contributions are reused under CC BY-SA 4.0.

Open bundled PDF ↗Scan-backed Wikisource index ↗Wikisource reading edition ↗CC BY-SA 4.0 ↗Internet Archive scan record ↗

पाठशाला · pāṭhaśālā

Reading Lab

Turn a dense line into a sequence of small decisions.

Reading Lab · Module 01

A Repeatable Reading Method

Expose the transmitted word forms first, then build a grammatical reading. The same source-attested sentence appears at every step so that each decision develops one continuous analysis. The order below is a project workflow, not wording attributed to Whitney or a claim that every sentence resolves uniquely.

Recover external sandhi.Propose citation-form boundaries before parsing. Here paśyataḥ + b- → paśyato b- (§175a) and nakulaiḥ + bh- → nakulair bh- (§174).पश्यतो बकमूर्खस्य नकुलैर् भक्षिताः सुताः।paśyato bakamūrkhasya nakulair bhakṣitāḥ sutāḥ.Terms · domain: external combination · padaccheda · phonology. Limit: a visible surface form can allow more than one split; preserve alternatives until morphology and syntax decide.nakulair sandhi §174 paśyato sandhi §175a exact sentence §300bAI-assisted source-checked · no human review
Locate the predicate; mark finite verbs.A personal ending identifies a finite form’s person and number. If none appears, test a participial predicate or licensed ellipsis instead of inventing a verb.पश्यतो बकमूर्खस्य नकुलैर् भक्षिताः सुताः।paśyato bakamūrkhasya nakulair bhakṣitāḥ sutāḥ.Terms · domain: finite verb · personal ending · predicate · morphosyntax. Condition: the worked line has no expressed finite verb; passive participle bhakṣitāḥ supplies the predication.finite-form test §541 passive participle §282a exact sentence §300bAI-assisted source-checked · no human review
Classify the verbal forms.Identify forms built from verbs and determine whether each one inflects. Here paśyataḥ and bhakṣitāḥ are both inflecting participles and must be parsed for agreement.पश्यतो बकमूर्खस्य नकुलैर् भक्षिताः सुताः।paśyato bakamūrkhasya nakulair bhakṣitāḥ sutāḥ.Terms · domain: participle · past passive participle · verbal morphology. Limit: both verbal forms in this sentence are inflecting participles; the sentence does not demonstrate an infinitive or absolutive.participles §537 exact sentence §300bAI-assisted source-checked · no human review
Parse nominal morphology.Record case, number and gender for nouns, adjectives and participles before assigning a sentence role.पश्यतो बकमूर्खस्य नकुलैर् भक्षिताः सुताः।paśyato bakamūrkhasya nakulair bhakṣitāḥ sutāḥ.Terms · domain: case · number · gender · nominal morphology. Limit: endings can be syncretic, and case identifies a relation family rather than one automatic English preposition.nominative §267 instrumental §278 genitive absolute §300bAI-assisted source-checked · no human review
Join agreement groups.Connect forms whose gender, case and number are compatible: genitive singular paśyataḥ … bakamūrkhasya, and nominative plural bhakṣitāḥ sutāḥ.पश्यतो बकमूर्खस्य नकुलैर् भक्षिताः सुताः।paśyato bakamūrkhasya nakulair bhakṣitāḥ sutāḥ.Terms · domain: concord · modifier · participle · syntactic grouping. Limit: matching endings constrain a parse but do not by themselves prove a unique dependency.nominative agreement §267 passive participle §282a genitive pair §300bAI-assisted source-checked · no human review
Build construction and case relationships.Bracket the genitive absolute, then read the main passive: nakulaiḥ is the instrumental agent and nominative sutāḥ the patient.पश्यतो बकमूर्खस्य नकुलैर् भक्षिताः सुताः।paśyato bakamūrkhasya nakulair bhakṣitāḥ sutāḥ.Terms · domain: genitive absolute · passive agent · patient · clause structure. Condition: Whitney limits this absolute construction to the later language and a living genitive actor, usually with a participle of seeing, hearing or uttering.instrumental agent §278 passive construction §282a genitive absolute §300bAI-assisted source-checked · no human review
Translate in two explicit passes.First preserve the parsed relations; then produce natural English without silently changing agent, patient, time relation or degree of certainty.पश्यतो बकमूर्खस्य नकुलैर् भक्षिताः सुताः।paśyato bakamūrkhasya nakulair bhakṣitāḥ sutāḥ.Terms · domain: literal rendering · natural translation · interpretation. Limit: the two-pass method is project pedagogy; Whitney supplies the source witness and translation evidence, not these workflow instructions.passive relation §282a source and translation §300bAI-assisted source-checked · no human review

Source-attested interactive line · select a word

।

paśyato bakamūrkhasya nakulair bhakṣitāḥ sutāḥ.

