I write poetry in Hindi. Not professionally and not prolifically, but consistently enough over enough years to have noticed something that reading poetry alone would never have taught me: when I try to carry one of my own poems into English, the part that survives is the part I cared about least.
The statement survives. The image usually survives, a bit dulled. What goes is the sound, and with the sound goes most of the reason the poem existed. That experience is the most useful thing I know about how poems work, and it is where I would start anyone who wants to write them.
The line is the unit, not the sentence
Prose is built out of sentences. Poetry is built out of lines, and the two do not have to agree.
This is the first thing that separates a poem from a paragraph with the returns pressed in early. A line break is a piece of punctuation with no equivalent in prose. It creates a pause the grammar did not ask for, and it puts weight on whatever word is left hanging at the end of the line and whatever word is exposed at the start of the next.
Which means that where you break is a decision with consequences. Break on a noun and the noun gets held up for inspection. Break mid-phrase, so the sentence has to reach across the gap, and you get tension and a small acceleration when the reader arrives at the resolution. Break at every grammatical boundary and the poem becomes a list of clauses, technically arranged as verse and doing none of verse's work.
A blunt diagnostic: read your poem aloud, pausing very slightly at every line end. If those pauses add nothing, or worse, fight the sense for no reason, the lineation is decorative and the poem is prose in costume.
Sound is language-specific, and this is not a small point
English verse organises itself around stress. The pattern that feels natural, the one iambic pentameter formalises, is an alternation of lighter and heavier syllables, and English rhythm is essentially about where the beats fall.
Hindi verse counts differently. The traditional prosody is moraic: syllables are weighed as light or heavy according to vowel length and what follows, and a line is measured by the total weight rather than by a count of beats. Urdu and Hindi poetry inherits from that tradition a set of forms in which the pleasure is partly arithmetic — the ghazal, with its repeating end-word and the rhyme that lands just before it, sets up an expectation the ear starts predicting two couplets in, and then satisfies.
None of that transfers. When I move a Hindi line into English, the weights are meaningless, the repetition that felt inevitable sounds like a tic, and the compound words that carried three ideas in four syllables become a limp English phrase of nine. What is left is the paraphrase. This is not a complaint about English; the reverse is just as lossy. It is an argument that the music is not an ornament on top of the meaning. In a poem that is working, the music is a substantial fraction of the meaning, and you only discover how much when you try to remove it.
Practically, this means the first thing to learn about the language you are writing in is how it makes rhythm, and the way to learn it is to read a great deal aloud rather than to study a metrics chart. Alliteration, assonance and consonance are usually taught as devices to be sprinkled on. They are better understood as the ways sound holds a line together when the rhythm is doing something irregular.
Image, and the argument about statement
The most durable piece of advice in the English tradition is roughly: prefer the thing to the comment on the thing. Ezra Pound's two-line "In a Station of the Metro" is the standard demonstration — faces in a crowd, petals on a wet black bough, and no instruction about how to feel.
I would offer it as a strong default rather than a law, because it is a law that would exclude a lot of poetry worth having. Plenty of poems in English state their feelings outright and survive it; a great deal of devotional poetry in every language I can read is direct assertion and is not weakened by it. What the image rule is really guarding against is the abstract adjective doing the work in place of anything observed — the poem that says a thing was beautiful, or painful, or profound, and gives you nothing to be moved by.
The reliable version of the advice is therefore narrower. Give the reader something specific enough that they could point at it. Then decide whether the poem also needs you to say what it meant. Sometimes it does.
Form, and the haiku misunderstanding
Constraint is not the enemy of expression. It is usually the reason a poem finds a phrase the writer would not otherwise have reached, because the requirement blocks the first and most obvious move.
The sonnet is the durable English example. Fourteen lines, arriving in English from Italian and in continuous use for centuries, with an internal architecture that sets something up and then turns against it. The turn is the point; the fourteen lines are just the room it happens in.
The haiku is the form most often taught wrongly. The 5-7-5 rule that everybody learns in school is a mistranslation of the Japanese count, which is measured in a smaller unit — not syllables in the English sense but shorter sound-units — so an English poem of seventeen syllables is generally noticeably longer and baggier than the Japanese original it is imitating. Most serious haiku written in English uses fewer syllables and concentrates instead on the features that actually matter: two images set beside each other with a cut between them, present tense, a season, and no explanation. If you want to practise compression, that is the version to practise.
The point of any of these is not the credential of having written one. It is that working inside a fixed shape teaches you what your own default habits are, because the shape will not permit them.
Revision, and the honest ratio
Almost nothing arrives finished. The draft is a way of finding out what the poem is about, and the poem is what you build afterwards from the two or three lines that were alive.
The most useful revision I do is subtractive, and the most common thing I cut is the line that explains. Poems tend to accumulate a final couplet that tells the reader what to take away. Removing it is almost always an improvement, and almost always feels like a loss for about a day.
Reading aloud catches what the eye forgives: the clumsy consonant cluster, the line that has one syllable too many, the rhyme that arrived because it was available rather than because it was right. If you have a language you speak but rarely write in, try reading the draft in your own accent rather than a neutral one; the rhythm you actually hear is the one you should be writing to.
Sending it out
At some point people want to publish, and the arithmetic of that is worth understanding before it demoralises you.
Rejection is not the exception in this field, it is the ordinary condition. The University of Arizona Poetry Center maintains a guide to reading rejection letters, which tells you something in itself: the volume is high enough, and the gradations between a form rejection and an encouraging one are subtle enough, that learning to read them is a genuine skill. Competitive journals decline the overwhelming majority of what they receive, and a great deal of what they decline is perfectly good.
None of which is a reason not to submit. It is a reason not to treat an acceptance as the verdict on whether the poem was worth writing. Most of the poems I have written will never be seen by anyone, and they were still the thing that taught me how to write the next one.
That is the actual practice. Read aloud in the language you are writing in, until its rhythm is in your ear. Write the thing you can point at. Cut the line that explains. And if you have more than one language, translate yourself occasionally, because nothing else will show you so quickly what you were relying on.
Tags
Taresh Sharan
support@sharaninitiatives.com