I have been thinking about the following question quite a bit recently: why do other languages, which often do not even use the Latin alphabet, seemingly get to decide on the way their words get Latinized in English, even if the original pronunciation of the letters chosen for their Latinization does not match the pronunciation of those words.

As an example of the mess this gives, let's take the word "chi", which is the Latinization of the name of a greek letter. However, because a 'ch' is usually pronounced like 'tsh', the greek letter would more accurately be represented by "ki", or "khi" for that matter if you want to emphasize the aspiration.

While that one's pretty 'tame', so to speak, it sets me up to talk about the Chinese word "qi", which is pronounced like 'che' (or 'tshe'). The martial art Tai Chi has the exact same pronunciation as 'qi' (or at least, for all intents and purposes, they are indistinguishable in the English language), but that one uses the 'ch' instead, making it even more inconsistent.

Another example is the name of the current president of China, "Xi". His name is pronounced like 'Zhe', so why is the letter X used there? I understand that this resembles the pronunciation of the first letter in "Xavier", but while Xavier simply has a silent 'k' sound at the start (similar to how the k is silent in words starting with 'kn'), the Chinese name "Xi" doesn't have a silent k at all. It's just a 'zh' sound at the start.

Another example is Arabic words with a Q not being used in the same way older English words (loanwords from Latin, mostly) are pronounced. In all of those cases, the q is always followed by a 'u', forming 'kw' when used together. In Arabic words like the countries of "Qatar" and "Iraq", they simply use the q to indicate a 'k'-like sound.

To get to my question: why do other countries, where the Latin alphabet is not used, get to decide how their words are transcribed into English, using the Latin alphabet, even if the letters they choose for their transcriptions do not match the pronunciation typically associated with those letters? If the country of origin gets to decide, what's stopping them from transcribing the word pronounced like 'tshi' into 'po', and claiming that the 'p' is pronounced 'tsh', and the 'o' is pronounced like 'e'?

I understand that English pronunciation and spelling are barely connected at all even when it comes to words that have been in the language for a long time, but as far as I know, that's mostly due to historical reasons, not due to some arbitrary choice of transcription that doesn't have to be this complicated, but seems to just be rather randomly picked.

  • 5
    Xi Jinping’s name is not pronounced ‘Zhe’. Xi is pronounced [ɕi] in Chinese, and the normal Anglicisation of that is something along the lines of the English word ‘she’. And those other countries aren’t deciding transcription (representing spoken language in writing) into English (you don’t transcribe something into a language) – they’re deciding transliteration, which goes from one writing system (not language) to another. In other words, what they decide is rules for how their writing system is represented in the Latin alphabet. If English uses that unchanged, that’s English’s choice. Commented Sep 24, 2023 at 19:24
  • 1
    Of course Xi isn't his first name. It's his last name. Commented Sep 26, 2023 at 13:50

7 Answers 7


In the end, nobody gets to decide except the people who do the writing! When a lot of people start doing something the same way, it eventually becomes a convention. The convention in English used to be to respell foreign words based on their pronunciation, which is how we get "jaunty", but now it's more common to spell them the way the source language spells them, which is how we get gentil (from the same French source).

At the moment, most people transcribing Mandarin words and names follow the pinyin system. Other systems used to be more common, which is why you see Taoism alongside Daoism, or Peking alongside Beijing. But currently, English-speakers generally like spelling things "authentically" (to a certain extent), and leaving the problem of pronouncing them correctly up to the readers. And pinyin is favored by the PRC (and the RoC and the United Nations), so that's what's generally used.

Why do we do it this way? Well, "making it easy to figure out the pronunciation of a word" is not a very high priority for English orthography. If you're using a word like Gesamtkunstwerk while writing about Wagner, it's considered more prestigious to demonstrate that you know German (or at least know German enough to reproduce the spelling) than to help your readers pronounce it; spelling things out phonetically tends to be very stigmatized, and considered a mark of stupidity or ignorance (look into "eye dialect" for examples). Other languages can, and do, make different choices.

(Who exactly is making these choices? The community of writers as a whole. There's generally not a single person deciding on these things; which conventions win out comes down to a whole lot of complicated social factors.)