Relation-preserving: “Of the foolish heron looking on, the young were eaten by ichneumons.” Natural: “While the foolish heron looked on, its young were eaten by ichneumons.”

Sandhi §174 §175a passive §282a exact witness §300bSource-attested witness · AI-assisted source-checked · no human review

Your sentence

Sentence breakdown

Paste or type a sentence, or load a plain-text file. The built-in analyzer proposes endings and verb roots; ambiguous forms remain visibly marked.

This is a transparent morphology aid, not a full statistical parser: sandhi and context can produce more than one valid analysis.

पदच्छेदः

Your word-by-word analysis will appear here.

Reading Lab · Module 03

Recitation and Practice

Generate a Sanskrit-aware reference reading for Classical, non-Vedic text, then practise the same line aloud.

Reference chant

Vāgdhenu connection

A configured pronunciation service uses the Vāgdhenu model to generate the reference chant.

Sanskrit verse or sentence in IAST or Devanāgarī

Open Vāgdhenu demo ↗

Checking the pronunciation backend…

श्रु

Listen analytically

Track vowel length, aspiration, retroflexion and mora timing. Repeat in short groups before attempting the complete line.

वद्

Practice aloud

Vāgbodhinī/Su-śrotā supplies Sanskrit speech-recognition research tools for guided practice. Its full scoring model remains a separate service.

Open the Vāgbodhinī project ↗

Scope: Vāgdhenu is intended for metrical, non-Vedic Sanskrit chant. Vedic recitation needs accent marks, a named śākhā or recitation tradition, and a dedicated model; it is never silently treated as Classical Sanskrit.

Vedic · Classical Sanskrit

Literature

Move from source-planned collections to annotated texts whose words can be inspected line by line.

Literature · Corpus

Corpus Status

See which source-recorded readers are present and which collection areas remain unavailable. This page reports status only; it does not change the adopted corpus.

Vedic

Available now

Ṛgveda Saṃhitā

All ten maṇḍalas use VedaWeb’s accented University of Zurich text and morphology. The public-domain Griffith translation is supplied through VedaWeb’s stanza alignment.

1,028 hymns10,552 stanzasVedic accent preserved

Open reader →

Currently Unavailable

Principal Upaniṣads

Name each Upaniṣad and adopted recension separately. Record whether its source passage is prose or verse before designing its reader display.

Classical Sanskrit

Available now

Bhagavad Gītā

The Bhagavad Gītā reader retains aligned Devanāgarī, IAST, translation and word analysis.

Open reader → Available now

Mahābhārata

The Mahābhārata reader retains aligned Devanāgarī, IAST, translation and word analysis.

Open reader → Available now

Rāmāyaṇa

All seven kāṇḍas use DCS word analysis of the Baroda critical text. Secure Sanskrit matches add M. N. Dutt’s public-domain English translation without hiding recension differences.

606 sargasPublic-domain translation

Open reader →

Currently Unavailable

Selected Sūtras

This is a collection slot, not a single text. Each adopted sūtra work will receive its own source and coverage record.

Currently Unavailable

Selected Śāstras

This is a collection slot, not a single text. Each śāstra will be scoped and sourced independently.

Currently Unavailable

Selected Purāṇas

This is a collection slot, not a single text. Each Purāṇa will require a named edition and an explicit coverage boundary.

Annotated corpus

Read Sanskrit with the Grammar Beside It

Select a reader and chapter, then tap any word for its source-recorded lemma and morphology. Each stanza or verse includes an aligned English translation where the adopted source supplies one.

Book Chapter Go to line

Loading the annotated corpus…

महाकाव्यम्

Loading the annotated reader…

Vedic and critical texts, analysis and translations

The Ṛgveda uses VedaWeb 1.0’s accented University of Zurich text and morphosyntactic annotation under CC BY 4.0. VedaWeb identifies this as a version by A. Lubotsky annotated at the University of Zurich and modified for the 2020 digital edition. Its exact accented Latin tokens are preserved; the synchronized unaccented Devanāgarī is mechanically generated and clearly labelled as a reading aid.

