From The Times

February 20, 2008

Professor Geoffrey Lewis

Scholar who introduced Turkish studies to Oxford and worked tirelessly to foster friendship between Britain and Turkey

Professor Geoffrey Lewis, who was the prime mover in introducing Turkish studies at Oxford and whose 1953 Turkish primer is still the classic introduction to the language, took up Turkish only “as a hobby” on the advice of his Latin tutor. It was to change his life.

He was reading classics at St John's College when his studies were interrupted by the war, during which he served in the RAF as a radar operator from 1940 to 1945, mainly in Egypt and Libya. He pursued his new hobby whenever possible while in the Middle East.

On his return to Oxford after the war, he took the advice of Sir Hamilton Gibb and read Arabic and Persian, being awarded first-class honours in 1947. There was no one to supervise a thesis in Turkish, and he therefore chose to do his doctorate in Islamic philosophy.

In 1950 he successfully submitted a thesis on the so-called Theology of Aristotle, which formed the backbone of his Plotiniana Arabica, published in 1959. His Turkish studies had continued apace, and in 1950 he was appointed university lecturer in Turkish at Oxford. He became in turn senior lecturer in Islamic studies, senior lecturer in Turkish, and finally Professor of Turkish. He was first elected to a Fellowship at St Antony's College, Oxford, in 1961, where he later served as sub-warden and senior tutor. He retired in 1987.

He taught the whole range of the Turkish syllabus from the Orkhon inscriptions to modern Turkish poetry. He was also fully at home in the main Turkic languages and, when required, taught Azeri, Chaghatay, Kazakh and Uzbek. He was a remarkably good teacher in whom a thorough grasp of his subject, ease of manner, and a fine sense of humour made a happy combination. Those who studied under him remember him with admiration and affection.

In addition to his teaching at Oxford, in 1959-60 he inaugurated and directed the bilingual humanities course at Robert College in Istanbul, a course designed to give engineering undergraduates a background in the history and traditions of Eastern and Western thought. 

He continued to act as Visiting Professor of Humanities at Robert College from 1960 to 1968. He was always much in demand both in Turkey and abroad and remembered with affection his spells as Visiting Professor at Princeton (1970-71, 1974); UCLA (1975); as a Leverhulme professor in Turkey (1984), and as Gunnar Jarring lecturer in Stockholm (2002).

Lewis maintained an active interest in a wide variety of subjects within the Islamic field, particularly in philosophy and science, culminating in the publication in 1973 of Albucasis on Surgery and Instruments, written jointly with Dr Martin Spink. He was also a fine Koranic scholar. However, his main academic interests lay in Turkish language and literature and current affairs in Turkey. His first publication in this field was Teach Yourself Turkish (1953, revised 1989), still the classic introduction to the language. This was followed by Modern Turkey (1955). Books like this are now taken for granted, but it was the first book in English written by a scholar who knew Turkish and Turkey. In it Lewis made clear his sympathy for, and admiration of Kemal Atatürk's reforms, though he was the first to agree that much yet remained to be done. The book appeared in several editions, with the 1978 reprint having been almost completely rewritten to keep up with the times.

More books followed, but his magnum opus was his Turkish Grammar, first published by the Oxford University Press in 1967 and revised several times. His final work was The Turkish Language Reform: A Catastrophic Success (1999, Turkish edition 2003). There were also numerous articles.

Geoffrey Lewis was born in 1920 and educated at University College School, in North London, before going up to Oxford. He was chiefly responsible for the establishment of Turkish studies at Oxford, this culminating in 1964 with the acceptance of Turkish as a main subject of study in the Honour School of Oriental Studies. This was achieved despite vociferous opposition orchestrated by two die-hard conservatives from the Queen's College.

The Lewises had a wide range of friends and were generous with their hospitality. They kept open house for Turkish scholars visiting this country and for anyone who sought knowledge of Turkey. For the Turks who found themselves in hospital in Oxford or, rather more rarely, were detained by the Oxfordshire police, they willingly acted as interpreters.

Lewis was much in demand to serve on bodies outside Oxford, and he was unstinting of his time. He gave much service to the British-Turkish Mixed (Cultural) Commission of which he was a member from 1975 to 1995. He was particularly fond of the Anglo-Turkish Society, of which he was vice-president from 1972 to 2003 and president from 2003 until his death; of the British Society for Middle Eastern Studies, which he helped to found and of which he was president, 1981-83; and of the E.J.W. Gibb Memorial Trust, which publishes works on Arabic, Persian and Turkish. His last public appearance was at the trust's launch last month for a new translation of the Koran, which it had just published.