For your specific examples:

  • Greek chi (χ) originally represented an aspirated /kʰ/, in contrast to kappa (κ) which was unaspirated /k/. So the Romans adopted the transliteration "CH", with the "C" representing the /k/ sound (as the letter "K" fell out of favor), and the "H" representing the aspiration. In Modern Greek, chi represents /ç/ or /x/, which happens to spelled "ch" in German, perhaps helping to reinforce the traditional "ch" transliteration.
  • In Chinese Pinyin:
    • x comes from the Portuguese/Catalan/Old Spanish convention to represent the consonant /ʃ/.
    • q is more obscure. Wikipedia says it's similar to Albanian /c͡ç/, but personally I think that Albanian is too obscure a language for this to be the actual origin. I have my own conjecture below.
  • Arabic distinguishes between kaf (ك) for velar /k/ and qaf (ق) for uvular /q/, the same as in the Phoenician alphabet (𐤊 and 𐤒). European languages generally don't distinguish these sounds, so we just pronounce both letters as /k/, but to Arabic speakers they're different, so they're often distinguished in transliteration as k and q.

As for who chooses the Romanization scheme: Well, there are various possible answers to that:

Also, informal systems get developed by people who have the need to write text in the language but only have a Latin keyboard available. For example, the Arabic chat alphabet or Russian Volapuk encoding.

Sometimes, a system will get the official blessing of the national government of the country where the language originates, as Pinyin did in China. Sometimes, multiple competing systems will coexist, and so, for example, the title of the Islamic holy book can be either Qur'an or Koran.

Typically, the people who develop Romanization systems (or constructed languages like Loglan or Iqglic they'll want their alphabet to minimize surprises so that it's easy to learn. So they'll typically choose letter-sound correspondences from existing languages. Though, they don't necessarily take all of their letters from the same language (which may be your source of confusion).

That's why nobody writes /tʃi/ as "po": It would subvert everyone's expectations.

Now, for some letter-sound pairs, the choice is more-or-less obvious. The letter m is pretty much always a bilabial nasal. The letter t is pretty much always a dental or alveolar plosive, even if its exact phonetic distinction from d (voicing? aspiration? both?) varies.

And then there are other letters that are less consistently used:

  • c is used for both /k/ and /s/ in English, depending on context. Some languages have /ʃ/ or /tʃ/ for the "soft" c. European Spanish famously "lisps" soft c's as /θ/. Other languages use c for /ts/, /dʒ/, /c/, etc.
  • j is typically /dʒ/ in English, /ʒ/ in French, /x/ or /h/ in Spanish, and /j/ in German.
  • q is often dropped from the alphabet altogether, as was the equivalent Greek letter qoppa. In many of the languages that do use it, it's a “redundant” spelling of /k/, which could be written with k or hard c instead.
  • r is variously /r/, /ɹ/, /ɻ/, /ɾ/, /ʀ/, /ʁ/, /ʐ/, or even silent (after a vowel in non-rhotic dialects).
  • w is not used in very many languages. When it is, it can represent /w/, /v/, /ʋ/, or /β/.
  • x variously represents /ks/, /gz/, /s/, /x/, /z/, /ɕ/, /ʃ/, or /ʒ/. Some languages don't use it at all, and respell loanwords using "ks".
  • y tends to be a vowel or semivowel, variously /i/, /j/, /y/, /ə/, /ɨ/, /ɪ/, or /ʏ/. Often, it's a redundant spelling of i.

So, when a person is devising a new spelling system, they'll take one or more of these “redundant” letters and arbitrarily use it for a sound that doesn't have a letter matched with it (like English /tʃ/, /ʃ/, and /θ/, if we didn't have the convention of using the digraphs "ch", "sh", and "th").

And that's my personal conjecture of where Pinyin's use of q for /tɕʰ/ comes from. While using x for /ɕ/ and j for /tɕ/ isn't too far removed from Portuguese x = /ʃ/ or English j = /dʒ/, there wasn't an obvious Latin letter to use for Mandarin's remaining alveolo-palatal consonant, so the linguists who developed Pinyin spelled it with q for no other reason than that q was available.