The Mahābhārata follows the Bhandarkar Oriental Research Institute Critical Edition in Muneo Tokunaga’s electronic recension. The Rāmāyaṇa follows the Oriental Institute of Baroda Critical Edition in Tokunaga’s electronic transcription as represented in Oliver Hellwig’s Digital Corpus of Sanskrit (DCS). Interactive surface forms and word-level analysis come from DCS under CC BY 4.0. GRETIL’s Tokunaga–Smith files are used only for scholarly comparison under their posted reference-only terms.

Ṛgveda translations use Ralph T. H. Griffith’s public-domain The Hymns of the Rigveda (1890); VedaWeb’s specific transcription and stanza alignment are CC BY-NC-SA 4.0. Bhagavad Gītā translations use the Unlicense gita/gita data and Shri Purohit Swami’s English rendering. Rāmāyaṇa and other Mahābhārata translations use M. N. Dutt’s public-domain English through the published Itihāsa corpus, only where the Sanskrit content matches the adopted critical text.

VedaWeb 1.0 TEI dataset ↗ · VedaWeb Ṛgveda ↗ · Zurich text licence ↗ · Griffith alignment licence ↗ · Digital Corpus of Sanskrit ↗ · Itihāsa Dutt alignments ↗

संख्या · saṃkhyā

Numbers

Read the digits, then learn the words that count and order.

Search · Convert

Number Converter

Enter Arabic or Devanāgarī digits, English, IAST, or Harvard–Kyoto transliteration. The converter accepts values from 0 to 100,000,000.

Enter a number to see every form together.

Numbers · Module 01

Cardinal and Ordinal Numbers

Cardinals answer “how many?”; ordinals answer “which in order?” The table keeps English, Devanāgarī and IAST together.

How cardinals 1–4 inflect

eka is singular and declines like a pronominal adjective. dvi is dual only: masculine dvau, feminine and neuter dve. tri and catur are plural in ordinary counting and have distinct masculine, neuter and feminine forms, shown in the table.

How ordinals inflect

Ordinals are adjectives agreeing with their noun in case, number and gender. Masculine and neuter forms follow the ordinary a-stem adjective patterns. prathama, dvitīya and tṛtīya form their feminine in -ā; caturtha and later ordinals ordinarily form it in -ī.

How ordinals are written

Devanāgarī has the digits ०–९, but no separate ordinal-digit signs corresponding to English 1st, 2nd or 3rd. Sanskrit normally writes the ordinal adjective; the new abbreviated-form column is therefore an English reference aid.

Whitney §482 · §486a · §§487–488

Arabic digit	Devanāgarīdigit	Cardinal			Ordinal
Arabic digit	Devanāgarīdigit	English	Devanāgarī	IAST	1st, 2nd…	English	Devanāgarī	IAST

Forms above twenty are built regularly but show several competing conventions in actual texts. This core table includes 0–20, the tens, one hundred and one thousand.

Source-attested reference

Cardinals and Ordinals in Sentences

Bold type marks the agreeing numeral or ordinal. Translations are concise project translations, not wording attributed to the source editions.

Gendered cardinals 1–4

माम् एकं शरणं व्रजmām ekaṃ śaraṇaṃ vraja

Go to me alone for refuge.

Bhagavad Gītā 18.66 · ekaṃ agrees with masculine mām.

द्वौ भूतसर्गौ लोकेऽस्मिन्dvau bhūtasargau loke ’smin

There are two kinds of created beings in this world.

Bhagavad Gītā 16.6 · masculine dual.

गुणान् एतान् अतीत्य त्रीन्guṇān etān atītya trīn

Having gone beyond these three qualities.

Bhagavad Gītā 14.20 · masculine accusative plural.

महर्षयः सप्त पूर्वे चत्वारो मनवस्तथाmaharṣayaḥ sapta pūrve catvāro manavas tathā

The seven ancient seers, and likewise the four Manus.

Bhagavad Gītā 10.6 · masculine nominative plural.

