Subway Announcements from Around the World.

Bathrobe sent me these links, adding:

It’s interesting to actually hear and savour the sound of these languages (rather than just stare at the written forms) as they are used in making announcements. Of course, the enunciation is much clearer than everyday conversation, which is nice if you don’t know the languages.

Metro Announcements in European Languages: English, French, German, Spanish, Italian, Occitan, Basque, Catalan, Portuguese, Dutch, Danish, Norwegian, Sweden, Finnish, Czech, Polish, Hungary, Romanian, Greek, Bulgarian, Ukrainian, Belarusian, Russian. (The Czech “dveře se zavírají” brings back my happy visits to Prague.)

Various Europe metro announcements: Budapest, Vienna, Berlin, Paris, Oslo, Stockholm, Bucharest, Minsk, Prague, Barcelona, Rome.

Subway announcements from around the world: part 1, part 2 (European Edition), part 3 (European Extra Edition), part 4 (American Edition). As Bathrobe says, these are not as clear or well done; the transcriptions are often incorrect or missing, and Catalan is mistaken for Spanish. Furthermore, each comes with an annoying half-minute introduction. On the plus side, they show the trains.

Thanks, Bathrobe!

Comments

  1. LA’s “Mantengase alejado, las puertas están por cerrar” is etched into my memory even though I never actually heard it clearly, the ending in my head was “porterán”

  2. Has anyone compiled a list of how many languages each city feels the need to include in its subway announcements? The announcements for each stop in the Nagoya subway system are given in Japanese, English, Chinese, Korean, and (at least during earlier visits) Portuguese, in that order. Gov’t PSAs in Honolulu’s TheBus often include Japanese, Chinese, Korean, Tagalog, Ilokano, and several Micronesian languages (the last often hampered by irregular orthographic standards).

    I have lots of nostalgia for Japanese train announcements, and probably know them better than most other domains of language use.

  3. There’s a trend in China, Korea, Singapore, and some US cities to assign station letters and numbers.
    For example, Eunos station in Singapore is EW7, the theory being that the city is crawling with helpless foreigners for whom “Eunos” would be impossibly complex but a random set of letters and numbers would be easy. In Japan they’ve taken to putting these numbers into the English versions of their announcements (but not any other languages).
    Honestly, I really wish they would just shut off the announcements. Riding long-distance trains with no announcements in India, it was so nice not to get squawked at every 3 minutes. And somehow I managed to know where I was even in the dead of the night.

  4. They rendezvoused in the rail terminal east of town, where at least a dozen tracks converged. The train tracks, coming in from secondary stations located at various points inland, dipped underground, passing along ancient tunnels cut from the bedrock and lined with flaking concrete. First the lines ran inward, like long spokes; then they bent, and lined up in parallel just as they reached the passenger hub, forty feet beneath the asphalt streets.

    Heavy plate glass doors, which hummed and slid back whenever a human form or a gust of subterranean breeze passed near, separated pairs of tracks. There were few passengers present. The trains ran infrequently, and there was little reason to tarry at the station. But as Yarec and Ris traced their way cautiously, inconspicuously towards their outbound track, they were surrounded by voices. Scratchy voices, of conductors probably long dead, echoed off the slick tile walls. Their messages, announcing arrivals and departures hours or days ahead of time, overlapped to become an unintelligible morass.

    On track nine… Arriving from Rand… Track six… At fourteen oh seven… The train from Stardhaven will be delayed until… Arriving from… Track change, please take notice… Nine forty-six…

  5. Lars Mathiesen says:

    If the Danish one makes you doubt your hearing, it’s husk at medbringe alle ejendele, not husk at tage alle Deres ejendele med as transcribed. The second one is perfectly cromulent and may well be an earlier or later version; tage med is actually a better pragmatic fit to the situation of leaving a train, but more colloquial than medbringe — on the other hand, it cannot sidestep the T/V distinction and it comes down on V. So both are a bit stilted in very different ways.

  6. For Honolulu, with its special requirement of getting the Hawaiian street names right,

    http://www.honolulumagazine.com/Honolulu-Magazine/March-2011/Voice-of-TheBus/

  7. Lars Mathiesen says:

    I could hold forth for at least, oh, 75 seconds* about the mangling of Copenhagen street names perpetrated by somebody’s voice synthesis software. Which is installed in all the buses, to general annoyance. (Never mind that it ignores the Old Copenhagener shibboleths that used to keep the Jutes in their place).

