Why Do Some Letters Have Marks on Them? Extending an Alphabet That Ran Out

The Latin alphabet was developed for Latin and has about twenty three letters, which is fewer than the distinct sounds in most languages that later adopted it. A language with more sounds than letters has three options, namely inventing new letter shapes, combining letters into pairs that stand for one sound, or modifying existing letters with added marks. The third produces the smallest disruption, since the modified letters remain recognisable and can be sorted and typed alongside the originals, which is why it became the dominant solution across Europe and beyond. Each language chose its own set, which is why the same mark means different things in different places.

Several distinct functions are served by what look like similar marks:

• •Indicating a different vowel sound entirely, as in several Germanic and Slavic languages
• •Indicating vowel length, which distinguishes words in Czech and Hungarian
• •Indicating which syllable is stressed, as in Spanish and Greek
• •Indicating tone, which is essential in Vietnamese and in transcribed Chinese
• •Distinguishing otherwise identical words, which is their only job in some cases
• •Indicating that a letter is pronounced separately rather than as part of a pair

Treating these marks as embellishment produces errors that range from confusing to serious, because in most languages that use them the marked and unmarked letters are different letters rather than variants of one. They are listed separately in dictionaries, sorted separately in alphabetical order, and taught as distinct. Removing them creates different words, and the examples are numerous enough that native speakers find writing without them genuinely hard to read. The habit of dropping them in English usage of foreign names is therefore a spelling error rather than a simplification, and it has practical consequences in databases, legal documents and identification where a name is legally one thing and recorded as another.

Several characters that look like modified letters have become independent letters in their own right, and the distinction matters for sorting and for identity. The Scandinavian languages treat certain marked vowels as separate letters placed at the end of the alphabet rather than alongside their unmarked forms, so a name is sorted in a position that surprises speakers of other languages. Spanish treated one marked consonant as a separate letter for centuries and still sorts it separately. Polish, Czech, Turkish and Vietnamese all have letters formed with marks that are counted as distinct in their alphabets. In each case the mark is not an addition to a letter but part of the letter, which is why keyboards in those countries give them dedicated keys.

Machines have been an obstacle to these marks for a century and a half. Typewriters offered a limited set of keys, and layouts were built per language, so a document mixing languages was frequently impossible. Early computing used character sets of a hundred and twenty eight slots that held unaccented Latin letters and nothing else, which forced ugly workarounds including writing marks as following characters. A proliferation of incompatible extended sets followed, so a document written on one system displayed as nonsense on another, which is where the familiar corrupted characters come from. The Unicode standard resolved this by assigning every mark and marked letter a unique number, and residual problems now concern sorting, searching and matching rather than display.

The Latin alphabet has fewer letters than most languages have sounds, and adding marks to existing letters was the least disruptive of the three available solutions. The marks indicate vowel quality, length, stress, tone or word identity, and they differ in meaning between languages. Marked letters are separate letters, sorted separately in dictionaries, so dropping them is a spelling error rather than a simplification.