Ordinal adjectives 1st–4th

कर्ण इत्यभिविख्यातः पृथायाः प्रथमः सुतःkarṇa ity abhivikhyātaḥ pṛthāyāḥ prathamaḥ sutaḥ

Karṇa, renowned by that name, was Pṛthā’s first son.

Mahābhārata 1.61.89.2 · masculine nominative singular.

द्वितीयः शलभस्तेषाम् असुराणां बभूव यःdvitīyaḥ śalabhas teṣām asurāṇāṃ babhūva yaḥ

The second of those Asuras was Śalabha.

Mahābhārata 1.61.29.1 · masculine nominative singular.

तृतीयस्तु महाराज महाबाहुर्महासुरःtṛtīyas tu mahārāja mahābāhur mahāsuraḥ

The third, O king, was a mighty-armed great Asura.

Mahābhārata 1.61.48.1 · masculine nominative singular.

तेषाम् अन्यतमो यस्तु चतुर्थः परिकीर्तितःteṣām anyatamo yas tu caturthaḥ parikīrtitaḥ

Another of them is named as the fourth.

Mahābhārata 1.61.49.1 · masculine nominative singular.

शब्दाः · śabdāḥ

Core Vocabulary

Learn what the corpus uses most, with form, function and meaning together.

Vocabulary · Module 01

1,000 Most Frequent Sanskrit Lemmas

Corpus rank comes from the Digital Corpus of Sanskrit frequency layer. Search in English, IAST, Devanāgarī or SLP1, then select any table heading to sort by rank, Sanskrit dictionary order, part of speech, English meaning or attestation count; select it again to reverse the order.

Find a word

Reorganize this table: select any column heading to sort, then select it again—or use the arrow—to reverse the order. Part-of-speech sorting groups categories alphabetically and alphabetizes Sanskrit headwords within each category. Export downloads the current filtered and sorted rows.

Loading 1,000 corpus-ranked lemmas…

One thousand most frequent Sanskrit lemmas in the Digital Corpus of Sanskrit

Source, calibration and limits

Rank, count and grammatical tag reproduce the first 1,000 records of kosha data-v0.1.0, a frozen Digital Corpus of Sanskrit frequency snapshot. Concise English glosses are project-authored summaries or short extracts linked to the bundled Cologne Monier–Williams XML edition; they are not quotations from Pāṇini, Vasu or Monier–Williams and are not context-sensitive translations.

Schema 3.0 preserves the previously calibrated first 500 rows byte-for-byte and source-links the added ranks 501–1,000. In all, 923 rows link to exact MW records and 77 retain their exact corpus fields while showing an explicit warning because no complete POS-compatible learner sense passed the gate. The added rows are machine-checked, not row-by-row editorially calibrated; none of the 1,000 rows has independent human Sanskritist review. AI assistance: OpenAI Codex, GPT-5 model family, 23 August 2026. Corpus rank does not measure universal Sanskrit importance, and a headword gloss does not establish which sense a passage uses.

Research · Evidence

Sanskrit Research Tools

Move from a learner question to dictionaries, corpora and scholarly analysis.

Tools · Official Gateways

Trusted Resources for Sanskrit Research

Every link below is maintained by a university, library, government body, nonprofit research organization or the resource’s official project team.

Use more than one tool: automated analyses are hypotheses, and corpus coverage varies. Confirm important readings against a named edition, dictionary entry and grammatical context.

01

Dictionaries and Lexica

Check headwords, senses, citations and historical lexicographic evidence.

University of Cologne

Cologne Digital Sanskrit Dictionaries

Search digitized Sanskrit lexica, consult page scans and access documented downloadable data.

Open dictionaries ↗ University of Chicago

Digital Dictionaries of South Asia

Search the Sanskrit dictionaries of Apte and Macdonell within a university-maintained South Asian reference collection.

Open DSAL ↗ University of Hyderabad · external tool

Amarakośa Thesaurus

Explore Sanskrit synonym groups and semantic relations in the maintained Saṃsādhanī Amarakośa interface.

Open the thesaurus ↗ University of Hyderabad · external tool

Concordance of Kośas

Compare Sanskrit synonym sets and locations across Amarakośa and other kośa traditions in the university’s maintained concordance.

Open the concordance ↗ Sanskrit Documents · external files

Amarakośa Reading Files

Open the maintained Amarakośa text, PDF and related links at their source. Reposting restrictions mean these files are not mirrored here.