Away from the public eye, Lewis advised a succession of British and Turkish diplomats on ways to further the cause of friendship between the two countries.

Over the years the Turkish Government showed its appreciation of Lewis's work by awarding him a Certificate of Merit in 1973, the Exceptional Service Plaque of the Foreign Ministry in 1991 and finally the historic Order of Merit of the Turkish Republic in 1998. In the same year he was appointed CMG.

He visited Turkey whenever possible. The visits started in 1947, when he spent six months in Turkey to bring his knowledge of Turkish up to a standard that was acceptable to himself, and continued until his final visit to Ankara in 2007 when he lectured for the British Council on the problems of language reform. His contributions to Turkish studies were widely recognised in Turkey, and he was elected a corresponding member of the Turkish Language Society as early as 1953. Later manifestations of this esteem included honorary doctorates from Boaziçi University in 1986 and Turkey's oldest university, Istanbul University, in 1992. In this country he was elected a Fellow of the British Academy in 1979 and an honorary Fellow of St John's College, Oxford in 2000.

Lewis's academic work combined the best of the classically trained philologist with a sense of the importance of literary style, and displayed a range of interests and accomplishments that few could match. His scholarship, driven by a love of learning that owed much to his Jewish background, was informed by great analytic skills and by lucid exposition, and he had a style of learning that combined erudition with wit, sympathy and an old-fashioned breadth of education. Study and research were both his work and his recreation, and he had the humility of the scholar who knows that there is always the yet undiscovered. He said that he regarded the existence of language as proof of the existence of God - “We could never have done it on our own.”

In July 1941 he married Raphaela Rhoda Bale Seideman (Raff), his childhood sweetheart (who was to publish her own ground-breaking Everyday Life in Ottoman Turkey in 1971). She and their daughter predeceased him, and he is survived by his son.

Professor Geoffrey Lewis, CMG, scholar of the Turkish language and culture, was born on June 19, 1920. He died on February 12, 2008, aged 87

Uyghur Latin transcription

This is an article I posted on Unilang a while ago:

'There have been too many attempts to transcribe Uyghur with Latin letters and the idiosyncratic versions only lead to confusion. The Chinese even went so far as to suggest ‘x’ for the ‘sh’ sound and ‘q’ for the ‘ch’ sound, as if it was Mandarin pinyin! Here on Unilang I propose a single, standard transcription for Uyghur as discussed below. Criticisms are always welcome.

First of all, I propose that the principles of Latinising Uyghur should be 1. the transcription should be simple and aesthetically pleasing; and we should avoid digraphs as much as possible; 2. it should show its Turkic characteristics. So we should look to other Turkic languages which have succeeded in Latinising their writing systems for examples.

I suggest that we transcribe ‘sh’ ش as ş and ‘ch’ چ as ç, following the examples of Turkish. Letter ‘c’ should take the same value as in Turkish – voiced alveolar affricate while letter ‘j’ takes the value of voiced alveolar fricative (as in French ‘bonjour’). 

As for the pair غ and خ (voiced and voiceless velar fricative), I suggest that we use the Turkish ğ and the Azeri x respectively. The Turkish ğ, although having no phonetic value in the modern language, is the remnant of an old sound that used to be found in the language and was lost in the history of language evolution. Since the sound is well alive in Modern Uyghur, ğ deserves to be reassigned its old value when it’s used in Uyghur. A good example is the word çağ (time) which exists in both Turkish and Uyghur; the ğ is silent in the former but clearly pronounced in the latter. Different from the Turkish ğ, the Uyghur ğ can start a word, as in ğam ‘sorrow’ (غام) (the Turkish cognate is ‘gam’). There’s nothing wrong with ‘kh’ representing the voiceless velar fricative, as many people do when transcribing Persian, but adopting the Azeri x follows the principle of avoiding digraphs. Moreover, letter ‘x’ is itself the traditional symbol for the voiceless velar fricative: it ‘cognate letters’ are used to represent this sound in Greek and Cyrillic alphabets. Since Azeri, a Turkic language, also uses it, we should follow the example. 


I also suggest that we use the Azeri ‘upside down e’, ə, to represent the mid-open front vowel /æ/which occurs in both Azeri and Uyghur. Most transcription systems don’t distinguish between /æ/ and /e/ in Uyghur, which I find very misleading and irresponsible. They sound rather similar on their own but depending on the phonological context, one can sound more open and the other can sound more close. It is absolutely necessary to make a distinction in writing, as the Arabic system and the Cyrillic system both do. As I said, we look to Azeri because it is a Turkic language, and we need something simple and easy to write. If it was not for this reason, umlaut ä or æ would satisfy perfectly. Phoneticians would complain about the confusion with the IPA symbol for the schwa, but writing is different from IPA symbols that always have to stay in brackets. Since Azeri uses it for both its Latin and Cyrillic versions of writing, we can be confident about assigning this value to a letter that looks like an IPA symbol. 

