In the summer issue of the Jewish Review of Books is a short essay by José Martinez Delgado, professor of Hebrew at the University of Granada, telling of his recent identification of a bilingual word list—found years ago in the Cairo Genizah—as having been written by Maimonides. Although news of the publication by Delgado and two collaborators of a scholarly article on the subject has been in the media for weeks, this is the first time he has written about it for the general reader. Intriguingly, his brief account of his discovery concludes with the admission that, when it comes to Maimonides’ motives for compiling the list, “I simply don’t know what he was doing.”

But surely, I thought, one doesn’t just compose lists, whether of words, groceries, household chores, or whatever, for no reason at all, especially if one is Maimonides! I wrote Delgado asking for a copy of his article, which appeared in Spanish in the journal Sefarad, and was graciously emailed a PDF of it. After reading its 73 pages, which bear the title “Un nuevo autógrafo de Maimonides: un glosario judeo-arabe con glosas romances,” or “A New Handwritten Document of Maimonides: A Judeo-Arabic Glossary with Romance Language Glosses,” I can only say that Delgado’s bafflement is entirely justified.

The word list wasn’t discovered by Delgado. Rather, it was originally come across among the vast number of still unexamined Cairo Genizah fragments by the Israeli scholar Avihai Shivtiel and described by him, with no ascription to an author, in a 2007 paper in the British Journal of Middle Eastern Studies. It was Delgado, fifteen years later, who identified its script with known samples of the handwriting of Maimonides—who, born in Andalusia in southern Spain, lived most of his adult life in Fustat, the old part of Cairo, and died there in 1204. This identification, backed in the Sefarad article by comparative photographs, is convincing and has not been challenged, to the best of my knowledge, by any scholar in the field.

Written in ink on folio pages of thick paper characteristic of the late 12th century, Maimonides’ glossary consists of 86 Arabic words in Hebrew characters, which is how Arabic was spelled by Jews in the medieval Islamic world. (This is the main reason for the “Judeo” in “Judeo-Arabic,” since otherwise such texts do not differ appreciably from the ordinary Arabic of the period.) The words are divided into four groups, the first consisting of seven adjectives denoting different colors, the second of twelve adjectives for different tastes or flavors, the third of 22 verbs or gerunds for different activities like eating, sitting, and standing, and the fourth of 45 nouns for different fruits, vegetables, and other foods. Next to 40 of the words appears a gloss or translation of them, also in Hebrew characters, into a Romance—that is, Latin-derived—language.

The bulk of Delgado’s Sefarad article is devoted to examining each of these 40 Romance translations with an eye toward reading it correctly and determining the language or dialect it belongs to—a difficult and often impossible task, as he points out, both because of the problem of pronouncing Romance words written in Hebrew characters and because of the similar vocabulary of the many vernacular descendants of Latin spoken in Maimonides’ age. Thus, in his discussion of the translation of the Arabic word for garlic, tum, in the fruit-and-vegetable section of the glossary, written by Maimonides as אליו , and apparently meant to be pronounced alyo, Delgado reviews like-sounding words for garlic—Latin alium—in Italian, French, Old Occitan, Gallego-Portuguese, Friulian, Catalan, Castilian, and the long-extinct Romance dialect of Maimonides’ native Andalusia (in which the words happens to be alyo, too), without concluding which of these Maimonides was drawing on.

This might lead you to object, “But wait a minute! If alyo was the word for garlic in the Romance speech of Andalusia, in whose city of Cordoba Maimonides spent his childhood, why not assume that אליו reflects it? After all, although his mother tongue was Arabic, as it was of all Jews living in Muslim-dominated Andalusia at the time, he must have heard some Romance Andalusian spoken by local Christians as a child. Isn’t it sensible to suppose that, if years later in Egypt he decided for some reason to compile an Arabic-Romance word list, this is what he would have based it on?”

Alas, it might be sensible but it doesn’t fit the facts of the case. Take Maimonides’ translation of the Arabic word abyad, “white,” as ביאנק , with a tsere, the Hebrew sign for the vowel “ey,” under the bet—that is, beyank. Delgado comments that “this form does not come from the original Latin [word for white] albus, which was preserved in Romance Andalusian albo.” In fact, it was only in Romance Andalusian, Delgado observes, that Latin albus was preserved, since in all other Romance languages, such as in Castilian blanco, Portuguese branco, Catalan and French blanc, Italian bianco, Corsican and Sicilian biancu, Milanese bianch, and so on, it was replaced by a word deriving from German blank, “bright” or “gleaming.”