Open Amarakośa resources ↗

02

Corpora and Digital Texts

Locate passages, compare textual collections and work with machine-readable Sanskrit.

Göttingen State and University Library

GRETIL

Browse standardized machine-readable texts in Sanskrit and other Indian languages, with TEI and plain-text formats where available.

Open GRETIL ↗ Heidelberg University

Digital Corpus of Sanskrit

Search a large corpus of Sanskrit sentences with lexical and morphological annotation.

Open DCS ↗ Official SARIT Project

SARIT

Browse and search TEI-encoded Sanskrit and Prakrit texts, with project documentation for scholarly text encoding.

Open SARIT ↗ The Sanskrit Library

The Sanskrit Library

Use digitized primary texts, lexical resources and integrated research tools from a nonprofit digital library.

Open the library ↗ Sanskrit Documents · external index

Aṣṭādhyāyī Sortable Index

Browse sūtra text, padaccheda, rule type, terms, scope and traditional, alphabetical and Kaumudī order at the maintained source site.

Open the index ↗ Aṣṭādhyāyī.com · proofread text

Vasu English Commentary

Read the complete rule-linked English translation and Kāśikāvṛtti-based commentary in a searchable interface, with links back to Vasu’s scans.

Open Vasu English ↗ Internet Archive · external facsimiles

Vasu’s Aṣṭādhyāyī · Books I–VIII

Check Vasu’s 1891–1898 translation and commentary against all eight book scans. The public-domain wording is distinct from the hosted compilation’s later formatting rights.

Open the eight-book witness ↗ Internet Archive · copyrighted reference

Sharma’s Aṣṭādhyāyī · Volumes I–VI

Consult a modern translation, commentary, explanatory notes and derivational histories. Sharma remains all rights reserved, so use exact page citations and original paraphrase rather than copied prose.

Open Sharma’s commentary ↗

03

Parsers and Analysers

Test segmentation, morphology, sandhi, compounds, syntax and metre.

INRIA

Sanskrit Heritage Platform

Use the Heritage dictionary and Sanskrit Reader for segmentation, sandhi splitting, morphological tagging and parser-assisted analysis.

Open Sanskrit Heritage ↗ University of Hyderabad

Saṃsādhanī

Access transliteration, lexicons, morphology, sandhi tools, compound analysis, parsing and specialist e-readers.

Open Saṃsādhanī ↗ The Sanskrit Library

Sanskrit Library Tools

Use transcoders, paradigm generators, morphological analysers and a Sanskrit metre-identification tool.

Open analysis tools ↗ Sanskrit Documents · external collection

Sanskrit Learning Tools

Use source-hosted dictionaries, converters, indices, metres, grammar aids and related study tools without reproducing restricted files locally.

Open learning tools ↗

04

Philology and Research Collections

Extend a language question into comparative, Vedic and cultural research.

Goethe University Frankfurt

TITUS

Explore Indo-European text and language materials, including Sanskrit and Vedic corpora, lexica and retrieval tools.

Open TITUS ↗ IGNCA · Ministry of Culture, India

Vedic Heritage Portal

Consult institutionally curated Vedic texts, recitation recordings, ritual documentation and related reference material.

Open Vedic Heritage ↗ McComas Taylor · source reference

The Little Red Book of Sanskrit Paradigms

The site’s Sanskrit paradigm and Sandhi tables are derived from this 2009 sixth edition, produced and generously shared by McComas Taylor.

Open Taylor’s PDF ↗

Tools · Publication record

Rights, Licences and Reuse

This page records the publication policy, attribution, licence and local changes for the principal source layers bundled into Sanskrit Study Companion.

Release record

Release & source versions

The site and its source datasets are versioned separately. A new site release does not silently change a pinned dictionary, corpus or historical-text witness.

Website: 2026.08.27Current public site release.
Dictionaries: 2014 XML editionsMonier–Williams and Apte digital editions from The Sanskrit Library and Cologne project.
Vocabulary & verbs: Core schema 3.0.0 · Verb schema 2.0.0kosha data-v0.1.0; 1,000 vocabulary rows and 501 verb-reference entries.
Pāṇini: Sanskrit Library 0.1 · index snapshot 2026-08-11Separate complete Vasu historical-layer snapshot 2026-08-22.
Readers: Whitney 2026-08-11 · DCS reader 2 · VedaWeb 1.0Whitney uses 25 pinned Wikisource revisions; DCS and VedaWeb retain their own recorded source scope.

