Digraph (orthography)

From Wikipedia, the free encyclopedia
Jump to: navigation, search
In Welsh, the digraph Ll, ll fused for a time into a ligature.

A digraph or digram (from the Greek: δίς dís, "double" and γράφω gráphō, "to write") is a pair of characters used in the orthography of a language to write either a single phoneme (distinct sound), or a sequence of phonemes that does not correspond to the normal values of the two characters combined.

Digraphs are often used for phonemes that cannot be represented using a single character, like the English sh in ship and fish. In other cases they may be relics from an earlier period of the language when they had a different pronunciation, or represent a distinction which is made only in certain dialects, like English wh. They may also be used for purely etymological reasons, like rh in English. Digraphs are used in some Romanization schemes, like the zh often used to represent the Russian letter ж. As an alternative to digraphs, orthographies and Romanization schemes sometimes use letters with diacritics, like the Czech and Slovak š, which has the same function as the English digraph sh.

In some languages' orthographies, digraphs (and occasionally trigraphs) are considered individual letters, meaning that they have their own place in the alphabet, and cannot be separated into their constituent graphemes, e.g. when sorting, abbreviating or hyphenating. Examples are found in Hungarian (cs, dz, dzs, gy, ly, ny, sz, ty, zs), Czech and Slovak (ch), and Croatian and the Latin orthography of Serbian (lj, nj, dž). In Dutch, when the digraph ij is capitalized, both letters are capitalized (IJ).

Digraphs may develop into ligatures, but these are distinct concepts – a ligature involves a graphical combination of two characters, as when a and e are fused into æ.

Double letters[edit]

Digraphs may consist of two different characters (heterogeneous digraphs) or two instances of the same character (homogeneous digraphs). In the latter case, they are generally called double (or doubled) letters.

Doubled vowel letters are commonly used to indicate a long vowel sound. This is the case in Finnish and Estonian, for instance, where uu represents a longer version of the vowel denoted by u, ää represents a longer version of the vowel denoted by ä, and so on. In Middle English, the sequences ee and oo were used in a similar way, to represent lengthened "e" and "o" sounds respectively; these spellings have been retained in modern English orthography, but the Great Vowel Shift and other historical sound changes mean that the modern pronunciations are quite different from the original ones.

Doubled consonant letters can also be used to indicate length, that is, that the consonant sound is geminated. In Italian, for example, consonants written double are pronounced longer than single ones. This was the original meaning of doubled consonants in Old English, but during the Middle English and Early Modern English period this developed into a spelling convention whereby a doubled consonant serves to indicated that a preceding vowel is to be pronounced short. In modern English, for example, the pp of tapping differentiates the first vowel sound from that of taping. Consonants are only geminated in rare cases where they come from different morphemes, as with the nn in unnatural (un+natural).

Some doubled consonants represent distinct sounds. For example:

  • In Welsh, ll stands for a voiceless lateral consonant, while in Spanish and Catalan it stands for a palatal consonant.
  • In several languages of western Europe, including English, French and Catalan, ss is used between vowels for the voiceless sibilant /s/, since an s alone between vowels is normally voiced, /z/.
  • In Spanish and Catalan, rr is used between vowels for the alveolar trill /r/, since an r alone between vowels represents an alveolar flap /ɾ/ (the two are different phonemes in these languages).
  • In Spanish the digraph nn formerly indicated /ɲ/ (a palatal nasal); it developed into the letter ñ.
  • In Basque, double letters mark palatalized versions, as in dd, ll, tt. However, rr is a trill contrasting with the single-letter flap, as in Spanish, and the palatal version of n is written ñ.

In many European writing systems, including the English one, the doubling of the letter c or k is represented as the heterogeneous digraph ck instead of cc or kk respectively. In native German words, the doubling of z, which corresponds to /ts/, is replaced by the digraph tz.

Pan-dialectical digraphs[edit]

Some languages have a unified orthography with digraphs that represent distinct pronunciations in different dialects (diaphonemes). For example, in Breton there is a digraph zh that is pronounced [z] in most dialects, but [h] in Vannetais. Similarly, the Saintongeais dialect of French has a digraph jh that is pronounced [h] in words that correspond to [ʒ] in standard French. Similarly, Catalan has a digraph ix that is pronounced [ʃ] in Eastern Catalan, but [jʃ] or [js] in Western CatalanValencian.

