On method: translation

Arabic is notoriously difficult to translate into English. It's virtually impossible to do without introducting substantial distortions. Take a simple example: زَيْدٌ طَوِيْلٌ Zaydun ṭawīlun. Translating this to “Zayd is tall” seems obvious enough, but it also obviously introduces a copula “is” that is not in the original Arabic. Furthermore, “tall” is an adjective, while طَوِيْلٌ ṭawīlun is a “noun” (اِسْم ism) - it serves a descriptive function, but for Sībawayhi, “adjective” (صِفَةٌ ṣifatun) is not a first-class part of speech; every adjective is a noun. So a more accurate gloss would be “Zayd, (someone) tall”. It's up to the listener (and speaker) to make the inference to the identity of “Zayd” and “(someone) tall”.

Usually such finickiness is not needed – “Zayd is tall” is quite acceptable, in most circumstances. But when the topic under discussion is language itself, as it is in the Kitāb, a naive idiomatic translation may mislead the reader in fundamental ways, so a more fastidious gloss is sometimes in order. As it happens, the analysis I've just given of زَيْدٌ طَوِيْلٌ is due to Sībawayhi himself, and it is essential to understanding how Arabic works – for example, it is critical to the explanation of the difference between هذَا الرَّجُلُ مُنْطَلِقٌ and هذَا الرَّجُلُ مُنْطَلِقًا (see articles 119 and 120 for details).

Another source of difficulty is the semantic or etymological “commonality” (for lack of a better term) so characteristic of Arabic morphology. Arabic terms whose common “root” is explicitly expressed in the form of the term must often (usually?) be translated by multiple English terms without a common etymology. But that shared root is usually critical to a correct understanding of the text – I take it as a matter of principle that the Arabs (and Sībawayhi) were always sensitive to such root meanings, so idiomatic English translations will often obscure semantic relations that have significance in the Arabic. A simple example: اِسْتِفْهَامٌ istifhāmun, a term Sībawayhi uses frequently, is usually reasonably translated as “interrogation” or “asking” or the like. But it is clearly derived from the root فهم fhm, whose core sense is “understanding”; a more accurate gloss expressing the core meaning of the word would be something like “solicitation of understanding”. He sometimes uses اِسْتِخْبَارٌ to express the same notion of “asking (for something)”, but in this case to root is خبر ẋbr, “news” or “tiding” or the like, hence “to solicit news”.

A more striking example arises from Sībawayhi's use of the root حول ḥwl. Here are just a few of his uses, with commonly found translations: حَال ḥāl “status”; مُحَال muḥāl “wrong”; يَسْتَحِيْلُ yastaḥīlu “to be impossible”; حَالَ بَيْنَ ḥāla bayna “to intervene”; حَوَّلَ ḥawwala “to change”. All of these terms clearly derive from (and therefore express) the root حول ḥwl, yet translators virtually never base their English renderings on a common root. In particular I have yet to find a scholar who does not treat two of Sībawayhi’s most important terms, حَال ḥāl and مُحَال muḥāl as if they were completely unrelated. But for Sībawayhi (I submit) they undoubtedly were related, so a translation that effaces their commonality does the reader no favors.

So a rule of thumb I follow is roughly, always strive to find English terms that preserve the root sense, but when that leads to unacceptably awkward English, use a footnote or other device to alert the reader. Note that “preserving the root sense” does not necessarily imply using only terms with a common etymology. Sometimes – thanks largely to the mixed heritage of the English language, as the offspring of Germanic Old English and Latin (via Norman) – the sense of an Arabic root may be expressible by different English roots. Think of “guts” v. “intestines”. And sometimes it will suffice to find terms from the same ancestor language (usually Latin) whose root meanings are similar even if they do not have a common etymology. So for example I take the root sense of حول ḥwl to essentially involve some kind of cyclical or recurrent change-of-state or “(phase) shift” – it is used to express the passage of years or seasons, as well as change in general, etc. So I reject the interpretation of حَال ḥāl as merely “state” or “status” in favor of “transient state”, and for a translation I'm going (for now) with “circumstance” – derived from the Latin circum “ring”, thus at least somewhat linked to a notion of cyclicality (I may change my mind about this one). For مُحَال muḥāl, which is the past-participle of the “form VI” verb أَحالَ (loosely “to make something shift/change”) I currently favor “distorted”, from the Latin torquere “to twist”, again suggesting the cyclicality of the root sense.