    _______
    (*) It gets old fast.

  8. AJP Crown says:

    ‘Ere! German London Underground announcements (Victoria Line). They’ve translated all the station names. Ziegelstadt Brixton? Vauxhall is the best.

  9. AJP Crown says:

    Although Kaninchengehegestrasse ist auch komisch.

  10. AJP Crown says:

    “Ich finde es immer noch schade, dass die Viktoria-Linie nicht an Wassertoilette (Waterloo) hält, dafür aber an des Königs Kreuz.”

  11. John Cowan says:

    husk at medbringe alle ejendele […] husk at tage alle Deres ejendele med

    GT in fact recognizes the second one and renders it with an equivalent American English version “please remember to take your belongings with you”, whereas the first it mangles into “remember to bring all belongings”. This is completely wrong because of bring: it wants to be completed with “to your destination” or something, rather than the actual rest of the cliche “when leaving the train”. The authorities don’t give a damn where you bring your belongings to; you can heave them in the trash cans found in better train stations everywhere for all they care. What matters to them is that you take your belongings away from their train.

    This distinction does not prevail in Hiberno-English, where (reflecting different verbs in Irish), take means only ‘transfer possession’, and everything else is bring. “Don’t forget to bring your umbrella” can be used not only when speaking to your absent-minded friend before they visit, as elsewhere in the Anglosphere, but also just before their departure. On the other hand, “Hold my hand, I don’t want anyone to take you” is cromulent from adult to child; I’d probably say “take you away” or “steal you”.

  12. Lars Mathiesen says:

    GT is right here, that was exactly the bad pragmatics I alluded to. Medbringe is more like ‘have along’ — the focus is on having it during a journey (like an umbrella), or at the destination (beer), not on the act of taking them with you.

    It may be a mangling of husk deres medbragte ejendele = ‘remember the belongings you had with you’ which would have been totally unremarkable in the PSA register when I was younger but feels a bit stuffy now.

    I’m still not sure I have completely understood when to use llevar or traer, though.

  13. The Seoul Metro has added announcements in Mandarin Chinese and in Japanese in recent years to the existing ones in Korean and in English, but there are complaints about the usefulness of these announcements because the names of the stations are simply pronounced as in Korean for most stations.

    You can listen to an example here for the station 고속터미널 Gosok Teomineol. The announcement in English gives the name of the station in English as “Express Bus Terminal”, but those in Chinese and Japanese give the name in the original Korean pronunciation.

    This wouldn’t be very helpful for the average Chinese-speaking tourist who will be listening in vain for 高速巴士客运 Gāosù Bāshì Kèyùn. The Japanese-speaking tourist might fare better listening for 高速ターミナル Kōsoku Tāminaru, and katakana spellings of the stations that approximate the Korean pronunciation are available, but it is still jarring to hear largely unmodified Korean pronunciation in the Japanese announcement. To be fair, this also happens to the untranslated Korean names of stations in English announcements. Adjusting the pronunciation of Korean names to the phonology of other languages is still an unfamiliar concept to many Korean speakers.

    Last I heard, they were testing pronouncing the station names as in Chinese for select stations, such as 강남 Gangnam which would be 江南 Jiāngnán in Chinese. Hopefully they will roll this out to the entire network so that the announcements can actually help Chinese-speaking tourists rather than being an empty display of how the city is becoming more international.

  14. Interesting, and I agree with your hope!

  15. Jen in Edinburgh says:

    Scotrail tell you to ensure that you have all your luggage and personal belongings with you before leaving the train, which is nonsensical – what they really want me to make sure of is that I still have everything with me *after* I’ve left it!

  16. But it’ll be too late to make sure of it then. Please ensure you have correctly resolved any PP attachment ambiguities when interpreting train announcements.

  17. Michael Prytz says:

    Back in my dissolute days in Prague, it was my party trick to turn to the stranger sitting next to me at the bar and clearly enunciate “Ukončete výstup a nástup, dveře se zavírají”. It cracked them up every time.

    It was all the Czech I knew.

  18. This is a short example of the Riga tram announcements: https://www.youtube.com/watch?v=GC9w90VxkWw

    When I was in NYC last year, I was surprised to see posters in Haitian Creole about planned maintenance. It took me some time to figure out which language these notices were in.

  19. What I remember from the Seoul Metro (how quickly these things fade) is the fact that station names were given in Korean, English, Chinese, and Japanese (note: Japanese last). Korean is, of course, in Hangul, no kanji. The Chinese is all in characters. The Japanese is a katakana transliteration of the Korean pronunciation.

