Turkish — more specifically, the standard Turkish dialect of Istanbul — is a very boring-sounding language, at least superficially. What I mean by this is that it has no distinct, uniquely-identifying sounds. For example:
- even if you don’t know any Russian, you can tell if people are speaking Russian because of its rampant palatalization (that is, its many y sounds, as in nyet);
- Arabic is easily identifiable by its baffling array of throat sounds; and
- Georgian (along with other Caucasian languages) is distinguished by its varieties of consonants and its many consonant clusters.
Turkish has nothing like this. It’s a very generic blend of Southern European and Middle Eastern sounds, shorn of anything remarkable. The first time I went to Turkey, I was shocked by how uninteresting the language sounded.
I still feel that way to a great extent, but it turns out that there’s a little more to the sound of Turkish than can be heard by a casual listener, namely vowel harmony, a process whereby the vowels within a word match each other in certain qualities.
First, there are eight vowels in the official version of Turkish. They are:
- a, e, i, o, and u, which all sound the same as in Spanish, Japanese, Georgian, etc.,
- ı, which sound like the vowel in “gull”,
- ö, which sounds like the vowel in the English pronunciation of the word Turk, and
- ü, which sounds like the vowel in the French word peut.
These vowels are partitioned three ways:
- Back: a, ı, o, u; and Front: e, i, ö, ü;
- Low: a, e, o, ö; and High: ı, i, u, ü;
- Unrounded: a, e, ı, i; and Rounded: o, ö, u, ü.
Most Turkish suffixes (for noun declension, verb conjugation, etc.) change their form based on the nature of the last vowel of the word to which they attach. These suffixes are of two kinds:
- Low vowel suffixes (A-type). Here the vowel of the suffix is a or e according as whether the vowel of the preceding syllable is back or front.
- Example: The plural marker on nouns is lar / ler, giving kuşlar, “birds”, but dişler, “teeth”.
- High vowel suffixes (I-type). Here the vowel of the suffix is ı or i or u or ü so as to match both the frontness and roundedness of the vowel of the preceding syllable.
- Example: The postposition lı / li / lu / lü means “with” or “containing”, so we have ballı, “with honey”, biberli, “with pepper”, çocuklu, “child-having”, and köylü, “villager”.
(Note that o and ö cannot be the vowel of variable-vowel suffixes, although there is at least one invariable suffix containing o.)
Turkish words tend to employ lots of suffixes one after the other, and vowel harmony affects all the vowels in succession. Consider the sentences Et donduruldu mu? and Et döndürüldü mü?. The first means “Did the meat get frozen?” and the second mean “Did the meat get turned (i.e. rotated)?”. Identical verb suffixes are used (four, if you include the I-type question particle mu), but they all take on different forms because the root don “freeze” has a back vowel and the root dön “turn” has a front vowel. Another nice illustration of this is the word çarpıştırıcılar, meaning “colliders”, as in “atom colliders”. Here the root is çarp “crash, hit, collide” and all the suffixes match it in being back and unrounded.
Actually, these rules apply not just to suffixes, but also to multi-syllable root words. Within such roots, the frontness of the first vowel dictates that of later ones. Animal examples: inek “cow”, boğa “bull”, öküz “ox”, köpek “dog”, deve “camel”, balık “fish”, koyun “sheep”, ördek “duck”, tavuk “chicken”, domuz “pig”.
I don’t know if following these rules is difficult for all English-speakers, but personally I find that that the rhythms of English stress and vowel reduction really interfere with my ability to adhere to them. What do I mean by “vowel reduction”? Consider the word Galatasaray, the name of a high school and a soccer team in Istanbul. English-speakers (from all countries) generally say something like Gılatısıray, putting stress on the second and fifth syllables and reducing the vowels of the other syllables to the weak sound ı (again, like the vowel in “gull”). This sounds ridiculous to Turks.
I find that I tend to do this for many harmonizing suffixes, especially I-type suffixes. For example, the question “Are you Turkish?” should be Türk müsün?, but when I say it it comes out like Türk mısın?. This kind of thing is bad and frequently not understood.