  • 2
    I have mainly objection to 「Greek chi (χ) represents a sound that doesn't exist in English, but is written as "ch" in German, which may be the origin of that transliteration.」. This transliteration was actually invented by ancient Romans. In classical Ancient Greek, letter <Κ, κ> was pronounced as /k/ and letter <Χ, χ> was pronounced as aspirated /kʰ/. Ancient Romans were transliterating them as <C, c> and <CH, ch> (here with modernized distinction between upper-case and lower-case letters). Modern transliteration of these Ancient Greek letters is usually <K, k> and <KH, kh>.
    – Arfrever
    Commented Sep 26, 2023 at 21:38
  • 1
    In Modern Greek, <Χ, χ> is pronounced as fricative /x/, and spelt as <CH, ch> in transliteration of Modern Greek, but most words of Greek origin in English and other European languages were borrowed from some form of Latin, or created from Latinized Greek elements, and therefore use traditional transliteration rules invented by ancient Romans. E.g. word <diachronic>, with medial part from Ancient Greek χρόνος, would be spelt *<diakhronic> if it used modern transliteration rules of Ancient Greek.
    – Arfrever
    Commented Sep 26, 2023 at 21:51
  • @Arfrever: You're right. A lot of Latinized Greek words are transliterated according to ancient, rather than modern, pronunciation. Hence, we have β=b (instead of v), φ=ph (instead of f), and υ=u or y (instead of merging it with ι=i). I've edited my answer accordingly.
    – dan04
    Commented Sep 26, 2023 at 21:59
  • 2
    Regarding Pinyin's use of <Q, q> for /t͡ɕʰ/, besides <Q, q> being simply available, there is some "diachronic similarity", because <Q, q> is often used for /k/ or /q/, while palatalized /kʲ/ and /kʰʲ/ can develop into /t͡ɕ/ and /t͡ɕʰ/ (or many other consonants), as it happened in recent centuries in Mandarin. (Compare also older borrowing of 北京 as Peking, before /kʲ/ → /t͡ɕ/ change in Mandarin.)
    – Arfrever
    Commented Sep 26, 2023 at 22:28
  • Well, the Pinyin use of bdfgklmnprstwy is unsurprising to English speakers (other than distinguishing plosives by aspiration instead of by voicing). And affricate cjz and fricative hx have precedent in European languages. That leaves just q and v as the only unassigned consonant letters. Maybe /tɕʰ/ is etymologically closer to /k/ or /q/ than to /v/, but I still think that it was an arbitrary choice.
    – dan04
    Commented Sep 26, 2023 at 23:36

Different languages have different phonetics. Not only may it sound differently, but also some differences may have particular importance in other languages. What sounds more or less like "k" or "h" to English, may be a bunch of completely different phonemes in other language. So, when people adopt Latin script to their language they may feel that usual letters are not enough for them. And they start using "unusual" ones. Like X or Q or Y.

For example, letter Q is derived from old Greek letter Koppa (Ϙ, see Koppa), which initially meant a different sound (not "k"). But as it came from Phoenician alphabet, it was actually pretty close to some current Arabic sound, that sounds similar to English "k", but differs. So, it makes total sense to use letter Q for it in Arabic transcription. Or in Qazaq.

In modern Chinese, AFAIK, there is an opposition between different syllables (probably it is called "by the place of articulation"): ci-qi, si-xi, zhi-ji. To distinguish differently sounded (from Chinese language point of view) syllables, they had to use different Latin letters. They tried not to use diacritics, as they need it to show tones (xī, xí, xǐ, xì).

  • 4
    Sibilants in Mandarin actually have a three-way distinction in place of articulation – you’ve shown all three, but inconsistently. It’s between ⟨chi-ci-qi⟩, ⟨zhi-zi-ji⟩ and ⟨shi-si-xi⟩. The ones with ⟨h⟩ are retroflex; the middle ones are plain dental-alveolar; and the ones with the ‘unusual’ letters are alveolo-palatal. Commented Sep 26, 2023 at 12:40

The exact reason can be mostly traced to the choices of a few individuals. For example, Pinyin (a system that uses Q and X) was developed in the 1950's by a known group of linguists, then was adopted and promulgated by the government. The Somali Latin spelling system was initiated by Muse Galaal and Shire Axmed. The details of the first Zulu writings are somewhat shrouded in historical clouds, but we know that the first Zulu Bible was put together by Hans Schreuder, and if one can find a copy of that translation, one can determine whether the x/q/c system for writing clicks was introduced at that point (the most likely hypothesis). In the case of Lushootseed, spelling follows standard transcriptional practices and phonemic analysis carried out by Tom Hess, following his mentors who followed their mentors all working in the Americanist tradition where x = [x] and q = [q]. In other words, the answer will come from an extended historical hunt, and not a particularly onerous one (it may require some travel to archives). There may not be satisfactory historical records regarding the "why".

The primary use of q stems from Ancient Greek where the ancestor of q represented /kʷ/, distinct from /k/. In Romance languages, the digraph qu (from Etruscan practice) was retained while the languages changed (/ki/ → [kʲi] but /kʷi/ → [ki]), leading to the current high dependency of q on following u in English. Use of q in transliteration schemes for Semitic languages is rather old (medieval, not ancient), to the point that it may be hard to determine who first used q for Semitic qoph in Latin transliteration. The historical source of the letter q is in fact the written form of the Semitic phoneme q, though that is probably not a causal factor in the decision to write ق as q.