    I’ve never been to Gangnam Station, but Wikipedia tells me that it gives the following:

    강남
    Gangnam
    江南
    カンナム

    What struck me is the Japanese. I interpreted the use of katakana rather than Chinese characters as the symptom of an extreme allergy to Korean names being given Japanese pronunciations, probably dating back to the colonial period when this was the norm. Make them pronounce names in Korean; don’t let them pronounce them as though they were Japanese. The Koreans would rather have the Japanese puzzle over カンナム kannamu than have them pronounce it kōnan (the Japanese reading for 江南). But with the Chinese name right there and much easier to read than katakana, I wonder whether Japanese tourists etc. are now more likely to use kōnan than kannamu?

    (It struck me, of course, because the English gave me the pronunciation while the Chinese characters told me the meaning of the name. I can read some Hangul so the Korean helped me confirm the pronunciation. The Japanese was virtually useless.)

  20. AJP Crown says:

    English gave me the pronunciation while the Chinese characters told me the meaning of the name

    Pronunciation of names is possibly of more use for a tourist but meaning (is Waterloo Wassertoilette or Schlachtfeld for battlefield?) is a lot more interesting.

  21. AJP Crown says:

    in NYC last year, I was surprised to see posters in Haitian Creole about planned maintenance. It took me some time to figure out which language these notices were in.
    Anything Frenchish on the NY subway would likely be for Haitians not French people. When I lived there, New York subway carriage advertising was (literally) only in the languages of poor people and usually about crime or STDs or diseases for which non-Europeans are at higher risk, like lupus. The transit authority stopped it in the end on the grounds that subway trips were already way too depressing, but it made me wonder why they thought white people only took taxis.

  22. AJP Crown says:

    Here someone with a yearning for more German in their daily life has translated the entire London Underground map for no other apparent reason. And as a bonus for John & Language, here‘s a year-by-year picture of the folded New York subway map.

  23. Bathrobe says:

    This is weird. The link to the German-language map took me to a forbidden-access page at: https://radarsubmissions.apple.com/tube/TfLSillyMaps/londonundergroundmapgerman.png

    In order to access the page, I actually had to manually input the link as http://steveprentice.net/tube/TfLSillyMaps/german_map.jpg

    I’m becoming increasingly disturbed by this kind of strange behaviour on the Internet. Why do links take you to somewhere other than you asked for?

  24. AJP Crown says:

    Ah. I ought anyway to have credited http://www.steveprentice.net/tube/TfLSillyMaps/ where I found it, because he has some great maps. He also says at the top that there’s a mirror site, so that might be a reason for any weirdness.

  25. My dad took me to the World’s Fair in ’64; I wish I’d saved a copy of the map.

  26. AJP Crown says:

    Bathrobe: Why do links take you to somewhere other than you asked for?
    In my case, it’s because I use UBlock Origin which blocks advertising and in combination with chrome ‘Preferences’ blocks dodgy sites. One negative consequence is that it sees LanguageHat and a few other quite reasonable sites as dodgy unless I tell it otherwise; I’ll click on a site and my machine refuses to go there.

  27. AJP Crown says:

    My dad took me to the World’s Fair in ’64
    Lucky. I’d have loved that. It was peak World’s Fair, all downhill thereafter. Expo 67 was ok if you like Moshe Safdie but it didn’t have the technological range, afaik.

  28. Athel Cornish-Bowden says:

    Pronunciation of names is possibly of more use for a tourist but meaning (is Waterloo Wassertoilette or Schlachtfeld for battlefield?) is a lot more interesting.

    A few years ago I was driving from Munich airport on the way (not on the Autobahn) to the Chiemsee, and got a bit confused close to Wasserburg. I stopped at a garage where the attendant spoke little English (and I speak little German) but tried hard to be helpful. She told me to go into the centre of Water Castle and turn right. Water Castle? Ah, right, Wasserburg.

  29. Bathrobe: I interpreted the use of katakana rather than Chinese characters as the symptom of an extreme allergy to Korean names being given Japanese pronunciations, probably dating back to the colonial period when this was the norm.

