3 ms·
> I find it fascinating that languages have different sets of sounds. For example, in English p and b are two different letters, and pat and bat are two differe
by thaumasiotes 7y ago
> I find it fascinating that languages have different sets of sounds. For example, in English p and b are two different letters, and pat and bat are two different words. But other languages don't distinguish them, and their speakers really wouldn't notice a difference if you said pat or bat.
This is related to a question I've been wondering about for a long time.
One of the first stages in language processing is converting a sound's raw physical waveform into a sequence of phoneme tokens. Different languages divide different sounds differently -- as you note, [p] and [b] are distinct in many languages, but not all of them.
What I'd like to know is how nonexistent sounds get classified. Take [θ], the thorn sound. It is rare, but physically similar to several other common sounds -- the stop at the same place of articulation would be [t̪], the closest nearby tongue-based fricative is [s], and [f], while not tongue-based, is even more similar.
All of those are attested substitutions for the difficult [θ] sound. And they appear to be predictable -- English-speaking babies reliably substitute /f/, which they can pronounce, for /θ/, which they can't. Mandarin speakers hear /θ/ as /s/. Cantonese speakers hear it as /f/. (As, I believe, do Russian speakers.) In Ireland, the sound is replaced by some kind of /t/. But all of the languages I just mentioned offer the full set of /f/, /s/, and /t/ as available options.
To a certain degree this is conventionalized. I'd believe that modern Irish pronunciation is based more on what you hear from other Irish people than on a difficulty with thorn. And in English, the Mandarin sound /tɕʰ/ has sometimes been recorded as "ts" and sometimes as "ch", and if you pronounce it to untrained listeners and ask them what they heard, you'll get a mix of both answers. (Today it is less helpfully spelled "q".)
But nobody trains babies that if they can't produce a correct /θ/, the proper thing to do is to fall back on /f/, because /s/ would be a faux pas. The babies just do their best, and they all independently agree that their best approximation to /θ/ is /f/.
I think it should be possible, if you know the phonology of a language, to determine which phoneme (or possibly phonemes) a particular foreign sound would be understood as by a speaker of that language. As things stand today, it isn't -- we know that different languages will cause the same sound to be heard differently, and we know experimentally what the likely options are for a given sound, but we can't predict an individual mapping. And there's enough consistency in what people hear that we _should_ be able to do that.
- fortran77 7y agoB and V switch around in many Spanish dialects, and also often in Hebrew where they are represented with the same letter "ב". (Also P and F, etc.)
- thaumasiotes 7y agoSure, every language collapses multiple sounds into unified phonemes. In modern Mandarin, [v] is a kind of /w/. In Old English, [v] was a kind of /f/. But that's a little different from what I asked about. These correspondences can be learned directly -- a child growing up to speak Old English would have known to treat [v] as a form of /f/ because he would have heard [v] all the time, as a form of /f/. The language had an explicit opinion, and [v] was a common sound. But that isn't the case for foreign sounds. A child growing up in a Mandarin-speaking environment will never hear [θ] at all. Instead, the correspondence of [θ] to /s/ is an implicit rule, driven by the rules that actually do exist. I want to use the explicit rules, like [v] = /w/, to predict the implicit ones, like [θ] = /s/.
- mantap 7y agoTo add to your note about dental fricatives. Arabic has separate consonants for /ð/ (ذ) and /z/ (ز). But it also has a consonant ظ that can be pronounced as either /ðˤ/ or /zˤ/. For me as an English speaker trying to learn Arabic it was confusing because /ðˤ/ and /zˤ/ don't sound the same at all, but evidently to Arabic speakers they do and the pharyngealization makes all the difference. Malay/Indonesian also turns fricatives into sibilants. So /θ/ becomes /s/ e.g. ath-thulathaa /aθ-θulaθa:ʔ/ (الثلاثاء) becomes Selasa. /ð/ would become a /z/ but that sound isn't native to most languages in that part of the world either, so it can become all kinds of sounds including /dʒ/ and even /l/!
- squiggleblaz 7y agoBe careful to distinguish borrowings and substitutions. They are not the same. Contemporary Indonesians substitute /ts/ for Arabic /θ/ if they have only so-so Arabic pronunciation. But this is on the basis of the contemporary transliteration of Arabic. Indonesians do not have the English speakers' concept of spelling - only transcription. In English, they're more likely to use /t d/ but sufficiently many people who speak English have learnt proper Arabic in school that they can pronounce the English sounds. The use of /s/ is not a contemporary substitution, it's a historical borrowing. It's like saying "English substitutes french /s/ with /ʃ/" on the basis of words like "fashion". Yeah, at one point, French had an /s/ sound that sounds like English /ʃ/, but the pronunciations of both has changed and that doesn't happen any more. The replacement of /ðˤ/ with /l/ in e.g. lohor is likewise a borrowing - but it seems to be ancient. ض is reconstructed as a velarised voiced lateral fricative in Wikipedia. And that borrowing is widespread; Indian Muslims going full Indian on me are surprised that I recognise "Lohor" from Indonesian. For my part, I pronounce it as thoohoo (book), but most native Australian English speaking Muslims seem to use /s z z/ for Arabic /θ ð ðˤ/ - they learnt the Arabic words from their Pakistani/Indian parents - and therefore say "zohaw" (hot). Even in recitation - Arabs seem to ask them to stop leading salah~sholat, but that hasn't happened to me. Also, it's not actually /dʒ/ but more of a palatal stop. My wife - and even I - can certainly hear my accent in Indonesian. I also can't hear the difference between c and j unless I already recognise the word.