The African Reference Alphabet is a largely defunct continent-wide guideline for the creation of Latin alphabets for African languages. Two variants of the initial proposal (one in English and a second in French) were made at a 1978 UNESCO-organized conference held in Niamey, Niger. They were based on the results of several earlier conferences on the harmonization of established Latin alphabets of individual languages. The 1978 conference recommended the use of single letters for speech sounds rather than of letter sequences or of letters with diacritics. A substantial overhaul was proposed in 1982 but was rejected in a follow-up conference held in Niamey in 1984. Since then, continent-wide harmonization has been largely abandoned, because regional needs, practices and thus preferences differ greatly across Africa.

Through the individual languages that were its basis, the African Reference Alphabet inherits from the Africa Alphabet and, like its predecessor, uses a number of IPA letters. The Niamey conference built on the work of a previous UNESCO-organized meeting, on harmonizing the transcriptions of African languages, that was held in Bamako, Mali, in 1966.

1978 proposals

Separate versions of the conference's report were produced in English and French. Different images of the alphabet were used in the two versions, and there are a number of differences between the two.

The English version was a set of 57 letters, given in both upper-case and lower-case forms. Eight of these are formed from common Latin letters with the addition of an underline mark. Some (the uppercase letters alpha, eth (x12px|class=skin-invert), esh, and both lower- and upper-case 20px|class=skin-invert, 20px|class=skin-invert) cannot be accurately represented in Unicode (as of version 15, 2023). Others do not correspond to the upper- and lower-case identities in Unicode, or (e.g. Ʒ, Ʃ) require character variants in the font.

This version also listed eight diacritical marks (acute accent (´), grave accent (`), circumflex (ˆ), caron (ˇ), macron (¯), tilde (˜), trema (¨), and a superscript dot (˙) and nine punctuation marks (? ! ( ) « » , ; .).

The letters presented in the Annex 1 of the 1978 Niamey meeting report are slightly different from the ones presented on page 34 (page 32 in the French version) which omitted the hooktop-z but included two apostrophe-like letters (for ʔ and ʕ). Five of the letters were written with a subscript dot instead of a subscript dash as in the English version (ḍ ḥ ṣ ṭ and ẓ). The French and English sets are otherwise identical.

thumb|upright=1.6|African Reference Alphabet, as presented on the 1978 Niamey conference Annex 1 (printed English version)|class=skin-invert-image

thumb|upright=1.6|African Reference Alphabet, as presented on the 1978 Niamey conference (handwritten French version)|class=skin-invert-image

[[File:African reference alphabet as presented 1978 on the Niamey meeting (English).png|thumb|upright=1.6|African Reference Alphabet, as presented on the 1978 Niamey conference (handwritten English version)

  • The pharyngeal ḥ and pharyngealized ḍ, ṣ, ṭ, ẓ are presented with lines below as h̠ and d̠, s̱, t̠, z̠ in the Annex 1 but with dots in the other parts of the 1978 Niamey meeting report (both in the French and English versions) These represent Arabic-style emphatic consonants.
  • c, j represent either palatal stops or postalveolar affricates. ɖ, ʈ are the retroflex stops, as in the IPA.
  • ƒ, ʋ represent bilabial fricatives.
  • ө is a dental fricative, not a vowel.
  • Although digraphs using h are normally used to represent aspirated consonants, in languages in which those are absent, the digraphs can be used instead of ʒ, ʃ, ө, ɣ...

thumb|upright=1.6|African Reference Alphabet (revised version 1982) as proposed by Michael Mann and David Dalby|class=skin-invert-image

{|class="wikitable Unicode"

|+Mann & Dalby's revised African Reference Alphabet It is not specifically supported in Unicode (), but can be represented by or . and are written without ascenders (thus esh is a mirror of ; is written with a right-hooking tail, like the retroflex letters in the IPA; and has a top hook to the left, like a squashed . , inspired by the shape of insular t, is meant to complete the ejective letters with hook series ; in practice, is used instead.

Because no language has all the consonants, the consonant letters are used for more than one potential value. They can be reassigned when there are conflicts. For instance, ɦ may be a voiceless pharyngeal, a voiced glottal fricative, or even (in the Khoekhoe table) an alveolar nasal click to avoid the digraph ɖɴ.

{|class="wikitable Unicode"

|+Consonant chart (1982)