Telephones are no hi-fi audio systems. In fact, their sound quality is rather poor, as everybody knows who has ever been trapped on hold listening to musically accompanied hold-on messages. But it’s sufficient to transmit spoken words intelligibly. Is this true for every language? I’m not a linguist, but I know that in many non-indoeuropean the pitch which with the word is pronounced has a crucial significance. Are there any languages where the way the words are pronounced is so important, and so hard to tell apart, that a spoken sonversation via telephone could not, or hardly, be understood?
Well, I have seen people telephone in Vietnamese that to my tin ear is about as tonal as you can get. So I think the answer is no, the telephone can handle any language.
I’m not a sound engineer, but I think Paul is right. However there’s work being done on voice/sound compression for VoIP applications where the algorithms are so fine tuned they might only work for Japanese and will make another language unintelligible. Or so I think I’ve heard… can’t find a site.
Or cite.
The telephone can’t handle sign languages for the deaf, such as American Sign Language, so the deaf have to use other devices for telecommuncation (such as teletypewriters).
The pheremone chemical language of ants 
Yeah, but these language recognition algorithms are being developed for Internet voice portal application. InfoTalk’s e-t@lk engine is one such language recognition algorithm. They’ll never be used for pure telephony applications - even VoIP telephony.
I don’t know if they can be understood over the telephone or not, but the whistled languages come to mind. Whistled languages developed as a way to communicate over long distances (up to a few miles). As far as I know, all people who use whistled languages use a more familiar sort of spoken language for face-to-face communication. The most famous whistle language is Silbo, used in the Canary Islands, but there are several others around the world.
Well, they don’t NEED a telephone, since the languages were “developed as a way to communicate over long distances”. Seriously, though, what would be the problem of whistling over the telephone?
There seems to be an impression here that pitches aren’t transmitted accurately over the telephone. This is not the case; if it were, the lo-fi music referred to in the OP wouldn’t be recognizeable. The bandwidth of a phone signal may be narrow, but it easily encompasses the fundamentals of any tones that the human voice can produce.
IIRC, the standard for telephone transmission (in the U.S.) tops out at 3 khz, although some devices have a higher frequency capacity.
According to this http://www.peruvianwhistles.com/jasa.html the primary frequencies of Peruvian bottle whistles are comfortably below 3k.
This paper (it’s a pdf) http://academic.sun.ac.za/as/journals/akro/Akro45/vanstekln.pdf says that most human whistling falls between 1k and 4k.
And this site http://www.me.psu.edu/cgi-me82/Feedback/whistlelst.pl in an experiment which I can only guess came from someone with WAY too much time on his hands, shows that only about 5.5% of the 600 or so subjects he tested could even whistle at a frequency above 3k.
So, although I Am Not An Anthropologist, I’d wager that most whistling languages are set with most sounds below 3k – thus, can be transmitted over a standard U.S. telephone.
Yeah, there’s no reason a tonal language would be transmitted any less precisely than any other language. After all, in non-tonal languages, tone of voice is transmitted just fine.
However, it’s always tough to tell /s/ from /f/ over the phone. I’ve read that the difference between the sounds is a sound of a frequency that’s not transmitted by phone lines. So a language could conceivably have similar problems in which the distinctions between several different pairs of phonemes aren’t transmitted, making the language much more difficult to hear over the phone. I’m not aware of any, and I sort of doubt that any exist, but it’s really a matter of degree, and it’s hard enough to spell my last name (which has an S and an F in it) over the phone.
Especially when one uses the long s - ƒ
The problem wouldn’t be the tones, but the blowing. You’d have to make sure that they didn’t blow into the mic/receiver.
The s and f problem is caused by someonewhat by the bandwidth problem. Sibilance is speech occurs at higher frequencies that the phone system isn’t designed to handle. I’ve notice my cell phone (SprintPCS) has a much clearer sound and I’m assuming it’s because they allocate more of their data stream so they can market “digitally clear calls”. I have some recording stuff at the house I can use to record a sample line, something with a lot of “ssss” and “ffff” and “pppppp”, then I could filter and compress it to telephone specs. Oooh, I could also see what it looks like frequency wise on a graph. This might turn out to be a fun little experiment for myself. I hope I don’t get too side tracked this weekend.
I could see some difficulty arriving in understanding many of the African languages that use clicks, such as !Kung or Xhosa, over the phone. From my cursory readings on their phonetics, some of the consonant distinctions they make are exceedingly fine, such as aspirated pharyngeal labialized lateral click versus aspirated pharyngeal palatalized lateral click, and things of that nature. I remember reading a few years back of some wax cylinder recordings that were made of one of the Khoisan languages that has since gone extinct; the researchers were having a hard time puzzling out exactly what was being said, because the snaps and pops on the recording sounded just like the clicks the speaker was making.
I +|-|0u9|-|+ U ƒr0//||3|) 0|| £33+5þ3@|<
(Dang, that was hard to code!)
Outside of English there are no languages that I can telephone intelligibly… And even that has been disputed.
Right, but part of the question is how much the fine distinctions matter. The Khoisan languages, at least, have enormous sound inventories (I don’t know about Xhosa, though - it only picked up clicks through contact with the Khoisan languages, which means it might noot use as many). One of the Khoisan languages holds the record for the most sounds, with I think well over a hundred different consonants (compare to 24 in English.)
That many sounds means that the distinctions are very, very fine. They might be more distorted by the phone, but you’d experience the same thing, to a lesser extent, in any speaking situation. There’s probably either highly restricted syllable structures, or highly multi-syllabic words, or something along those lines anyway - for the language to work, there’d have to be an increase in redundancy elsewhere. All languages are inherently redundant, and I just suspect that a language with as many fine distinctions as the Khoisan languages probably has features to deal with the inevitable confusion of sounds. Which might help some with the phone as well.