The Languages of Africa is a seminal 1963 book of essays by Joseph Greenberg, in which he sets forth a genetic classification of African languages that, with some changes, continues to be the most commonly used one today. It was first published in 1963 as Part II of the International Journal of American Linguistics, Vol. 29, No. 1, as an extensively revised version of his 1955 work Studies in African Linguistic Classification. Its second edition of 1966, in which it was published (by Indiana University, Bloomington: Mouton & Co., The Hague) as an independent work, is more commonly cited.

Its author describes it as based on three fundamentals of method:

• "The sole relevance in comparison of resemblances involving both sound and meaning in specific forms."
• "Mass comparison as against isolated comparisons between pairs of languages."
• "Only linguistic evidence is relevant in drawing conclusions about classification."

The second point, mass comparison, is controversial in historical linguistics, particularly as practiced by Greenberg. The third is completely uncontroversial in modern times, and is directed against previous African linguists (notably Meinhof) who had classified languages on typological and even racial grounds.

It classifies Africa's languages into four stocks not presumed to be related, as follows:

I. Congo-Kordofanian

I.A Niger-Congo

I.A.1 West Atlantic

I.A.2 Mande

I.A.3 Voltaic

I.A.4 Kwa

I.A.5 Benue-Congo

I.A.6 Adamawa-Eastern

I.B Kordofanian

I.B.1 Koalib

I.B.2 Tegali

I.B.3 Talodi

I.B.4 Tumtum

I.B.5 Katla

II. Nilo-Saharan

II.A Songhai

II.B Saharan

II.C Maban

II.D. Fur

II.E. Chari-Nile

II.E.1 Eastern Sudanic

II.E.2 Central Sudanic

II.E.3 Berta

II.E.4 Kunama

II.F Koman

III. Afroasiatic

III.A Semitic

III.B Egyptian

III.C Berber

III.D Cushitic

III.D.1 Northern Cushitic

III.D.2 Central Cushitic

III.D.3 Eastern Cushitic

III.D.4 Western Cushitic

III.D.5 Southern Cushitic

III.E Chad

IV Khoisan

IV.A South African Khoisan

IV.A.1 Northern South African Khoisan

IV.A.2 Central South African Khoisan

IV.A.3 Southern South African Khoisan

IV.B Sandawe

IV.C Hatsa

His Niger-Congo family was substantially foreshadowed by Westermann's "Western Sudanic", but he changed the subclassification, including Fulani (as West Atlantic) and the newly postulated Adamawa-Eastern, excluding Songhai, and classifying Bantu as merely a subfamily of Benue-Congo (previously termed "Semi-Bantu").

Semitic, Berber, Egyptian, and Cushitic had been generally accepted as members of a "Hamito-Semitic" family, while Chadic, "Nilo-Hamitic", Fulani, and Hottentot had all been controversially proposed as members. He accepted Chadic (while changing its membership), and rejected the other three, establishing to most linguists' satisfaction that they had been classified as "Hamitic" for purely typological reasons. This demonstration also led to the rejection (by him and by linguistics as a whole) of the term Hamitic as having no coherent meaning in historical linguistics; as a result, he renamed the newly reclassified family "Afro-Asiatic". This has since been emended by changing the status of "Western Cushitic", making it an independent subfamily of Afroasiatic called Omotic.

Following Schapera and rejecting Meinhof, he classified Hottentot as a member of the Central Khoisan languages; to this group he added the much more northerly Hatsa and Sandawe. This change remains controversial, and is widely considered inadequately founded.

His most revolutionary step was the postulation of the Nilo-Saharan phylum; this is still highly controversial, despite the publication of claimed reconstructions of the family, but is widely used. Within this, he identified a major subgroup termed Chari-Nile , containing Eastern Sudanic, Central Sudanic, Kunama, and Berta; this has been generally rejected. On a lower level, he placed "Nilo-Hamitic" firmly within Nilotic, following a suggestion of Köhler, and placed Eastern Sudanic on a firmer foundation.

Finally, he assigned the non-Nilo-Saharan languages of the Nuba Hills of Kordofan to a single subfamily (Kordofanian), which together with Niger-Congo formed a new phylum, Congo-Kordofanian. This is generally accepted, with the exception of the "Tumtum" group, although it is unclear whether Kordofanian or Mande was the first branching.