How to turn any Persian text into a reading lesson.

Researched and edited by VowelMarks · Last reviewed August 29, 2026

A translation question is not automatically a reading exercise

Imagine a learner sees a Persian sentence, pastes it into a translator, and copies the English answer. The meaning may be correct, but the learner never had to notice the short vowel, follow an Ezafe connection, identify the verb, or decide which words in the sentence support that meaning. The task tested access to a translation, not reading.

A stronger question changes what the learner does with the source. It may ask for a reading, a grammatical relationship, a meaning that fits this context, a concise paraphrase, or the exact phrase that proves an answer. In every case, the learner should have to look back at the Persian rather than escape from it.

VowelMarks uses 15 families, grouped into four readable jobs

The internal taxonomy is specific because Persian reading problems are specific. The public interface keeps the choices simpler: mixed practice, reading and pronunciation, or meaning and language. Behind those choices, the planner can select from the families below when the source contains enough evidence.

These are VowelMarks product families, not official PIRLS or PISA labels. Broader reading frameworks distinguish processes such as retrieving stated information, making an inference, integrating ideas, and evaluating evidence. VowelMarks adds Persian-specific decoding and pronunciation work beside those comprehension processes because a learner can understand the topic of a sentence and still be unable to read it aloud.

Not every passage deserves every kind of question

A two-line family message may contain a useful Ezafe chain and one ambiguous spelling, but it may not support a serious tone analysis. A public notice may support register, instructions, dates, and exact evidence, but no colloquial-formal contrast. Forcing the same checklist onto both texts produces filler.

One Persian passage, several different reading jobs

The passage below is an original VowelMarks example. It is short enough to inspect, but rich enough to support pronunciation, Ezafe, object marking, vocabulary, and comprehension.

The questions are not interchangeable. The first asks the learner to recover a missing sound. The second follows a Persian phrase connection. The third tests a grammatical marker. The fourth requires proof from the passage. The last asks the learner to integrate several actions without copying the full source.

A reusable template for writing a better question

You can use this template by hand for a class reading, family message, news paragraph, or story. It also explains what a generation system has to decide before wording a question.

1. Name one objective. For example: recognize the Ezafe in one noun phrase.
2. Copy the exact evidence. Keep the complete phrase or sentence that makes the answer defensible.
3. Write the answer before the prompt. If the answer is vague, the question will be vague too.
4. Choose the response form. Use multiple choice for a clean distinction and writing for paraphrase, explanation, or more than one acceptable wording.
5. Build distractors from real mistakes. A wrong option should reflect a plausible vowel, boundary, grammar, or meaning error, not nonsense.
6. Explain the deciding clue. Feedback should say why this answer fits the Persian and what the learner can reuse next time.
7. Return to the passage. After feedback, reread the original line without the answer in view.

Why answer first and feedback second?

Retrieval-practice research distinguishes trying to bring an answer to mind from simply seeing it again. Kang, Gollan, and Pashler found benefits when learners attempted foreign-word retrieval before feedback. Li and colleagues likewise studied retrieval with corrective feedback in additional-language vocabulary learning. These findings support asking for a real attempt before revealing the explanation.

They do not prove that every quiz format is useful, or that one correct answer means a learner has mastered a Persian pattern. The practical consequence is narrower: require a commitment, give prompt feedback, and let the reader use that feedback in another encounter with the text.

Hints should preserve the work, not do all of it

A learner who is stuck on pronunciation may need vowel marks, Pinglish, or speech. A learner who understands the sound but not the sentence may need meaning or a grammar explanation. Those are different problems. A useful hint solves the smallest blocking problem and leaves the learner with something meaningful to retrieve.

This is also why VowelMarks keeps marks, Pinglish, meaning, and Listen as separate help choices inside a Practice session. The support can remain connected to the same source without turning the whole task into an English translation.

How VowelMarks builds a source-grounded set

VowelMarks does not begin by asking a model to invent a quiz from an undifferentiated block of text. The system first plans a learning objective, an exact source occurrence, the supporting quote, the question family, and the order of the lesson. That plan keeps a larger set distributed across the passage instead of clustering every question around the first paragraph.

AI then writes the prompts, plausible choices, written-answer rubrics, and teaching feedback that require language authoring. Reading questions keep a trusted pronunciation target outside the authored response. Exact evidence quotations must occur in the original source. Duplicate prompts, invalid answer structures, or a changed trusted reading can cause the set to be repaired or rejected instead of silently saved.

What Free proves and what Pro is for

A Free account can experience complete source-grounded lessons, save the results, review the answers, and retake the same set without spending another generation. Pro is designed for making this a routine across many more texts. It adds deeper 15- and 20-question lessons, many more new sets, and more learning-history capacity so practice can continue across what the learner reads. The exact current allowances are listed on the pricing page.

Quality controls should not become a paid advantage. Correct answer validation, source evidence, and clear feedback have to belong to the lesson itself. The paid value is frequency, depth, continuity, and the time saved when real Persian keeps becoming usable practice.

The evidence boundary matters

The research cited here supports design principles such as retrieval before feedback, corrective feedback, active engagement, and practice attached to meaningful input. It does not prove that VowelMarks improves Persian fluency, that every generated question is pedagogically optimal, or that the internal scoring represents mastery. Those claims would require outcome studies, live provider sampling, Persian-reader review, and learner data collected for that purpose.

The narrower claim is still valuable: a reading lesson is more defensible when every question has a named objective, exact evidence, a supported answer, and feedback that returns the learner to the Persian source.

Frequently asked questions

Sources and review notes

Sources are listed for the claims they support. Original practice examples are identified in the article. Product capabilities and access terms were checked on the review date and can change.

1. Kang, Gollan, and Pashler, Don’t Just Repeat After Me. Used for the finding that attempting retrieval before feedback can support later foreign-word comprehension and production. The study did not test VowelMarks or Persian reading lessons.
2. Li et al., Retrieval Practice With Corrective Feedback in Additional-Language Vocabulary Learning. Used to support immediate corrective feedback after a learner commits to an answer.
3. Uchihara, Webb, and Yanagisawa, second-language incidental vocabulary learning meta-analysis. Used for the value of keeping vocabulary work attached to meaning-focused input rather than separating every word into an isolated list.
4. Chi and Wylie, The ICAP Framework. Used to distinguish passive exposure from active answers, constructive explanations, and interactive discussion. It is a general learning framework, not a validation of this product.
5. IEA, PIRLS 2021 Assessment Frameworks. Used to cross-check broad comprehension processes such as retrieving stated information, making inferences, interpreting ideas, and evaluating textual elements. VowelMarks is not a PIRLS assessment.
6. University of Texas at Austin, Persian Online, The Writing System. Used to ground the Persian-specific problem of short vowels normally being absent from ordinary writing.
7. Doostmohammadi, Nassajian, and Rahimi, Persian Ezafe Recognition Using Transformers. Used to verify the linguistic importance of Ezafe and the fact that it is commonly absent from written Persian.