    Interesting observation. My take is that both Korean and Japanese, having phonetic writing systems at their disposal, are moving more and more towards respecting the pronunciation of the names in the original language. Once upon a time, 伊藤博文 Itō Hirobumi and 豐臣秀吉 Toyotomi Hideyoshi would have been called 이등박문 Ideung Bakmun and 풍신수길 Pungsin Sugil respectively using the Sino-Korean readings of the characters. Now it is universal to follow the Japanese pronunciation and write them as 이토 히로부미 Ito Hirobumi and 도요토미 히데요시 Doyotomi Hideyosi respectively. Likewise, the name of the South Korean president Moon Jae-in (문재인 Mun Jae-in) is written in Japanese in 文在寅 using the Chinese characters but pronounced as ムン・ジェイン Mun Jein instead of the Sino-Japanese ぶん・ざいいん Bun Zaiin.

    In my mind it’s comparable to what has happened with European languages. Apart from some royals and religious figures, names of individuals are not translated to their equivalents but given in their original forms. Whereas Pierre, Pietro, Pedro and so forth might have been anglicized as Peter in the past, nowadays they are referred to in their original forms in English.

    This is obviously more difficult for the Chinese languages that use the traditional writing system. In Chinese, they continue to pronounce both Korean and Japanese names according to the Chinese readings of the characters that make them up where these are available. This means that Moon Jae-in becomes Wén Zàiyín and Japan’s Shinjo Abe (安倍晋三 Abe Shinzō) becomes Ānbèi Jìnsān.

    There must be reasons of practicality as well. Some characters have multiple readings in Korean, and it isn’t always obvious which one to choose. Japanese with its multiple kanji readings must face similar issues to an even greater degree. If Wikipedia is to be believed, the Seoul district named after the old East Gate 동대문 東大門 Dongdaemun has the Sino-Japanese reading とうだいもん Tōdaimon, but someone would have to explain to me the reason for the mix of Kan-on and Go-on daimon. Maybe it’s obvious in most cases for Japanese speakers, but it still introduces a source of error. Simply going by the pronunciation in the original language makes the conversion a lot simpler and unambiguous.

    Of course, the etymology of certain names are obscured if only the pronunciations are given. But in the case of the Seoul Metro stations, the Chinese characters will be given (in their traditional form) as well in multilingual signage, so Japanese speakers can in practice see both the Chinese characters and the katakana rendering. So it is easy for them to figure out that Gangnam or kannamu is the Korean equivalent of the Japanese 江南 kōnan.

  30. In recent visits to Japan I’ve noticed that Japanese names of foreign citizens of Japanese ancestry, like Alberto Fujimori or Isamu Noguchi are now written in katakana. (I wonder if Alberto has a Japanese middle name, as many people of Japanese ancestry do–or once did–in Hawai‘i.) The ideological reason may be recognition that Japanese emigrants need not remain Japanese, either in cultural practices or national loyalties. But there’s also a practical reason: the many-to-many relationship between the pronunciation and writing of Japanese names, especially given names. The kanji for Fujimori and Noguchi can be guessed with very little chance of error, but many other surnames are trickier, and many, many given names have huge potential for error.

  31. @Joel: If the Web is to be believed, Fujimori’s middle name is Kenya (Japanese, but not terribly common).

  32. Interesting. His middle name appears in Japanese and Spanish Wikipedia, but not English, and his names and those of his wife and children are only in katakana in Japanese Wikipedia. I couldn’t find Kan’ya in my old Japanese names dictionary, but managed to find a dozen ways to write it at a Japanese baby-naming site online.

  33. Lars Mathiesen: I could hold forth for at least, oh, 75 seconds* about the mangling of Copenhagen street names perpetrated by somebody’s voice synthesis software. Which is installed in all the buses, to general annoyance. (Never mind that it ignores the Old Copenhagener shibboleths that used to keep the Jutes in their place).

    I wonder about the effect of the pronunciations used in these announcements in the case of names with variable pronunciations. The Seoul Metro stations 선릉 Seolleung (선 Seon + 릉 reung) is often pronounced as if written 선능 Seonneung by many if not most speakers of Standard Korean nowadays, but the announcements as well as the romanization use the prescriptive pronunciation.

    An interesting case is that of the station 학여울 Hangnyeoul (학 Hak + 여울 yeoul). As a compound name, the semivowel y in the second element is supposed to turn into ny to be euphonious and the preceding k at the end of the first element automatically turns into the nasal ng. But while this n-insertion is still observed in familiar words like 색연필 saengnyeonpil (색 saek + 연필 yeonpil, “coloured pencil”), it is no longer automatic for many speakers in unfamiliar compounds (let alone when they don’t catch that it is a compound name).