In short, while most closely corresponding to Milanese bianch, Maimonides’ ביאנק is similar to words for “white” in every Romance language except Romance Andalusian! Statistically, writes Delgado, while twelve of Maimonides’ 40 glosses are attested to in Romance Andalusian, and another fourteen, though unattested, might conceivably be Romance Andalusian words of which no written record has survived, at least eight others cannot possibly be Romance Andalusian at all.

Can the Romance glosses on Maimonides’ list possibly be taken from Castilian Spanish, which he might also have heard spoken in Andalusia? Here, too, the facts don’t bear this out. Milanese bianch as opposed to Castilian blanco is not the only instance in which some of the glosses are closer to Italian than to Spanish. This is especially true of their plural formations, in which, when plurals appear in the glossary, Maimonides prefers their Italian to their Spanish endings.

Consider the case of Arabic tin, “figs,” which Maimonides translates as ,פיגי fijey, with a ḥirik or “ee” vowel beneath the peh, a tsere beneath the gimmel, and a rafeh or horizontal line above both peh and gimmel to indicate that they are to be pronounced “f” and “j” respectively. Throughout the Iberian Peninsula, with its Portuguese figo, Old Castilian figa, and Romance Andalusian figo or fiko, this word was pluralized with a final “s” as figos, figas, and fikos. So are Spanish and Portuguese nouns generally. Italian nouns, on the other hand, are pluralized by an “i” or an “e,” as in Italian fico/fichi—and this, as elsewhere, is the plural form preferred by Maimonides. (The i-e/s distinction derives from the difference between the plural endings of the Latin first and second declensions, which prevailed as the sole plural forms in Italy, as opposed to the ending of the Latin third declension, which came to be the norm in Portugal, Spain, and France.)

And yet apart from the fact that nothing in Maimonides’ biography suggests even a passing exposure to Italian, he wasn’t translating his Arabic words into it, either. To take one of many examples: when he renders Arabic gulban, “pea” or “bean,” as פזול or fazol, with a rafeh over the peh, this closely resembles Castilian fàsol and Old Occitan faizol (from late Latin fasiolus) but not Italian fagiolo. The Italian-like elements in the glossary are only partial.

On what identifiable Romance language, then, is the glossary based? As strange as it may seem, the answer would seem to be: none at all. The list is eclectic. Some of its words appear to be taken from one Romance language or dialect, others from others. There is no consistency.

If Maimonides was recording words remembered from his Andalusian childhood, this would make no sense at all. There is, however, Delgado suggests, a second possibility—namely, that he was relying not on his own knowledge or memory but on one or more Romance speakers whom he consulted and who might have lived in different Romance-speaking regions before coming to Egypt and have provided him with words from them. Delgado even tentatively nominates a candidate: the southern French rabbi Anatoli bar Yosef, who settled in Fustat, where Maimonides befriended him, sometime before 1200, before which he lived in Sicily and possibly other parts of Italy. And yet, Delgado concedes, though Bar Yosef’s profile for the role is “ideal,” there are few traces in the glossary of either Sicilian or the Provençal of southern France.

No less puzzling than the question of where the Romance words in the glossary came from, or of why half of its Arabic terms have no Romance translation at all, is the question of why Maimonides compiled the list in the first place. What made him do it? It wasn’t, as we have seen, to record words remembered from his childhood. Nor could it have been to learn to speak or read a Romance language, because the list is far too limited for that and has no elements of grammar, without which no language can even begin to be learned. If anything, the glossary reminds one of those language booklets for tourists that contain words and phrases like “hello,” “goodbye,” and “Can you tell me where the nearest bank is?” Was Maimonides, at the age of sixty, planning a European trip, for which he wanted to be able to communicate with vendors in the marketplaces? This hardly seems likely.

In the end, Delgado concludes in his Jewish Review of Books essay, the only explanation he can think of is that “perhaps Maimonides was just playing around.” This, too, strikes me as improbable. As I wrote to Delgado, “Maimonides was hardly a playful type, and to tell the truth, I can’t recall a single line in his Guide of the Perplexed that reveals an even slight sense of humor.”

Delgado wrote back:

It saddens me that you do not see Maimonides as capable of having fun. Do you really believe that he is only and exclusively what he wrote? I, too, take my work seriously and separate it from my personal life, but this doesn’t mean that one day someone won’t find a piece of my personal writing and realize that I was not just a deeply boring gentleman who only knew how to write about Hebrew morphology. Please give Maimonides a chance as a human being: you will discover a wonderful person capable of improvising very complex lists that he translates with linguistic games, a real challenge to himself and to everyone who reads it.

A challenge it is. But while Maimonides may have been quite capable of “having fun”—one can certainly imagine him getting down on all fours to let his grandchildren ride on his back—there is nothing in his Judeo-Arabic/Romance glossary to indicate a “linguistic game” of any kind. It’s all one small mystery.