CC BY-NC-SA 3.0

Monier–Williams and Apte digital editions

The bundled Monier–Williams 1899 and Apte 1890 browser dictionaries use the 2014 XML editions published by The Sanskrit Library and the Cologne/University of Cologne project. The source headers record copyright © 2014 The Sanskrit Library and Thomas Malten.

Local changes: source records were reindexed and split into browser chunks; source encoding is converted for IAST and Devanāgarī display; entries are linked to local routes; and selected lexical fields, short extracts and source-grounded summaries appear in learner references. Those source-derived adaptations are distributed under CC BY-NC-SA 3.0. No source party endorses this project.

Full digital-edition credit

The Monier–Williams XML edition names Jim Funderburk, Thomas Malten and Peter M. Scharf as editors and credits Sharon Ben-Dor, Pawan Goyal, Malcolm D. Hyman, Susan J. Moore, Ramaswamy Chandrashekar, Wendy Teo, Sampada Savardekar Thomas and Jason Lyle Reeder for specified digital work. The Apte XML edition names Jim Funderburk, Thomas Malten and Peter M. Scharf as editors and Sampada Savardekar Thomas for correction work. The underlying printed works name Monier Monier-Williams with Ernst Leumann and Carl Cappeller, and Vaman Shivram Apte, respectively.

Monier–Williams source ↗Apte source ↗CC BY-NC-SA 3.0 ↗

CC BY-NC-SA 3.0

Pāṇini source index

Pāṇini, Aṣṭādhyāyī as in the Kāśikā: First XML edition, edited by Peter M. Scharf, version 0.1 (The Sanskrit Library, 2010). Data entry, XML markup and TEI template are by Peter M. Scharf; copyright © 2010 The Sanskrit Library.

Local changes: the source readings were adapted into a searchable browser index with mechanically generated IAST and Devanāgarī reading layers and local record links. This derived source-index layer remains CC BY-NC-SA 3.0. The separately sourced public-domain Vasu layer and modern project explanations retain their own stated status.

Sanskrit Library record ↗CC BY-NC-SA 3.0 ↗

CC BY-NC-SA 4.0 · CC BY 4.0

Ṛgveda text and Griffith alignment

The displayed accented text, lemmas and morphology use the University of Zurich Rigveda source in VedaWeb 1.0, licensed CC BY 4.0. The underlying 1890 Griffith translation is public domain; VedaWeb’s 2020 transcription and stanza alignment are CC BY-NC-SA 4.0.

Attribution and changes: VedaWeb 1.0 was prepared by Francisco Mondaca and Patrick Sahle. Its Zurich source identifies A. Lubotsky’s text and annotation work subsequently modified by Jakob Halfmann and Natalie Korobzow. The project converted the TEI into hymn chunks, synchronized the source tokens with mechanically generated unaccented Devanāgarī and indexed the supplied Griffith alignments. The CC BY-NC-SA alignment-derived layer remains under CC BY-NC-SA 4.0.

VedaWeb 1.0 ↗CC BY 4.0 ↗CC BY-NC-SA 4.0 ↗

Other open source layers

Whitney, DCS and kosha

Whitney’s 1889 work and scan are public domain; the adopted Wikisource transcription and this project’s transcription adaptations remain CC BY-SA 4.0. Oliver Hellwig’s Digital Corpus of Sanskrit data are CC BY 4.0. The pinned kosha frequency release is CC BY-SA 4.0.

Local changes: the Whitney transcription was cleaned of site navigation, split into local reading chunks, linked to local routes and indexed for search. DCS data were converted into reader chunks and lemma links; kosha rows were selected and displayed with their pinned rank, count and grammatical fields. Each adapted layer keeps its named source licence.

Whitney transcription ↗CC BY-SA 4.0 ↗DCS source ↗kosha release ↗

Public domain and project layers

No rights are claimed over public-domain source works

Public-domain texts, translations and historical explanations remain public domain. Project-authored site code, artwork and teaching prose are not placed under a blanket Creative Commons licence by this notice. Where project contributions form part of an adaptation governed by a source’s ShareAlike licence, that source-specific licence applies to the adaptation as stated above.

This notice describes the adopted release policy and source attributions; it is not legal advice or a warranty that every possible reuse is permitted.

