Regexes to help fix lowercase ls that have been turned into uppercase Is

Such as in the case of some subtitle files that have been obtained by OCR'ing images written in a font that doesn't distinguish between l and I.

These don't cover every case (these miss ls at the start of lines but in the middle of sentences, and might convert some proper nouns that should begin with an I), nor have I tested them for languages other than English, but this does an effective job of getting most of the easy cases.

Remember to use regex search mode in your text editor, and remember to set case-sensitive mode on! Otherwise you'll just replace every lowercase i with an l.

If you have the reverse problem, of all uppercase Is being represented as lowercase ls, my suggestion is to first globally replace l→I, then run the below replacements. (I haven't tested it yet as of first posting this, though.)

The replacements here use backslash syntax (\1, \2 etc.) to represent capture groups in replacements. Of course, if your regex language uses dollar signs instead ($1, $2 etc.), then replace backslashes with dollars.

1. • Find: ([a-z']+)I([a-z]*)
   • Replace with: \1l\2
   • Run this one twice to handle double-Ls
   • What the regex searches for: occurrences of capital-I in the middle or end of otherwise lowercase words, or right after an apostrophe (like in words such as we'II→we'll).
2. • Find: ([A-Z]+)I([a-z]+)
   • Replace with: \1l\2
   • What the regex searches for: occurrences of capital-I that are between an uppercase letter and a lowercase letter.
3. • Find: (([a-z",]|I) [a-z]*)I([a-z]+)
   • Replace with: \1l\3
   • What the regex searches for: words that start with a capital-I, but are followed by lowercase letters, and which follow a word ending in lowercase characters, or punctuation that could reasonably be found in the middle of a sentence.
   • This will also convert proper nouns in the middle of sentences (Call me Ishmael → Call me lshmael). The workaround is to just fix those manually afterwards (or, first do a search for text matching this regex that would be replaced, and take a note of all the names that it finds, so they're easier to fix); there are usually fewer proper nouns like these than there are other words that begin with l.
   • Your regex processor could number nested capture groups differently – what is wanted here is the larger, outside capture group before the capital-I, not the smaller one within.
4. • Find: ([a-z]['-]+)I([a-z]*)
   • Replace with: \1l\2
   • What the regex searches for: a capital-I that follows some punctuation which itself follows a lowercase letter (e.g. fleur-de-Iis→fleur-de-lis). Looking for a lowercase letter is important, because otherwise this could replace correct capital-Is that are at the beginning of speech lines that are marked with a dash.

First posted 2025-08-20, last edited 2025-08-20. index