Polyglotta Africana is a study published in 1854 by the German missionary Sigismund Wilhelm Koelle (1823–1902), in which the author compares 280 words from 200 African languages and dialects (or about 120 separate languages according to today's classification; several varieties considered distinct by Koelle were later shown to belong to the same language). As a comparative study it was a major breakthrough at the time. Koelle based his material on first-hand observations, mostly with freed slaves in Freetown, Sierra Leone. He transcribed the data using a uniform phonetic script. Koelle's transcriptions were not always accurate; for example, he persistently confused with and with. His data were consistent enough, however, to enable groupings of languages based on vocabulary resemblances. Notably, the groups which he set up correspond in a number of cases to modern groups: Although Koelle's was not the first such study comparing different African languages, (for example, a missionary called John Clarke had produced a similar work in 1848, and still earlier Hannah Kilham had produced her Specimens of African Languages, Spoken in the Colony of Sierra Leone in 1828), yet in its accuracy and thoroughness it outclassed all the others and still proves useful today.

Value of the work

The Polyglotta Africana was the second work carried out by Koelle during his five years in Sierra Leone, the first being a grammar of the Vai language in 1849. The idea of this was to use the fact that Sierra Leone was a melting pot of ex-slaves from all over Africa to compile a list of 280 basic words (a sort of early Swadesh list) in some 160 languages and dialects. These were then grouped as far as possible in families. Most of the informants who contributed to this work came from West Africa, but there were also others from as far away as Mozambique. One area that was lacking was the Swahili coast of Kenya and Tanzania, since it seems that slaves from this region were generally taken northwards to Zanzibar and Arabia rather than southward towards America and Brazil. The pronunciations of all the words were carefully noted using an alphabet similar, though not identical, to that devised by Karl Richard Lepsius, which was not yet available at that time. The name of the book was imitated from a well-known work called Asia Polyglotta (1823) by the German scholar Julius Klaproth. The value of the list is not merely linguistic, since the work not only includes the words themselves, but Koelle also added a short biography of each informant, with geographical information about their place of origin, and an indication of how many other people they knew in Sierra Leone who spoke the same language. This information, combined with a census of Sierra Leone conducted in 1848, has proved invaluable to historians researching the African slave trade in the 19th century. Of the 210 informants, there were 179 ex-slaves (two of them women), while the rest were mostly traders or sailors. An analysis of the data shows that typically Koelle's informants were middle-aged or elderly men who had been living in Freetown for ten years or more. Three-quarters of the ex-slaves had left their homeland more than ten years earlier, and half of them more than 20 years before; and three-quarters of the informants were over 40 years old. Another interesting facet of the book is the manner in which the informants had been made slaves. Some had been captured in war, some kidnapped, some sold by a relative, others condemned for a debt or sentenced for a crime. Included with a book is a map of Africa showing the approximate location, as far as it could be ascertained, of each language, prepared by the cartographer August Heinrich Petermann.

The transcription

It was Koelle's aim not to use any previously published material on the languages he was writing down, but to achieve uniformity by having one person using a single phonetic system for every language. The orthography he eventually chose, after discussions in London, was not that of Karl Richard Lepsius (as is sometimes claimed), since it had not yet been published, but was based on a short document issued in 1848 by Henry Venn of the Church Missionary Society entitled Rules for Reducing Unwritten Languages to Alphabetical Writing in Roman Characters With Reference Specially to the Languages Spoken in Africa. The aim of this was to produce a simple practical system of orthography for teaching purposes with the use of as few diacritics as possible. Koelle, however, sought a more accurate phonetic system, and added diacritics. He retained seven of the eight vowels of Venn's system (i, e, ẹ, a, ọ, o, u, omitting ạ as in "but") but added length marks, a dot for nasalisation, and an accent to indicate the prominent syllable. (Unlike in Lepsius's alphabet, the dotted ẹ and ọ are open not closed sounds.) He modified Venn's alphabet by writing dṣ for the sound of "judge" or "church" (apparently confusing these two), and n followed by a dot (n˙) for the "ng" sound of "sing". When Koelle learnt of Lepsius's alphabet in 1854, he made immediate use of it in his Kanuri grammar, in which he wrote:

Koelle's word list

In the introduction Koelle tells us that he wanted a selection of words that would be simple enough for each informant to be interviewed on a single day, and for this reason he omitted pronouns, which would have taken much longer to elicit. He adds that a few years earlier during a long vacation he had made a similar such list, of just 71 languages, and that in making the present list he had learnt from that experience. The actual list (the spelling is Koelle's) is as follows:

The languages

As the list of languages and countries below shows, most of Koelle's languages came from West Africa. This is mainly because the majority of the slaves themselves who were intercepted by the British Navy and taken to Sierra Leone were from that region. Another factor was that the number of different languages in West Africa is greater than in some other parts of Africa. For example, Cameroon alone is said to have 255 different languages. One area missing is the Swahili coast of Kenya and Tanzania, apparently because slaves intercepted there were taken not to Sierra Leone but to Zanzibar. Koelle's language names are given in the left-hand column of the table below: some of the diacritics (such as the dot beneath ẹ and ọ, and the acute accent) have been omitted. The groupings are Koelle's own. The larger groups are subdivided by Koelle into smaller groups, which are not shown in the table. Names in square brackets such as [Aku] are subheadings of a group of languages, and do not themselves have any words. The number of languages or dialects represented on each double-page spread of Koelle's book is therefore exactly 200, divided into four columns of 50 languages each.