Trigraphs

Back to The Compiler

C also provides a 3 char representation of punctuation marks all starting with two question marks.

??( is translated into [

??) is translated into ]

??< is translated into \{

??> is translated into \}

??= is translated into #

??/ is translated into \\

??! is translated into !

??' is translated into ^

??- is translated into ~

The plain idea

Trigraphs are three-character stand-ins for punctuation marks that some very old or non-English keyboards and character sets could not type. In the 1980s not every terminal had {, }, [, ], #, \, ^, ~ or | available, so C let you write, for example, ??( wherever you meant [. The compiler replaced every trigraph before doing anything else, in the very first translation step.

Why it matters

Mostly as history and as a trap. You will almost never want a trigraph, but because the replacement happened so early and so blindly, an innocent ?? inside a string could change your program’s meaning without warning, which is exactly why they are worth knowing about.

An everyday analogy: a typewriter workaround

Old typewriters without a “1” key taught typists to use a lowercase “l” instead. It looked right on paper but confused anything that read the text literally. Trigraphs were that workaround, baked into the C language itself for keyboards missing certain symbols.

A worked example: the surprise

printf("What??!\n");   // prints: What|

That is not a typo in the output. ??! is the trigraph for |, so the compiler silently rewrote the string. This kind of accidental substitution is the whole reason trigraphs became infamous.

A 2026 update

Trigraphs were removed from C in C23 (and from C++ in C++17). Modern compilers ignore them by default and often warn if they spot a sequence that used to be one. Today, with Unicode keyboards everywhere, they are purely a historical curiosity, and that is good news: one fewer silent surprise.

Common misconceptions

• “They’re still active in C.” Not since C23. Even before that, GCC and Clang disabled them unless you asked.
• “They’re the same as digraphs.” No. Trigraphs are replaced character-by-character anywhere, even inside strings; digraphs are proper tokens and never touch string contents.

How it connects

• Digraphs: the safer, token-level alternative.
• Tokens and The Compiler: where this substitution sits in translation.
• Unicode in C: why the problem trigraphs solved no longer exists.

The truth revealed: trigraphs are a fossil from an age of limited keyboards, kept alive just long enough to bite the unwary, and finally retired in C23.