There are myriad uses as velar or uvular, as well as a handful where q represents glottal stop. These are typically recent developments that arose with the need to develop a Latin alphabet writing system (e.g. Somali). The most divergent version, represented in Pinyin, is the palatal usage of q, also found in Albanian, where the q spelling stems from an orthographic reform in the early 20th century (there being 32 known participants in that congress). It is entirely possible that the similarity between Pinyin and Albanian spelling is non-coincidental, but establishing / refuting such a relation would take substantial historical research.

The story of x is similar. Latin letters derive from Etruscan letters which derive from Greek letters... but there have been variants in what letter was used for what phoneme in what language from the very beginning – due to the excess number of consonant letters, where one can be free to adopt earlier q as the representative for a non-Semitic phoneme. Since x and q are rather non-essential in Latin writing systems, that creates a degree of freedom to press q into service as a systematic spelling of /ʔ, q, ʕ/ or whatever is required.

I have observed (unsuccessful) attempts to press x and q into service as vowels in African language writing systems, but these have never progressed beyond the blog-chat stage.

I do not know how non-linguists pronounce the word Xhosa, though I am certain that is is not with a lateral click (I guess "Hosa" and "Kosa" is what I have noticed from people who can't manage the click). There aren't any widespread Nguni loanwords in English that use x, c, q, but on occasion, newscasters have to utter names of individuals with those letters. Tokyo Sexwale was called "Sex Whale" by one or more announcers, though in his case because he is Venda the letter represents a velar fricative rather than a lateral click as in Nguni languages. English speakers who don't know the details of local-language spelling will therefore invent a pronunciation based on spelling, which is why Beijing is not pronounced pèɪ.tɕíŋ in English, it's pronounced beɪʒɪŋ.

In other words, this is a traceable historical matter, but it depends on the actual source of a word.

  • 2
    I don't believe qoppa was ever used for /kʷ/ in Greek; the labiovelars disappeared before the alphabet caught on. The use for /kʷ/ in Latin comes from the Etruscan convention to use gamma before front vowels, kappa before low vowels, and qoppa before back vowels, which comes from an earlier Greek convention to use qoppa before back vowels and kappa before all others (hence qoppa being used as a symbol of Corinth).
    – Draconis
    Commented Sep 26, 2023 at 16:51
  • 1
    Note that Etruscan didn't have separate /k/ and /g/ phonemes (let alone /q/), so the letters C, K, and Q were interchangeable. This was problematic for the Latin language, which then introduced the letter G for /g/, but still had redundant spellings of /k/.
    – dan04
    Commented Sep 26, 2023 at 22:14
  • Beijing is a good example of an invented pronunciation-based spelling gone more wrong than it needed to: ⟨j⟩ from the Pinyin spelling has mostly become the voiced fricative /ʒ/ as you say, even though the default value of ⟨j⟩ in English (optionally/partially voiced affricate /dʒ/) is much closer to the Mandarin value. Same can be seen in Norwegian where ⟨ei⟩ normally represents /æɪ/, but quite a few speakers pronounce Beijing with /ɑɪ/ instead. Commented Sep 27, 2023 at 0:46

In general, this question may come down to case-by-case situations. You speak about this topic as if a country collectively decides how its language will be transliterated into the Latin alphabet. This is not generally the case. Typically, standard Romanization systems are devised by individuals who have a very good understanding of the phonology of the language as well as the existing functions of each Latin letter they want to use. You should also be aware that these individuals are not trying to transcribe their language for English readers alone - they are trying to transliterate their language for all users of the Latin alphabet.

Because of how the Latin alphabet is used to represent English, the usage of certain letters in other writing systems will seem odder than other letters. Creators of alphabetic writing systems usually want to make orthographies as phonemic as possible, so they will follow a guideline of assigning one letter or one digraph to each phoneme. In present-day English orthography, the letters ⟨q⟩ and ⟨x⟩ do not follow that guideline well. ⟨q⟩ typically represents clusters like /kw/ which could be spelled with other letters. ⟨x⟩ typically represents clusters like /ks/ or /gz/ (eXample) that could be spelled with other letters too. If orthography creators for other languages want to follow the 1 letter/1 phoneme correspondence guideline, they will not use those letters the way present-day English does. If they use those letters at all, it will probably be: reason 1) to represent phonological distinctions that English does not make, or reason 2) to represent with one letter sounds that English represents with multiple.

