How Speedtype Writes and Reviews English and Marathi Passages

What is in the current passage library

Speedtype currently ships 24 first-party practice passages:

Language	Basic	Moderate	Advanced	Total
English	4	4	4	12
Marathi	4	4	4	12
Total	8	8	8	24

Each entry has an internal identifier, difficulty label, passage text, source label, review status, and review date. The passages are stored with the application rather than fetched from an unknown quote feed at typing time.

The number is intentionally stated as the current library size, not as a permanent promise. Entries can be corrected, replaced, or expanded after review.

Original text rather than unattributed quotations

The passage source is labeled Speedtype original. The content is written for the product instead of being copied from a book, news article, social post, or quotation database.

This choice serves two purposes. It avoids presenting someone else's writing without context, and it lets passages be designed around complete, ordinary prose rather than a famous line that is too short for a timed session.

Original authorship does not automatically make a passage good. The release checks still examine structure, duplication, review status, and minimum length before a production build is accepted.

Automated release checks

The passage validation script checks facts that software can verify consistently. Current release requirements include:

• every identifier is unique;
• every full passage is unique;
• opening text is not duplicated across entries;
• the language and difficulty fields use supported values;
• each passage contains at least 70 whitespace-separated words;
• each passage contains at least five sentence endings;
• required source and review fields are present;
• English entries are marked reviewed;
• Marathi entries cannot remain in draft;
• reviewed Marathi entries include a review date.

A release build fails if these requirements are violated. The checks prevent obvious content and metadata mistakes from silently reaching the site.

They do not judge whether a sentence is elegant, culturally natural, or pedagogically ideal. Those require human reading.

What the difficulty labels mean

Basic, Moderate, and Advanced are editorial groupings inside Speedtype. They are not standardized exams and are not calibrated to a population score.

A reviewer can consider sentence length, punctuation frequency, vocabulary, visual complexity, and the amount of input-method work required. The labels help a user select different passage styles within one language.

The labels should not be interpreted as equal across English and Marathi. A Marathi basic passage and an English basic passage do not establish equivalent motor or language demand. The language comparison guide explains why WPM should be tracked within a consistent script and input method.

English review

English entries are read for coherent subject matter, complete sentences, spelling, punctuation, and a natural progression rather than keyword repetition. They are also checked for the minimum structure required by the release script.

The review record currently associated with the library is dated August 5, 2026. That date indicates when the stored text received its recorded review. It does not mean the article you are reading or every future passage was reviewed on the same date.

If an English passage is revised substantially, its review metadata should be updated after another read rather than preserving an old date.

Marathi review

Marathi requires an additional language-specific checkpoint. Automated checks can count words and detect missing metadata, but they cannot reliably approve grammar, idiom, spelling choices, or natural phrasing.

The release workflow allows a Marathi entry to be marked as needing native review while work is in progress, but production release blocks a draft. The current 12 entries are marked reviewed with recorded review dates.

That status is a process record, not a guarantee that no typo can exist. A reader who notices an unnatural phrase can provide more useful feedback than a generic automated score. Include the complete sentence and a proposed correction through Feedback and Support.

How passage selection works in the app

Practice and Race select a passage from the current language and difficulty group. Starting another text can select a different entry from that group. The selected passage is then the exact target for case, punctuation, and spacing.

The random selection means two runs with identical controls may still use different passages. That is useful for variety, but it introduces a variable into score comparison. If the goal is a controlled baseline, record the passage opening or use enough sessions that one unusually comfortable text does not determine the conclusion.

The seven-session baseline protocol includes a place to record this context.

How to evaluate one passage

A useful manual review separates language quality from typing behavior.

For language quality, ask:

1. Does the paragraph maintain one understandable subject?
2. Are the sentences complete and logically ordered?
3. Is the spelling appropriate for the chosen language?
4. Does punctuation support the sentence rather than create noise?
5. Does any phrase sound copied, promotional, or mechanically repeated?

For typing behavior, ask:

1. Does the text display fully at desktop and mobile widths?
2. Does the current character remain visible as the passage moves?
3. Are case and punctuation mismatches marked correctly?
4. Does a normal space commit the intended word?
5. Does Reset choose clean state without retaining attempts?
6. Does the result reflect the actual selected language and duration?

A passage can pass the first list and expose a product defect in the second. Reports should say which kind of issue occurred.

Why the library is not presented as research

Speedtype has not published population validation, readability calibration, controlled learning outcomes, or proof that one difficulty tier predicts another. The passages are practice material supported by editorial and software checks.

That boundary is important. A transparent small library is more useful than claiming scientific authority that the project has not established. The site publishes its scoring methodology, author identity on About, and limitations so a user can decide whether the tool fits an informal practice goal.

Reporting and correction workflow

For a passage problem, include:

• language and difficulty;
• the first six to ten words;
• the complete sentence containing the issue;
• the current wording;
• the suggested wording and reason;
• whether the issue also affected scoring or display.

Do not include private account information. Feedback is reviewed before any excerpt is made public, and publication requires explicit consent.

After a confirmed correction, the content should receive a new review date, pass the passage validation script, pass browser tests, and be included in a normal release. This creates a traceable path from reader observation to maintained content instead of silently changing text without review.