There is only one potential digraph left, ‘ng’, which is commonly used in many languages to represent the velar nasal stop. Since it’s a convention to use it in languages based on the Latin-alphabet we could well adopt it in transcribing Uyghur. However, confusion arises when it becomes ambisyllabic. In Uyghur, the velar nasal stop remains intact when it is in the coda position of the previous syllable while occupying simultaneously the onset position of the following syllable. It does not break up to n-g as the orthography would visually suggest. For example, the word yengi ‘new’ (cf. Turkish ‘yeni’) is read as ‘ye-ngi’ (the same pronunciation can also be broken down into two syllables as ‘yeng-i’ since it’s ambisyllabic, but no gap is heard), never ‘yen-gi’. However, since visual perception affects oral execution, learners of Uyghur or children would be confused about the pronunciation. Writing it as yeng’i would be worse, since it’d suggest a glottal stop between yeng and i. The Arabic writing system deals with it well, using only one single letter ڭ for this phoneme, and so does the Cyrillic alphabet which Uyghur briefly employed (unfortunately my keyboard doesn’t have this letter, but you can refer to the lyrics posted by zhiguli, where the velar nasal sound is represented by the Russian ‘n’ with a hook). So we should basically use the Latin letter ‘n’ and add a diacritic to it. The Spanish ñ is used by a lot of Native American languages to represent this sound, which I would not suggest that we do, since the Spanish ñ is too well known for its original value – palatal nasal and even though its value is not fixed, the use of it would remind people of Native American languages, whereas here we are dealing with a Turkic language. Turkmen uses ň to represent its velar nasal and it is a possible solution for Uyghur. However, I personally would hesitate to use it because the hacek is characteristic of Slavic languages written in Latin letters. I found in Latvian the letter ņ (n cedilla) which I’d personally prefer to use, because it harmonises well with the other cedillas – ş and ç and is aesthetically pleasing. It is also easier to write, since we write from top to bottom and it is more natural to mark the bottom part rather than going back to the top to add things there. Although the Latvian ņ is really used to represent the palatal nasal of the language, its value is less known outside the language community and consequently it is easier to assign a new value to it when it’s used in another language.'

A few words on Turkic languages

'... An eminent orientalist remarked: "We might imagine Turkish to be the result of the deliberations of some society of learned men"; but no such society could have devised what the mind of man produced, left to itself in the steppes of Tartary, and guided only by its innate laws, or by an instinctive power as wonderful as any within the realm of nature'

- Max. Müller 'Lectures on the Science of Language', Vol. I, Lect. VIII.

'The Turkish tongues are of singular interest to the student of language. They are to him, what the mountains which surround their birth-place are to the geologist; who there can observe many of the vastest operations of nature and their results, naked as it were, and not veiled by the superficial covering which in other less barren countries makes the investigation and tracing out of the various formations so laborious a task.

The Indo-European languages are like an ancient building, where frequent restorations have interfered with the original design, and where finally a universal coat of plaster has destroyed all outward distinction between old and new. In the Turanian structure, on the other hand, every tool-mark is still fresh, the places where the scaffolding has rested are still visible, and we can almost trace each course of the stone-work to its origin in the quarry whence it was hewn.

It may seem strange that a language developed by the rude and nomad tribes of Central Asia, who in their own home have never known how to reduce it to rule (or rather to distinguish the laws through which they themselves had unconsciously formed it), should present in fact an example of symmetry in complexity such as few of the more cultivated forms of speech exhibit. Although its own people would have one believe that it is subject to no rule and almost purely arbitrary (their only notion of grammar being that of Arabic and Persian with which the Turki cannot be made to fit); yet in reality a few simple and transparent rules suffice to account for all its permutations. These rules, possessing an accumulative power, are enough to produce the immense variety of forms noticeable in the Eastern Turki.

We are now learning to believe that even in languages such as Greek, German, or even English, every seeming irregularity is really the result of laws, some of which we know and can trace in their action, and some of which are yet to be discovered. But in Turki we can see them; it is as if the centuries were to flow backwards, and we could watch the building of the Pyramids and solve by ocular demonstration the doubts of the learned as to the method by which the vast blocks were transported form the quarries, and placed in their present positions. We can even detect in some instances a commencement in this Turanian tongue, of the process by which the Aryan languages have been polished down and enamelled, as it were, till they reached their present condition.