Discontinuous digraphs[edit]

The pair of letters making up a phoneme are not always adjacent. This is the case with English silent e. For example, the sequence a...e has the sound /eɪ/ in English cake. This is the result of three historical sound changes: cake was originally /kakə/, the open syllable /ka/ came to be pronounced with a long vowel, and later the final schwa dropped off, leaving /kaːk/. Later still, the vowel /aː/ became /eɪ/.

However, alphabets may also be designed with discontinuous digraphs. In the Tatar Cyrillic alphabet, for example, the letter ю is used to write both /ju/ and /jy/. Usually the difference is evident from the rest of the word, but when it is not, the sequence ю...ь is used for /jy/, as in юнь /jyn/ 'cheap'.

The Indic alphabets are distinctive for their discontinuous vowels, such as Thai เ...อ /ɤː/ in เกอ /kɤː/. Technically, however, these are diacritics, not full letters; whether they are digraphs is thus a matter of definition.

Ambiguous letter sequences[edit]

Some letter pairs should not be interpreted as digraphs, but appear due to compounding, like in hogshead and cooperate. This is often not marked in any way, so must be memorized as an exception. Some authors, however, indicate it either by breaking up the digraph with a hyphen, as in hogs-head, co-operate, or with a trema mark, as in coöperate, though usage of this diaeresis has declined in English within the last century. When it occurs in names such as Clapham, Townshend and Hartshorne, it is never marked in any way. Positional alternative glyphs may help to disambiguate in certain cases, e.g. when round s is used as a final variant of long ſ the English digraph resembling /ʃ/ would always be ſh.

In romanization of Japanese, the constituent sounds (morae) are usually indicated by digraphs, but some are indicated by a single letter, and some with a trigraph. The case of ambiguity is the syllabic , which is written as n (or sometimes m), except before vowels or y where it is followed by an apostrophe as n’. For example, the given name じゅんいちろう is romanized as Jun’ichirō, so that it is parsed as /jun.i.chi.rou/, rather than as /ju.ni.chi.rou/.

In several Slavic languages, e.g. Czech, double letters may appear in compound words, but they are not considered digraphs. Examples: bezzubý ‘toothless’, cenný ‘valuable’, černooký ‘black-eyed’.

In alphabetization[edit]

In some languages, certain digraphs and trigraphs are counted as distinct letters in themselves, and assigned to a specific place in the alphabet, separate from that of the sequence of characters which composes them, for purposes of orthography and collation. For example:

Most other languages, including English, French, German, Polish, etc., treat digraphs as combinations of separate letters for alphabetization purposes.


Latin script[edit]


English has both homogeneous digraphs (doubled letters) and heterogeneous digraphs (digraphs consisting of two different letters). Those of the latter type include the following:

Digraphs may also be composed of vowels. Some letters a, e, o are preferred for the first position, others for the second i, u. The latter have allographs y, w in English orthography.

English vocalic digraphs
...e ...i ¦ ...y ...u ¦ ...w ...a ...o
o... oe¦œ > e – /i/ oi¦oy – /ɔɪ/ (SAE: /ɔːə/) ou¦ow – /aʊ¦aʊ̯¦uː (¦oʊ¦əʊ¦ɔʊ¦ʌ)/
(AUS: /æɔ¦æʊ/; CDN: /ʌʊ/; SSE: /ɘʉ/)
oa – /oʊ¦ɔː¦əʊ/ (US: /ɔʊ/) oo – /uː¦ʊ(¦u¦ʌ)/
e... ee – // ei¦ey – /aɪ¦eɪ¦(iː)/ eu¦ew – /juː¦uː/ ea – /iː¦ɛ¦(eɪ¦ɪə) a... ae¦æ > e – /i/ ai¦ay – /eɪ¦ɛ/ au¦aw – /ɔː/ (US: /ɑ¦ɑː¦ɔ/)
(in loanwords: // )
(in afro-American loanwords and proper nouns: aa – /ə¦ɔː¦ɔl/ ) (in loanwords from Chinese: ao – // )
u... ue – /uː¦u/ ui – /ɪ¦uː/ ua – /(ɒ¦ɑ)/
i... iː(¦aɪ)}

Other languages using the Latin alphabet[edit]

In Bosnian, Croatian, Serbian:

Note that in the Cyrillic orthography, these sounds are represented by single letters (љ, њ, џ).

In Czech and Slovak:

In Danish and Norwegian:

  • The digraph aa represented /ɔ/ until 1917 in Norway and 1948 in Denmark, but is today spelt å. The digraph is still used in older names, but sorted as if it were the letter with the diacritic mark.

In Dutch:

In French:

French vocalic digraphs
...i ...u
a... ai – /ɛ¦e/ au – /o/
e... ei – /ɛ/ eu – /œ¦ø/
o... oi – /wa/ ou – /u(¦w)/

See also French phonology.