In any case: provide a glossary justifying your translations. I'll get that done eventually, I hope.

The subtlest and most difficult problem is deciding on what exactly Sībawayhi means when he uses a given word. Knowing the (dictionary) meaning is not enough. Even if we convince ourselves that we know exactly what a given word meant in Sībawayhi's day, we are still confronted by the question of just what he thought he was expressing by using it. An obvious example is نَصْب naṣb, the term Sībawayhi uses to refer to inflection using word-final /a/. This is often mistranslated as “accusative”, which is not even remotely accurate. The literal sense of the term seems clearly to be related to the notion of “erecting” or “making something stand up”. But just what he meant by it – and why he used it instead of some other term – is a mystery that continues to baffle Western scholars.

I take this lesson from Jens Høyrup[1]. His topic is ancient Babylonian mathematics, but the same principles apply to any textual interpretation:

Like every technical terminology, that of Babylonian mathematics was ultimately derived from daily language – but often technical meanings cannot be guessed from general meanings, even when these are known. Once we have analyzed the term ‘perpendicular’, it is easy to see how a pending plumb line suggests the idea of the vertical and hence – via its relation to the horizontal plane – of the right angle. Yet etymology alone could never tell us whether verticality or orthogonality is the technical significance; worse, the use of phrases like ‘raising the perpendicular’ support the wrong hypothesis of verticality. Ultimately, a technical terminology has to be understood from its technical uses, and the interpretation can at most be suggested and checked by, but never derived from, everyday meanings. Technical uses, however, may be difficult to understand as long as we do not understand the
terminology in which they are expressed. [p. 5]

Here Høyrup also touches on a principle of interpretation whose importance cannot be overstated: understanding the Kitāb requires a holistic approach. To understand one of Sībawayhi's terms one must examine all of its uses in the Kitāb. Furthermore to understand those situated uses, one must understand all the other words collocated with those uses. In a word, to understand one of his words, one must understand all of his words.

Høyrup introduced the notion of “conformal translation”:

The texts on which the rest of the book are based are bilingual, in Akkadian and in English translation... The translation is meant, firstly, to be a possible basis for discussion of the text without reference to the original; secondly, to serve as support for the reader who wants to be able to follow what goes on in the original without knowing more than the rudiments of its language. For both reasons, the translation should be ‘conformal’, that is, conserve the structure of the original,
rendering always a given expression by the same English expression, rendering different expressions differently... To the extent it does not entail exorbitant clumsiness, the translation is de verbo ad verbum, and word order is conserved... To the extent I have found it possible, terms of different word class but derived from the same root are rendered by derivations from the same ‘English’ (actually often Latin) root. In many cases, the extra- and intra-mathematical uses of a term are intimately connected ... via shared connotations; in order not to make this connection disappear from view in the translation, the conformal principle requires that the English word chosen to render a particular Babylonian term should not only be distinct from the translations of other terms but also inasfar as possible share the everyday connotations of the original term. The result undeniably is a betrayal of decent English style - and the acceptance of what seemed to me to be necessary deviations from the principle of conformality has certainly made me no less ‘falsely true’ than Tennyson's Lancelot. But experience shows me that it is possible for those who do not know the original languages to work in and with this artificial code. [p. 41-2]

My translations are not as fastidiously “conformal” as Høyrup's; a strict adherence to the principles he recommends too often results in barely-readable English. In particular, conservation of Arabic word order often results in indecipherable English. There may be cases where that does reveal something important about the original text, but frequently it does not, in which case preserving Arabic word order would be pointless. For example there is usually not much point in rendering something like رَجُلٌ طَوِيْلٌ rajulun ṭawīlun as “man tall” instead of “tall man”. And there are many cases where preserving Arabic word order would result in English that would be impenetrable to any readers not already capable of deciphering the original Arabic syntax. English with Arabic syntax rarely works. On the other hand, there are many passages where understanding how the Arabic syntax works is important, so in those cases a more conformant translation (with some explanatory notes) may be warranted.

[1] Høyrup, Jens (2002). Lengths, widths, surfaces: a portrait of old Babylonian algebra and its kin. Springer.