Conjuncts with half-forms and stacked shapes

When Devanagari writes consonants without a vowel between them, the first consonant is followed by virama ्. A font can display the resulting cluster in several ways: as a half-form beside the next consonant, as a compact ligature, or with one component stacked above or below another. These appearances are different solutions to the same structural problem, not three different grammatical categories.

English readers are used to letters keeping one shape inside every cluster. Devanagari instead reshapes consonants to show that no inherent a intervenes. The essential skill is therefore reversible parsing: see a complex glyph, recover its consonants, then identify the whole word and its morphology. Detailed cluster phonetics belongs on the syllable structure page.

Half-forms: the productive default

Many consonants have a full-height vertical stroke on the right. As the first member of a cluster, that stroke often disappears. The reduced consonant sits to the left of the following full consonant.

In मुख्य, there is no vowel between ख् and य. In व्यक्ति, both व्य and क्त have to be parsed. Reading vayakati by assigning an a to every visible consonant ignores the viramas and the lexical word.

Compact and stacked shapes

Not every consonant can form an obvious left-hand half-form. Fonts may combine the components more radically or stack them. Doubled retroflex stops are especially recognizable: ट् + ट → ट्ट and ड् + ड → ड्ड. Other sequences such as द् + ध → द्ध, ह् + न → ह्न, and ह् + म → ह्म can also receive distinctive forms.

Different fonts can render the same character sequence with more or less stacking. A mobile font may expose a small virama where a print font uses a tight ligature. Neither is a different spelling. Do not add spaces or invisible controls merely to make one font imitate another.

A minimal spelling contrast: पता and पत्ता

The pair पता patā “address/knowledge” and पत्ता pattā “leaf” isolates a single versus doubled consonant. In पत्ता, the first त carries virama and joins the second: त् + त → त्त. The difference is lexical and audible, not decorative.

The same principle applies to imperative सजा sajā “decorate” and learned noun सज्जा sajjā “decoration” (formal). A conjunct may distinguish words, grammatical forms, or registers; its meaning comes from the lexical item, not from stacking itself.

Vowel signs belong to the cluster's final consonant

A following vowel sign is encoded after the cluster, although pre-base ि displays before the cluster's base consonant. Compare क्ति in शक्ति, where ि follows the final त logically, with ट्ठी in चिट्ठी, where ी follows ठ. The screen position of a mark does not redefine the consonant order.

Here चिट्ठी contains ट् + ठ, not two independent syllables ṭa-ṭhī. The exact release and duration of such sequences vary; spelling securely gives the ordered consonants, while pronunciation must be learned with the word.

Morphology can create or preserve complex shapes

Conjuncts occur inside lexical roots, but grammatical suffixes also bring consonants together. लिख- + -ती → लिखती likhtī and समझ- + -ता → समझता samajhtā place the habitual ending directly after a consonant-final stem. The half-form is an orthographic consequence; gender and number are expressed by -ता, -ती, -ते, not by the conjunct.

Inflection can also follow an existing conjunct: व्यक्ति → व्यक्तियों, छुट्टी → छुट्टियाँ, and शब्द → शब्दों. The cluster remains part of the stem while the suffix supplies number and case. Parsing morphology therefore predicts where the dense glyph ends.

Unicode-safe display and actual reading

The safe sequence is always the logical one: consonant, U+094D virama, consonant, then any dependent vowel sign. ZWJ and ZWNJ can influence shaping in specialized typography, but they are invisible characters and are not ordinary Hindi spelling. Default educational and running text should not use them to force a half-form or stack.

Use genuine Devanagari code points throughout the sequence. A Latin y, l, or other lookalike is not a substitute for य, ल, or a Devanagari stroke, even if one font makes the mixed-script result look plausible.

Half-form recognition still does not settle all pronunciation. शब्द is śabd, not śabda; स्वागत is svāgat, with its internal and final vowel pattern learned as part of the word. A reader needs conjunct parsing, lexical knowledge, and schwa deletion together.

A short reading line

The line combines half-forms (व्य, स्त, ख्य, स्प, ष्ट, ब्द) with the compact doubled shape ट्ट. Read words and morphemes, not a procession of isolated alphabet names.

Common Mistakes

1. Inserting an a inside a half-form cluster

2. Breaking a cluster with a space

3. Losing a contrastive doubled consonant

4. Replacing a stacked sequence with full syllables

Key Takeaways

• Half-forms, ligatures, and stacks are renderings of consonant + virama + consonant sequences.
• The same Unicode spelling may look different across fonts without changing the word.
• Vowel signs follow the logical cluster even when they display to its left or around it.
• Morphology can create clusters at stem-suffix boundaries and helps readers separate dense shapes.
• Fluent reading combines glyph parsing with vocabulary and schwa deletion; it never assigns a mechanically to every consonant.