Viewed in this light the study of the Eastern Turki is seen to have an interest which is not to be measured by the amount of the commercial or other intercourse likely to be facilitated by it. For the Turkish tongues, a journey eastward is pretty nearly equivalent to a study of the earlier forms of an Indo-European language. In either case we get nearer to the source; and the less literary character of the former makes it easier to approach its origin in space than in time. Rémusat, in his "Langues Tartares", truly says: "Le dialecte de Constantinople est celui de tous qui s'est le plus enrichi, je pourrais dire appauvri, par l'introduction de mots Arabes et Persans; et l'on n'en rencontre que fort peu dans la langue des Turks voisins de la Chine, où l'on peut, pour cette raison, espérer de retrouver l'antique langue Turke dans un état plus voisin de sa pureté primitive."

Valikhanoff (the son of a Kirghiz chief in the Russian service, whose name, Valí Khán, with the affixed Russian patronymic ending off, is significant of Russia's progress among those tribes) writes: "The language... spoken in Kashgar is altogether unknown to European savants', and Prof. Vámbéry, in quoting him, adds that this language 'has incontestably the most primitive words and formations amongst all Turkish forms of speech."'

- Extracts from the Preface to Shaw's 'A Sketch of the Turki Language'

Arabic is science, Persian is sugar, Hindi is salt, and Turki is art!

عربي علم فارسي شکر هندي نمک ترکي هنر

Quote given on the opening page of Robert Barkley Shaw's A Sketch of the Turki Language. The source was noted as 'Oriental saying'. When I first saw it, it immediately reminded me of a similar quote I saw in my Xinjiang travel guide. The difference is that in the version give by the travel book, Uyghur was mentioned instead of 'Turki'. Since Shaw's grammar is for the 'Turki language spoken in Kashgar and Yarkand' the ambiguity doesn't seem to have been clarified. More research should be carried out in order to know whether Uyghur was referred to as 'Turki' at that time. Shaw's book seem to suggest, however, that '... the Turkish of Káshgar and Yarkand which some European linguists have called Uïghur' was 'a name unknown to the inhabitants of those towns, who know their tongue simply as Túrki'. Today we know that the demonym of the people living in those areas is 'Uyghur' and this piece of historical evidence coming from a descriptive grammar book has shed light on the sense of racial belonging shared by the Turkic Uyghur people. There is a tendency for Turkic peoples to refer to their mother tongues as 'Turkish'; for purposes of specification, they usually add the name of their ethnicity: 'Azeri Turkish', 'Uzbek Turkish', even the national language of Turkey, which is known in the West simply as 'Turkish', is esoterically named 'Turkey's Turkish', which seems to be a tautology. It seems that there is an incredible sense of unity felt by Turkic peoples, an extreme example of which is known as Pan-Turkism, an ideology deemed controversial even by some Turkic people. If the word 'Turk' is employed to describe anyone that belongs to a Turkic ethnicity, so should the word 'Turkish' be used to describe their language: any Turk speaks a kind of Turkish, thus the Turks in Turkey speaks Turkey's Turkish, the Turks in Kazakhstan speaks Kazakhstan's Turkish (Kazakh) and so on. The whole Turkic world, in fact, already has a (dormant) collective nomenclature: Turkestan.

However, taking into account how people in the West conventionally use the word 'Turkish' and how Turkic peoples think it should be used, we have to admit that the signifié of the word will become confusing. The word 'Turki', here proposed by Shaw's grammar of Uyghur published in as early as 1878, presents itself as a solution. It should serve as a collective term for Turkic languages. Given the sense of unity among Turkic peoples and the curiously great degree of resemblance among their languages, it wouldn't be an exaggeration to say they are one people speaking one same language with regional variations. In fact, the differences among Turkic languages should have been smaller in the past, which gave rise to the term 'Turki' adopted by some European scholars such as Shaw.

Revenons à la citation. If Shaw didn't make up the quote, which I'm sure he did't, it might have come from a homme de lettres with a good knowledge in at least the four languages. Which Turkic language would 'Turki' have been? Or was it rather used as a collective name for all Turkic langauges? Both are possible. If it's the first case, 'Turki' is most probably Ottoman Turkish or Uzbek, representing Western and Eastern Turkic languages respectively, both being well cultivated languages with a long oral and written literary tradition. Taking language evolution into account,it is probably Chagatay, the language of Ali-Shir Nava'i which Uzbek claims to be its predecessor, rather than Uzbek, that is being mentioned.

So why are Turkic languages, or Turki, regarded as an art? Surely the artistic values of Arabic, Persian and Hindi (here probably Urdu, but I won't go into a detailed discussion on the history of Hindi and Urdu) are not to be neglected. Why is 'Turki' singled out?