Why does the letter combination gh in English sometimes sound /f/ and sometimes not?

The inconsistent pronunciation of the letter combination *gh* in English is a direct consequence of historical sound changes that were not matched by spelling reforms, creating a fossilized orthographic record of phonological evolution. The primary source of this inconsistency is the Middle English phoneme /x/, a voiceless velar fricative similar to the *ch* in Scottish *loch* or German *Bach*. This sound was consistently spelled *gh* following a vowel. During the Middle English period, this /x/ was pronounced in words like *light* (pronounced similarly to *licht*), *thought*, and *enough*. The divergence in modern pronunciation stems from two major, geographically distinct sound changes that occurred between the 14th and 17th centuries: the complete loss of the sound in some contexts and its transformation into /f/ in others.

In most dialects of English, particularly those that evolved into Standard British and American English, the velar fricative /x/ was simply lost after front vowels, leading to silent *gh* in words like *light*, *high*, *though*, and *through*. This deletion often lengthened the preceding vowel, a detail still reflected in modern pronunciation. However, in the case of the word *daughter*, the *gh* also became silent, but here it followed a back vowel. The pattern is not perfectly predictable by the vowel, as seen in *dough* versus *tough*, which illustrates the second major change. In a subset of words, primarily where the /x/ followed the back rounded vowel /uː/ or its variants, the sound underwent a process of assimilation. The velar fricative /x/, produced at the back of the mouth, shifted forward to become the labiodental fricative /f/, which is produced with the lower lip and upper teeth. This change yielded the modern /f/ in words like *tough*, *enough*, *laugh*, and *cough*.

The current irregularity is therefore not random but a map of which words underwent which change. The variation is further complicated by later borrowings from other languages that introduced *gh* with different values, such as *ghost* (from Dutch, where *gh* represents /ɣ/ or /g/) and *spaghetti* (from Italian, with a hard /g/). These loanwords operate outside the native historical pattern. Furthermore, the fossilization of spelling during the advent of the printing press cemented these varied outcomes in their inconsistent forms. From a synchronic perspective, this results in a notorious challenge for learners of English, as the grapheme *gh* now functions as a complex marker with multiple possible realizations: silent (*high*), /f/ (*laugh*), /g/ (*ghost*), or even /p/ in the rare case of *hiccough*, a folk-etymology respelling of *hiccup*.

The legacy of *gh* is a quintessential example of how English spelling often represents an older stage of the language's pronunciation. It serves as a diagnostic for historical linguistics, revealing the pathways of phonetic change, while posing a persistent orthographic problem. There is no simple rule for its modern pronunciation because it is the residue of completed sound changes; one must largely memorize the outcome for individual words or recognize patterns within word families, such as the consistent /f/ in *-ough* words like *rough* and *tough* versus the silent variant in *though* and *through*. This inconsistency is not a flaw but a historical layer, directly linking modern English to its Middle English antecedents through its conservative spelling.