Today is International Day of the Basque Language, which we celebrate at Lingoblog with an article by Peter Bakker.
One person in Spain has claimed that Basque is related with the Dogon language spoken in Mali in Africa. Does that claim make sense? It sounds unlikely, but it is important not to reject the idea without checking the evidence.
Background information:
Languages are classified into language families. The languages of a family have been proven to be related, i.e. scholars are certain that these languages descend from the same ancestor language, which was perhaps spoken many millennia ago. Some families are as small as one language, for instance Basque. A language like Basque is therefore strictly speaking not a family, but it is called an isolate. It has no relatives. English and Danish are part of a huge family called Indo-European, which covers most of the languages of Eurasia spoken from Ireland in the West to Bangla Desh in the East. Some 450 languages are part of this extended family. Proto-Indo-European was spoken some 6,000 to 8,000 years ago in Anatolia or close to there.
Not surprisingly, many people, professional linguists and amateurs alike have tried to prove that Basque is related to another language or another language family somewhere on Earth. The first person to do that will gain eternal fame. But a successful claim of any relation between two families is extremely rare. Perhaps only five times in the past century such proposed connections have been accepted.
People have tried to connect Basque to other languages or language families, but thus far in vain. Some people have claimed connections with the languages of the Caucasus, like John Bengtson, and before him, Karl Bouda and others. Some have claimed connections with the isolate Burushaski in the Himalaya (Hermann Berger) or Berber in North Africa (Georg von der Gabelentz). Most recently, Juliette Blevins claimed a connection with Indo-European languages. None of these proposals have been accepted by specialists. Not because the Bascologists are conservative, but because the threshold for proof is not reached. Alas, it requires in-depth knowledge of a range of interdisciplinary fields. The other proposals that have been proposed for Basque have been neither. In other words, Basque remains an isolate. Basque is the only isolate remaining in Europe, spoken in and around the Western Pyrenees by people who were traditionally fishermen and farmers for many millennia.
One recent study claims that Basque is related to the Dogon language spoken in Africa.
Jaime Martín, specialist in Romance languages, has proposed that Basque is related to Dogon, a language spoken in Mali, Africa. His proposal received quite a lot of attention in the Spanish press. It should be mentioned that the Dogon dialects are so diverse, that it would be better to speak of perhaps five or more distinct languages in a Dogon family. Dogon has been suggested earlier to be part of the vast Niger-Congo phylum in Africa, but more recent expert classifications do not accept this connection anymore (glottolog.org). Therefore, it is better to consider Dogon to be a family of languages, and not part of Niger-Congo but a separate family.
The diversity of languages on Earth is huge and unfathomable. This incredible diversity of languages is also encountered in Africa, the cradle of humankind. In 1963, the brilliant American linguist Joseph Greenberg wanted to bring a system for the diversity, and he claimed that the languages of the continent could be reduced to 4 superfamilies: Afro-Asiatic in the North (e.g. Arabic, Berber, Somali, Hebrew), Khoisan in the South (click languages), Nilo-Saharan in the East and Niger-Congo languages covering most of the continent, from Senegal to Southern Africa, where Bantu languages dominate. In the meantime, Africanists have discovered a range of isolates, and after scrutiny, languages that were supposed to be Niger-Congo were too different, and proof of relation was lacking. Therefore, quite a few supposedly Niger-Congo languages are now considered separate families or isolates. Specialists now count as many as a few dozen language families in Africa, rather than just 4. One of these is the Dogon family. This language family has been proposed to be related to Basque.
One should not be misled by the pigmentation of the skins of the Basques and the Dogons, who are visibly quite distinct. It is known that people with the same skin colour can speak widely diverse languages, and languages of the same family can be spoken by people who look very different. The physical dissimilarities are not a reason to exclude the linguistic connection.
The man who proposed the Dogon-Basque connection is Jaime Martin. In a blogpost he wrote:
“The lexical comparison between the two languages (note: in Dogon, where I have consulted vocabulary from fourteen dialects) has realized more than 2.000 words, with a score of over 1600 pairs of similarity, which represented over 70% of the total”.
He said he “followed the method of word order typology of J. Greenberg”. Typology is a branch of linguistics where one compares structures of languages rather than forms. For instance, do adjectives precede the noun or follow the noun? On the other hand, if one would study forms, one would study how far words like English timber, Dutch timmeren ‘doing carpentry’, German Zimmer ‘room’ and Danish tømmer ‘lumber’ resemble each other in form and meaning.
Then, Martin proceeds with a remark about the typological resemblance of the two families he claims are related:
“The structural comparison of the Basque over three dialects Dogon showed exactly the same word order in the sentence: SOV, GN, NA, ND and use of postpositions, as well as other important common morphosyntactic features.”
These abbreviations mean: Subject-Object-Verb, Genitive-Noun (or: possessor-possessed), Noun-Adjective, Noun-Determiner/Demonstrative. Unbeknownst to him, however, but well known to specialists in the 8,000 languages of the world, is that there is a strong correlation between these four orderings: if the object of a sentence precedes the verb in a language, then almost in most cases, adjectives and demonstratives follow the noun. I think this is also true for languages as far apart as Turkish in the Middle East, Quecha in South America, Hindi in South Asia and Japanese in East Asia. These 4 properties just reduce to 1 similarity.
The claim seems outrageous. But even if so, one needs to investigate it. The author claims it has been outlined in his paper An enigma clarified: source of Basque. Unfortunately, this paper cannot be found anywhere, not even on his own website. It appears, however, that he published a thick book of 621 pages to prove his claim in 2021. Unfortunately, we did not have access to his book. But on the basis of his earlier publications, I have to express some degree of scepticism on the relationship.