The ⟨q⟩ example with Arabic is an easily-explained case of reason 1. Latin made a distinction in its voiceless dorsal plosives ([+ round] vs. [- round]) and used ⟨k ~ qu⟩ to represent it. Arabic makes a voiceless dorsal plosive distinction too (velar vs. uvular), so the two Latin letters representing voiceless dorsal plosive letters were chosen. The ⟨u⟩ is not used because labialization (which letters like ⟨u⟩ and ⟨w⟩ inevitably connote) is not a key feature of the distinction. English speakers who do not know Arabic may be confused by this since we do not make a velar vs. uvular distinction and never use ⟨q⟩ without ⟨u⟩.


The letter Q and X are kind of redundant in the Latin alphabet (Q could be replaced by c or k with almost no loss of information, exceptions like the pair cui/qui are just exceptions to that rule, X could be replaced by cs/ks) and so they are "free" for other assignments of sounds (at least, freer than other letters). Many orthographies made use of this freedom, each in its particular way, this is why we have so many different pronunciations of these two letters over the languages of the world.

Latin orthographies have sometimes deep histories, the use of IPA letters for new sounds is a rather new invention compared to these histories. And for practical reasons, it was always a good idea to stick to the basic Latin alphabet as much as possible (think of the costs of making new hardware from printer fonts to typewriters and computer keyboards for additional letters, or for the cost of getting a newly devised letter into international standards).

  • Latin needs the distinction between C and Q because /kʷ/ contrasts with both /k/ and /ku/ (quibus/cibus, qui/cui); if anything, it's K that's redundant. It's mainly in later languages using the Latin alphabet that Q becomes redundant, while C vs K often becomes significant.
    – Draconis
    Commented Sep 26, 2023 at 15:30
  • Right, that's why ubi is a backformation from necubi, from **qubi, because Latin very elegantly distinguished between these by removing all evidence of **qubi lol
    – vectory
    Commented Oct 2, 2023 at 21:11
  • @Draconis: It's use isn't very consistent other since it is more variable than people think it is. For example, in Turkish, soft K is a [c] sound which does confuse foreigners: ijahss.com/Paper/10802016/694782527.pdf. In Indonesian, it can make a glottal stop sound, especially at the end if a syllable. In Swedish and Norwegian, it has hard and soft sounds which both differ and in Icelandic, it represents multiple allophones variants of [k] such as [c] or [x] and even [hk] before l, m and n. In Danish, it can sound like a [g]. Commented Dec 20, 2023 at 5:21
  • In Czech, it is used as a preposition too when it can sound like [g]: Ex: k řece vs k vodě Commented Dec 20, 2023 at 5:22
  • @AkshatGoswami My comment was about Latin, not about Turkish or Indonesian.
    – Draconis
    Commented Dec 20, 2023 at 5:23

With respect to the spellings of (ultimately) Arabic-origin words in English, there are beliefs held by many that knowing the origin of the word is significant to the meaning of the word. I say ultimately because more often than not the word is not borrowed directly from Arabic, and the sounds it is spelled with are still not pronounced in the language of origin. For example, the name Iqbal is very popular with Punjabis in Pakistan but it is pronounced ikbāl. However in English, people with this name will spell it Iqbal in part because someone from another Muslim society will also spell that name Iqbal.

It can be added that this habit was reinforced historically by the British colonial administration, which tended to show a preference towards the Arabicized element of the Indian languages and Persian. Names and words original to the local languages were not given spellings like this. The spelling of Lahore is completely Anglicised since the “e” at the end has no value. The native pronunciation is /lɔ̀r/ but you rarely see it written in English like this even by people who live there.

  • 1
    Do Punjabis in Pakistan spell it ikbāl? English speakers have negligible contact with actual speech in the language of origin (be it Arabic, Punjabi or Persian). What matters most is spelling, and 99.99% of the time, writings like اِکبال are completely uninformative.
    – user6726
    Commented Sep 26, 2023 at 17:22
  • @user6726 Seeing as Pakistan is a former British colony, with a lower literacy rate than much of the world, the British administration was not using spellings used by Punjabi speakers but those used by Persian writers. Most Punjabis I know are illiterate in their native language, and when writing their name in English will use an ad hoc combination of pronunciation spelling and Arabicized spelling. My own name has a q in it but the vowel letters chosen were my parents’ creative decision Commented Oct 2, 2023 at 16:41

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.