For A-type suffixes, I often try to split the difference and use a vowel that sounds halfway between a and e. This means that I don’t have to think about which vowel to use, while ensuring that my vowel is not too far off from the prescribed one. The sound I use is basically a short a, as in “cat”. I had actually heard this short-a sound used by Turks, but inconsistently. For example, en “the most” rhymes with the English “pan”, but ekmek “food” rhymes with the English “Tex-Mex”. (And when they speak English, Turks generally make “help” rhyme with “scalp”, but they make “dad” sound like “dead”.)
It turns out that while this vowel is merely an allophone in Istanbul Turkish, in Azeri it’s its own vowel, denoted ə. Harmony-wise, ə has the same properties as e (i.e. they’re both low, front, and unrounded), but it remains distinct. In fact, ə replaces e in the rules of Azeri vowel harmony: evdə, rather than evde, for “in the house”.
Anyway, if vowel harmony were easily noticeable by the casual listener, Turkish might sound interesting. But it isn’t, and so it doesn’t. A major reason for this is that there are many common words that violate vowel harmony. According to the rules all the vowels in a sentence should cluster together by kind, but this doesn’t happen in practice, making the operation of vowel harmony somewhat obscure.
Harmony-violating words fall into three main categories:
- Compound words. If a word is composed of distinct words, those words don’t harmonize, as in bugün “today” (= bu “this” + gün “day”) and Karadeniz “Black Sea” (= kara “black” + deniz “sea”).
- We might also include invariable suffixes here, like the continuity marker yor, as in geliyorum “I’m coming”.
- Words of non-Turkic origin. (Note that I don’t say “foreign words”, since some of them are familiar, everyday words.) Turkish words of non-Turkic origin are mostly from three sources:
- Arabic: kitap “book”, beyaz “white”, vücut “body”
- Persian: ateş “fire”, şeftali “peach”, siyah “black”
- Ugly words like kütüphane combine Arabic and Persian.
- French: asansör “elevator”, kuaför “hairdresser”
- In Azerbaijan, vowel-harmony-violating words generally come from Russian rather than French: kino “movie”, spiçka “match”. (These words are often used in Georgian as well.)
- Pan-European words generally take their French forms in Turkey and their Russian forms in Azerbaijan, violating vowel harmony in most cases: televizyon and jeoloji and otobüs vs. televiziya and geologiya and avtobus.
- Oddballs. Some words just don’t follow vowel harmony, like anne “mother” and elma “apple”. I don’t know why they’re like this. From what I understand, it’s mostly an Istanbul thing, with ana and alma being more widespread in Anatolia and Azerbaijan.
It should be noted that different terms are used in Turkish to describe its vowels. The corresponding terms are:
- back – kalın (thick)
- front – ince (thin)
- low – geniş (wide)
- high – dar (narrow)
- rounded – yuvarlak (round)
- unrounded – düz (flat)
So ü would be classified as thin, narrow, and round, while a would be classified as thick, wide, and flat. [Great idea: Somebody should make a chart juxtaposing human bodies with vowels of the same description.]
Frontness harmony is known as büyük ünlü uyumu (greater vowel harmony), while roundedness / height harmony is known as küçük ünlü uyumu (lesser vowel harmony).
Vowel harmony also exists in other Turkic languages, like Kazakh and Uyghur. More generally, vowel harmony exists (in one form or another) throughout the so-called Altaic language family, which includes Mongolian. Vowel harmony is also seen in Finnish and other Uralic languages. It was partly on the basis of this common feature that Turkish and Finnish were once believed to be related to each other as members of a larger Ural-Altaic language family. This idea, however, is no longer widely believed.
By the way, vowel harmony can be observed in Georgian words of Turkic origin. For example, otakhi (ოთახი, room) comes from the old Turkic word otag. The Georgian nominative case ending -i doesn’t harmonize, but the vowels of the base do. Another example is gemi (გემი, ship). In this case the i on the end is not added in Georgian, but is part of the root, harmonizing with the first vowel.