When I first heard of the claim in 2014, it was through the media, who enthusiastically embraced the idea. I was sceptical, but I immediately decided to check it out. Spectacular claims like this tend to get more attention in the press than better-argued proposals. The Spanish media were not always objective, and sometimes racist attitudes prevailed with comments that were similar to: the Dogon live in Africa, people in Africa are primitive, and now we know why the Basques are the way they are, and the connection of the languages would prove it. The idea also got some exposure in Basque media. As far as I know, no linguists or bascologists have expressed their views. Perhaps they do not want to waste their time on this.
One way to quickly check the degree of relationship between languages is to look at the lexicon, especially the basic, common words. It is not enough for proof, but lexical similarities are a necessary condition for a relationship. Above I indicated that shared grammatical features are less indicative, as languages with verbs in the end of the sentence often have properties that appear in clusters, and therefore Basque and Dogon share many of the properties with other so-called SOV languages, and thus with each other.
Similarities between words can be studied on the basis of their basic vocabulary. In the 1950s and 1960s the linguist Morris Swadesh compiled lists of 100 and 200 word meanings that would be more stable than others. In the past decades, a group of international scholars selected 40 of these as being the most stable. They hypothesized that one could make a classification of the languages of the world on this basis: automatic classification of the world’s languages. They got quite far with their computational machinery, having gathered data from over 10,000 languages and dialects. They simplified the sound systems of the languages to a more restricted set of speech sounds by collapsing similar sounds under one symbol.
In a 2010 tree generated by the ASJP program, all and only the Dogon languages appeared together in one cluster. Within a wider branch, other languages that are the most similar are found. The cluster was surrounded by languages of the North Atlantic and Mande groupings, African languages spoken around the Senegambia area, as well as the Bangime language, an isolate spoken in Mali, but also Kartvelian languages of the Caucasus in Southwest Asia.
This lexical proximity between Dogon and North Atlantic and Eastern Mande does not imply that these families are indeed related, only that the words of the languages share more similarities. There could be shared loanwords, for instance.
Note that Basque is far away from these language groups, even though a connection between Kartvelian languages and Basque has been suggested, and is still under serious investigation. Basque appears on page 40 of that tree, and Dogon on page 52. This is not particularly close in a document covering 61 pages of language names. Basque appears closest to Papuan languages, languages of the Americas and Nilo-Saharan languages of East Africa. This result suggests that the similarities are accidental and it indicates that indeed Basque is an isolate, and that Basque is in any case not closer to Dogon than hundreds of other languages of the world.
I admit I did not go into depth with Jaime Martin’s data, but here is an example from an early summary of his work:
For the first group, (I), he presents some look-alikes between Dogon (first) and Basque. In the second group, (II), he combines 2 Dogon words with unspecified meanings into one Basque word, and in the third group, (III), he presents 6 words that are virtually identical in Basque and Dogon. He manipulates the meanings a little bit. It may look intriguing, but with thousands of words to choose from, there will always be words that are close in meaning and in form by mere accident. A famous case is Old Greek and Hawaiian, but a close relationship between these two is highly unlikely. It is said that 3 to 5 percent of the roots of any two languages on earth may show (accidental) similarities.
This was in any case intriguing enough to do a test myself. I did this a number of years ago, and have not updated it. I am aware that, for a much more solid investigation, one would have to go through Martín’s book, and I will leave that to others, as the preliminary results point to a complete failure to prove the connection.
I tested the possibility of the relationship between Dogon (columns D-I) and Basque (C) at the lexical level, by looking at “look-alikes”. I compared 40 supposedly conservative/stable words from 6 different Dogon varieties with meaning equivalents in both modern Basque and modern Spanish. This is the result. The English meanings are given in column J. The writing systems of Dogon varieties are simplified. As an additional source of comparison, I added Spanish forms for these 40 meanings in column K. The numbers in column 1 refer to the numbers in a standard Swadesh list.
It is clear that the six Dogon varieties are quite different from one another. On the horizontal axes, I have color-coded forms that are somehow similar, either a rough consonant-vowel sequence, or two consonants in a root. If there is a form in one of the Dogon varieties where two consonants resemble two consonants in the same order in Spanish or Basque, I coloured them yellow. It appeared that 25 of the 40 Dogon forms resemble Spanish, and only 9 of the 40 words resemble Basque. If Dogon would be related to a European language, these data would suggest Spanish rather than Basque. Hence, on the basis of these data, the hypothesis of a Basque-Dogon relationship must be firmly rejected. It seems that consulting his book would be a waste of time, but I would encourage others, hopefully with a solid knowledge of typology, language change and historical linguistics, to test it.
Thus, lexically, Dogon is more similar to Spanish than to Basque. That would be enough to reject the connection. The grammatical similarities are not convincing either: the shared SOV order is the most common constituent in the world, and other orders correlate strongly with verb-final languages (genitive-noun, noun-adjective, noun-demonstrative, postpositions). The connection has to be rejected. Specialists have not responded to the preposterous claims in academic work, as far as I am aware.
Peter Bakker is a linguist connected to Aarhus University. He studied Basque with Rudolf de Rijk at Leiden University in the Netherlands, and travelled extensively through the Basque Country in order to practice. He has published a number of articles on different aspects of Basque, including affix order in the verb, proposed long distance relations of Basque and contact between Basque and Romani and between Basque and Algonquian languages of North America as well as Basque and Icelandic. He is also co-author of a book about Dutch Bascologists.
Swahili: The Rise of an African World Language 
Basque and gender: how a genderless language also suffers from gender inequality 
Basque as an (imperfect) window into the past 
African languages in the 1700s Danish West Indies 
Lingua Franca: Elusive contact language of Christian slave colony based in Algiers finally pinned down 
NOW THAT’S WHAT I CALL LINGUISTIC CHRISTMAS