In German:

In Hungarian:

In Italian:

In Polish:

In Portuguese:

In Spanish:

  • ll is traditionally (but now usually not) pronounced /ʎ/
  • ch represents /tʃ/ (voiceless postalveolar affricate). Since 2010, neither are considered part of the alphabet. They used to be sorted as separate letters, but a reform in 1994 by the Spanish Royal Academy has allowed that they be split into their constituent letters for collation. The digraph rr, pronounced as a distinct alveolar trill, was never officially considered to be a letter in the Spanish alphabet, neither were gu and qu (for /ɡ/ and /k/ respectively before e or i).

In Welsh:

The digraphs listed above represent distinct phonemes. On the other hand, the digraphs mh, nh, and the trigraph ngh, which stand for voiceless consonants, but only occur at the beginning of words as a result of the nasal mutation, are not included in the alphabet.

Daighi tongiong pingim, a transcription system used for Taiwanese Hokkien, includes or which represents /ə/ (mid central vowel) or /o/ (close-mid back rounded vowel), as well as other digraphs.


Main article: Cyrillic digraphs

Modern Slavic languages written in the Cyrillic alphabet make little use of digraphs apart from дж for /dʐ/, дз for /dz/ (in Ukrainian, Belarusian, and Bulgarian), and жж and зж for the uncommon Russian phoneme /ʑː/. In Russian, the sequences дж and дз do occur (mainly in loanwords), but are pronounced as combinations of an implosive (sometimes treated as an affricate) and a fricative; implosives are treated as allophones of the plosive /d̪/, so these sequences are not considered to be digraphs. Cyrillic only has large numbers of digraphs when used to write non-Slavic languages, especially Caucasian languages.


Because vowels are not generally written, digraphs are rare in abjads like Arabic. For example, if sh were used for š, then the sequence sh could mean either ša or saha. However, digraphs are used for the aspirated and murmured consonants (those spelled with h-digraphs in Latin transcription) in languages of South Asia such as Urdu that are written in the Arabic script. This is accomplished with a special form of the letter h which is only used for aspiration digraphs, as seen with the following connecting (kh) and non-connecting (ḍh) consonants:

Urdu connecting   non-connecting
digraph: کھا /kʰɑː/ ڈھا /ɖʱɑː/
sequence:  کﮩا /kəɦɑː/ ڈﮨا /ɖəɦɑː/


In the Armenian language, the digraph ու corresponds to /u/, (close back rounded vowel).


The Georgian alphabet uses a few diacritics when writing other languages. For example, in Svan, /ø/ is ჳე "we", and /y/ is ჳი "wi".


Modern Greek has the following digraphs:

  • αι (ai) represents /e̞/
  • ει (ei) represents /i/
  • οι (oi) represents /i/
  • ου (oy) represents /u/
  • υι (yi) represents /i/

These are called "diphthongs" in Greek; in classical times most of them did represent diphthongs, and the name has stuck.

  • γγ (gg) represents /ŋɡ/ or /ɡ/
  • τσ represents the affricate /ts/
  • τζ represents the affricate /dz/
  • Initial γκ (gk) represents /ɡ/
  • Initial μπ (mp) represents /b/
  • Initial ντ (nt) represents /d/

Ancient Greek also had the "diphthongs" listed above although their pronunciation in ancient times is disputed. In addition Ancient Greek also used the letter γ combined with a velar stop to produce the following digraphs:

  • γγ (gg) represents /ŋɡ/
  • γκ (gk) represents /ŋɡ/
  • γχ (gkh) represents /ŋkʰ/

Tsakonian has a few additional digraphs: ρζ /ʒ/ (historically perhaps a fricative trill), κχ /kʰ/, τθ /tʰ/, πφ /pʰ/, σχ /ʃ/. In addition, palatal consonants are indicated with the vowel letter ι, but this is largely predictable. When /n/ and /l/ are not palatalized before ι, they are written νν and λλ.

In Bactrian, the digraphs ββ, δδ, γγ were used for /b/, /d/, /ŋg/.