Contact · Feedback

Leave a Message

Questions, corrections and thoughtful suggestions are welcome.

Contact · Message

Feedback or enquiries

For feedback or enquiries, please leave a message. You may include an email address if you would like a reply.

Support the project

dhanyavāda धन्यवाद

This independent study companion is being built as a careful, source-aware resource. Voluntary support will help with source access, research time and continued development.

Support on Ko-fi Voluntary support is accepted through Ko-fi.

शब्दकोशः · Śabdakośa

Sanskrit Dictionaries

Two field-standard Sanskrit–English references, searchable in this site.

Dictionary · Monier–Williams 1899

Monier–Williams Sanskrit–English Dictionary

Search 193,890 digitized entry records by English definition or Sanskrit headword. Sanskrit may be entered in IAST, Devanāgarī, plain Latin or the source edition’s SLP1 spelling.

अ

Search Monier–Williams

Select a result to read the complete digitized entry.

Digital data: Cologne Digital Sanskrit Dictionaries and The Sanskrit Library, CC BY-NC-SA 3.0. Edition and attribution details.

Dictionary · Apte 1890

Apte’s Practical Sanskrit–English Dictionary

Search 34,882 digitized entry records by English definition or Sanskrit headword. Sanskrit may be entered in IAST, Devanāgarī, plain Latin or the source edition’s SLP1 spelling.

अ

Search Apte

Select a result to read the complete digitized entry.

Digital data: Cologne Digital Sanskrit Dictionaries and The Sanskrit Library, CC BY-NC-SA 3.0. Edition and attribution details.

Dictionary · Research gateways

Texts, Lexica and Corpus Data

Use these gateways to verify a dictionary form, find a text or move from a learner table into primary evidence. Availability and redistribution rights vary by collection.

Lexica

Learn Sanskrit Online with Sanskrit Study Companion

Choose a Subject

Devanagari

Simple Characters

Vowels

Consonants

Common Devanāgarī Signs

Find a Conjunct

Complete Conjunct Table

Search the Script

Word Converter

Pronunciation Laboratory

Phoneme chart

Vowels

Consonants

Words and short phrases

Mora timing

Sandhi

Vowels Inside a Word

Complete internal vowel sandhi: simple first vowel

Complete internal vowel sandhi: complex first vowel

Same-quality vowels become long

a or ā before i, u, ṛ or ḷ

a or ā before e, ai, o or au

i, u, ṛ or ḷ before a different vowel

e, ai, o or au before a vowel

The chart gives a default—not one result for every context

Consonants in Close Contact

Two verified paths, plus one immediate blocker

Aspiration moves to an ending beginning in t or th

A dental shifts to the matching palatal in close contact

Families outside this bounded lesson

Vowels at a Word Boundary

Sandhi Rules: Vowels

Consonants at a Word Boundary

Stops

Visarga

Nasals

Sibilants

Final stops: voicing and place assimilation

Keep the corresponding voiced stop before a voiced sound

Visarga after a or ā

Visarga after i, u, ṛ—and before sibilants

Final m and anusvāra

The Little Red Book Consonant Chart

Sandhi Rules: Consonants

A reliable order for solving consonant sandhi

Which Words Join in Devanāgarī?

One written word or a word boundary?

Join the pieces of one inflected word

Join a lexical compound

Apply external sandhi between independent words

Read continuous manuscripts cautiously

“The people dispersed”

Put a Boundary Together

Test Possible Underlying Boundaries

Three Questions Organize Sandhi

Boundary position

Sound group

Rule family

1 · Vowels merge or become glides

2 · Consonants adjust in contact

3 · Visarga responds to the next sound

How to Work Through Sandhi

Verbs

Verb Roots

Forming guṇa and vṛddhi from zero grade (inserting a/ā BEFORE root vowel)

Forming guṇa and vṛddhi from zero grade (inserting a/ā AFTER root vowel)

Vocabulary examples: a/ā inserted BEFORE the root vowel

Vocabulary examples: a/ā inserted AFTER the root vowel

Other lexical and irregular root-grade series

Rule Examples

Verbs and Their Grades

Irregular and Lexical Verb Roots

Verb Root Classes and Paradigms

How each class builds its present stem

Finite and Non-Finite Paradigms

√गम् · √gam

Complete Verb Charts

How Preverbs Build a Complex Verb