CaseConv is a text case converter with one unusual commitment: where the published style guides disagree, it reproduces the disagreement instead of averaging it away, and where a rule cannot be applied without guessing, it says so on the screen and leaves the call to you.
Why four title cases and not one
Most case converters offer a single "Title Case" button. That button has to pick a rule set, and it almost never tells you which one it picked. The result is a headline that is correct for a guide you may not be writing under — and the failure is invisible, because the output looks perfectly reasonable.
The four guides here genuinely differ. AP and APA decide capitalisation by word length; Chicago and MLA decide it by word class. APA has no rule capitalising the last word of a title, and the other three do. Chicago changed its preposition rule in its 18th edition, in 2024, and a great many tools still implement the 17th. Those are not shades of preference; they produce different strings from the same input.
| Guide | Decides by | Source |
|---|---|---|
| AP (Associated Press Stylebook) | length | AP Stylebook, “composition titles” entry (wording of the 54th ed., 2019, onward); read from archived copies of the entry, never from a live subscription |
| Chicago Manual of Style, 18th edition (2024) | word class | CMOS 18th ed. §8.160 — the section text is paywalled; encoded from the publisher’s own online guidance (Chicago Manual of Style Online Q&A, “Capitalization, Titles”) and the most consistent secondary reproductions |
| APA 7th edition | length | APA Style, “Title case capitalization”, apastyle.apa.org — the publisher’s own free guidance page, read directly |
| MLA Handbook, 9th edition | word class | MLA Handbook 9th ed. §2.90 — the section text is paywalled; encoded from the publisher’s own online guidance and the most consistent secondary reproductions |
How the rules were built
Each guide is encoded as data in one file, with the rule it implements written in a comment immediately beside it. Word lists follow the wording each manual's publisher and the most consistent secondary reproductions give; no tidied-up composite is built from them. Chicago's rule names five coordinating conjunctions and MLA's names seven, so this tool holds two different lists and never reconciles them into one.
Where that wording came from is stated per guide in the source column above, because the four are not equally accessible. APA publishes its title-case rule free on its own site. AP's composition-titles entry is a subscription product, read here from archived copies. Chicago §8.160 and MLA §2.90 are both paywalled sections: what is encoded for those two follows the publishers' own online guidance and the secondary reproductions that agree with each other, which is a weaker footing than reading the section and is labelled as such in the table above.
Every rule set is covered by an automated test suite that asserts the exact strings printed in the comparison tables on this site. If a rule changes, the tests fail before the pages do.
Every table here is generated, not typed
The comparison tables, the word matrix, the developer-case matrix, the character-count examples and the Unicode expansion lists are all produced at build time by calling the same functions the buttons on the page call. Nothing is transcribed by hand.
The constraint is deliberate, and it is there for a reason. A converter whose documentation is written separately from its code drifts: the article says one thing, the tool does another, and the reader cannot tell which is wrong. Here the page and the tool are the same thing, so they cannot disagree.
The Unicode figures are measured the same way. The lists of characters that change length when their case changes are produced by scanning the Basic Multilingual Plane with the JavaScript engine's own case mappings at build time — the same mappings that will run in your browser. Nothing was copied from a reference table.
What this tool refuses to decide
Some title-case sub-rules depend on grammar rather than spelling. Whether "up" is a preposition or the particle of a phrasal verb, whether "to" marks an infinitive, whether "than" introduces a clause — a converter cannot see any of that, because the information is not in the letters. Where a rule depends on it, CaseConv applies the conservative reading and names the word in an amber panel so a human can overrule it.
- Is "up" in “Backing Up Your Laptop” a preposition or the adverbial particle of a phrasal verb? All four guides capitalise an adverbial particle. Deciding which one it is needs a parser, so CaseConv applies the preposition reading (lowercase) and flags the word under “Judgment calls” for you to fix.
- Does AP capitalise “to” when it marks an infinitive — “How To Win” or “How to Win”? It does, and this one is not in dispute: AP’s composition-titles entry explicitly capitalises “to” in infinitives — “What I Want To Be When I Grow Up.” Telling an infinitive marker from a preposition still needs a parser, so CaseConv lowercases mid-title “to” under every guide and flags it for you to fix. Chicago and MLA lowercase it either way.
- Does Chicago lowercase “so” and “yet”? Chicago’s rule names exactly five coordinating conjunctions — and, but, for, or, nor — and “so” and “yet” are not among them, so CaseConv capitalises them under Chicago. MLA names all seven FANBOYS, so MLA lowercases them. Secondary summaries of Chicago disagree with each other here; we follow the wording the manual’s publisher and the most consistent secondary reproductions give, and flag both words whichever guide you pick.
- Does AP capitalise the first word of a subtitle after a colon? AP publishes no rule for subtitles inside a composition title, so there is no AP text to follow. APA, Chicago and MLA all capitalise it, and AP’s own published copy does the same, so CaseConv extends that shared behaviour to AP and invents no AP-specific exception.
None of these is unique to this tool; every rule-based converter faces them. The difference is being told which words were judgement calls, so the five per cent that needs a human gets one.
Your text never leaves your browser
The whole site is static files. Every conversion is a JavaScript function running in your own tab — there is no server component, no account, no upload and no logging. You can verify it: open your browser's network panel and convert something, and not a single request fires.
That is not a nice-to-have. People paste unpublished manuscripts, client copy, internal headings and legal documents into case converters. It should be safe to do that, and the only way to make it safe is for the text never to be transmitted at all.
What it is not
This is a capitalisation tool and not an editor. It does not check grammar, fix spelling, or know which words in your sentence are proper nouns. It cannot restore a capital that was already folded away — once "nasa" is lowercase, nothing in the text says it was an acronym. And it applies a style guide's rules; it does not adjudicate your publisher's house exceptions to them.
Corrections
If a rule here is wrong, it should be fixed and not defended. If you can point to the wording in the relevant manual, the contact page is the fastest route — a rule with a citation gets encoded, tested and shipped; a rule without one does not, which is the same standard applied to everything already on the site.