In the Hebrew alphabet, תס and תש may sometimes be found for צ /ts/. Modern Hebrew also uses digraphs made with the ׳ symbol for non-native sounds: ג׳ //, ז׳ /ʒ/, צ׳ //; and other digraphs of letters when it is written without vowels: וו for a consonantal letter ו in the middle of a word, and יי for /aj/ or /aji/, etc., that is, a consonantal letter י in places where it might not have been expected. Yiddish has its own tradition of transcription, so uses different digraphs for some of the same sounds: דז /dz/, זש /ʒ/, טש //, and דזש (literally dzš) for //, וו /v/, also available as a single Unicode character װ, וי or as a single character in Unicode ױ /oj/, יי or ײ /ej/, and ײַ /aj/. The single-character digraphs are called "ligatures" in Unicode. י may also be used following a consonant to indicate palatalization in Slavic loanwords.


Most Indic scripts have compound vowel diacritics that cannot be predicted from their individual elements. This can be illustrated with Thai, where the diacritic เ, on its own pronounced /eː/, modifies the pronunciation of other vowels:

single vowel sign: กา /kaː/, เก /keː/, กอ /kɔː/
vowel sign plus เ: เกา /kaw/, แก /kɛː/, เกอ /kɤː/

In addition, the combination รร is pronounced /a/ or /am/, there are some words where the combinations ทร and ศร stand for /s/ and the letter ห as prefix to a consonant changes its tonic class to high, modifying the tone of the syllable.


Inuktitut syllabics adds two digraphs to Cree:

rk for q
qai, ᕿ qi, ᖁ qu, ᖃ qa, ᖅ q


ng for ŋ

The latter forms trigraphs and tetragraphs.


Two kana may be combined into a CV syllable by subscripting the second; this convention cancels the vowel of the first. This is commonly done for CyV syllables called yōon, as in ひょ hyo hiyo. These are not digraphs, as they retain the normal sequential reading of the two glyphs. However, some obsolete sequences no longer retain that reading, as in くゎ kwa, ぐゎ gwa, and むゎ mwa, now pronounced ka, ga, ma. In addition, non-sequenceable digraphs are used for foreign loans that do not follow normal Japanese assibilation patterns, such as ティ ti, トゥ tu, チェ tye / che, スェ swe, ウィ wi, ツォ tso, ズィ zi. (See Katakana and Transcription into Japanese for complete tables.)

Long vowels are written by adding the kana for that vowel, in effect doubling it. However, long ō may be written either oo or ou, as in とうきょう toukyou [toːkjoː] 'Tōkyō'. For dialects which do not distinguish ē and ei, the latter spelling is used for a long e, as in へいせい heisei [heːseː] 'Heisei'.

There are several conventions of Okinawan kana which involve subscript digraphs or ligatures. For instance, in the University of the Ryukyus system, ウ is /ʔu/, ヲ is /o/, but ヲゥ is /u/.


As was the case in Greek, Korean has vowels descended from diphthongs that are still written with two letters. These digraphs, ㅐ /ɛ/ and ㅔ /e/ (also ㅒ /jɛ/, ㅖ /je/), and in some dialects ㅚ /ø/ and ㅟ /y/, all end in historical ㅣ /i/.

Hangul was designed with a digraph series to represent the "muddy" consonants: ㅃ *[b̥], ㄸ *[d̥], ㅉ *[d̥z̥], ㄲ *[ɡ̊], ㅆ *[z̥], ㆅ *[ɣ̊]; also ᅇ, with an uncertain value. These values are now obsolete, but most of these doubled letters were resurrected in the 19th century to write consonants which had not existed when hangul was devised: ㅃ /p͈/, ㄸ /t͈/, ㅉ /t͈ɕ/, ㄲ /k͈/, ㅆ /s͈/.

In Unicode[edit]

Generally, a digraph is simply represented using two characters in Unicode.[1] However, for various reasons, Unicode sometimes provides a separate code point for a digraph, encoded as a single character.

The DZ and IJ digraphs and the Serbian/Croatian digraphs DŽ, LJ, and NJ have separate code points in Unicode.

Two Glyphs Digraph Unicode Code Point HTML
DZ, Dz, dz DZ, Dz, dz U+01F1 U+01F2 U+01F3 DZ Dz dz
DŽ, Dž, dž DŽ, Dž, dž U+01C4 U+01C5 U+01C6 DŽ Dž dž
IJ, ij IJ, ij U+0132 U+0133 IJ &#307
LJ, Lj, lj LJ, Lj, lj U+01C7 U+01C8 U+01C9 LJ Lj lj
NJ, Nj, nj NJ, Nj, nj U+01CA U+01CB U+01CC NJ Nj nj

See also Ligatures in Unicode.

See also[edit]

Notes and references[edit]

  1. ^ "FAQ – Ligatures, Digraphs and Presentation Forms". The Unicode Consortium: Home Page. Unicode Inc. 1991–2009. Retrieved 2009-05-11.