    The station opened in 1993 and was named after 학탄 鶴灘 Haktan from old maps, with 灘 (which in this context refers to rapids) given the native Korean rendering 여울 yeoul. Hangnyeoul may have indeed been the original name of the stream, but it has not been in use for decades, so I don’t think even long-time residents of the area would have been familiar with it. So this name usually ends up being pronounced as if written 하겨울 Hagyeoul, without the n-insertion. I myself might be pronouncing it that way if not for the romanization reminding me about the n-insertion.

    Apparently, after the Korail announcements recently switched to a text-to-speech system instead of using human voice actors, they are using the pronunciation Hagyeoul. One bus line also uses this pronunciation. The katakana spelling used in the signage was ハギョウル Hagyouru based on this pronunciation. The Seoul Metropolitan Government issued a list of standard katakana renderings of Seoul place names, according to which ハンニョウル Hannyouru was correct, so they were supposed to replace the signage. I don’t know if this has been done yet.

    In Hong Kong, 銅鑼灣, the Chinese name for Causeway Bay, is pronounced in Cantonese either as Tung4lo4waan1 or Tung4lo4waan4. The original reading of 灣 is waan1, but this often changes to waan4 in certain place names, e.g. in 長沙灣 (Cheung Sha Wan), pronounced Coeng4saa1waan4. I seem to recall reading that the pronunciation Tung4lo4waan1 is more popular now for Causeway Bay because that is what the Hong Kong MTR announcements have opted for, but I have trouble finding this information again. I do see that the English Wikipedia gives for Tung4lo4waan1 for Causeway Bay station but Tung4lo4waan4 for Causeway Bay.

  34. after the Korail announcements recently switched to a text-to-speech system instead of using human voice actors

    Why do so many of these use text-to-speech systems? They sound unnatural and are annoying (surely not just to me); what was wrong with humans?

  35. My take is that both Korean and Japanese, having phonetic writing systems at their disposal, are moving more and more towards respecting the pronunciation of the names in the original language.

    This was at least partly politically driven. I had always felt that it wasn’t necessarily because the Japanese preferred it that way, but that China (and perhaps Korea) put pressure on Japan to make it that way. However, this is only my perception.

    Chinese don’t seem to feel the same about this, perhaps because, as you implied, the they still haven’t got a decent, non-obtrusive method of showing pronunciation in a hanzi text. Pinyin is highly obtrusive whereas the old bo po mo fo, which fits in reasonably well, has been phased out (except in Taiwan).

    In fact, quadruple signage in Tokyo stations is fairly similar to that in Korea. Kanji on top; hiragana underneath to give the reading in Japanese; English towards the bottom; and Chinese and Korean in small lettering off to the right of the kanji (Chinese in simplified characters, Korean in hangul giving the pronunciation).

    If you know the characters you will be able recognise the name anywhere, even on a Japanese-language map lacking indication of pronunciation. If you read the English, Japanese (kana), and Korean you’ll know how to pronounce it.

    Only the Chinese end up relying on Chinese characters alone, and have no idea how to pronounce it unless they can read the kana, Korean, or English.

  36. what was wrong with humans?

    Too expensive? Too much trouble to change or make later adjustments when needed?

    In Mongolian banks etc., where people get a number and wait for it to be called, I think I’ve heard synthetic voice announcements ignore vowel harmony. So what should be neg-dugeer (no. 1), hoyor-dugaar (no. 2), gurav-dugaar (no. 3), döröv-dugeer (no. 4), etc., just becomes neg-dugaar (no. 1), hoyor-dugaar (no. 2), gurav-dugaar (no. 3), döröv-dugaar (no. 4) etc.

  37. Too expensive? Too much trouble to change or make later adjustments when needed?

    I guess those would be the excuses, but I can’t believe it’s that expensive, or that much trouble. Cheap bastards.

  38. Bathrobe: This was at least partly politically driven.

    When the current Loanword Transcription Rules (외래어 표기법 Oerae-eo Pyogibeop) for Korean were introduced in 1986, one of the most contentious points was the decision to write contemporary Chinese names (names of people active after the 1911 Chinese Revolution, or place names in use today) according to their pronunciation in Standard Mandarin. These were previously written according to their Sino-Korean readings. So 毛泽东/毛澤東 Máo Zédōng was now to be written 마오쩌둥 Mao Jjeodung instead of 모택동 Mo Taekdong.

    This caused a lot of consternation at the time, but nowadays people seem used to it, not least because South Koreans are becoming less and less familiar with Chinese characters. South Koreans are more likely to encounter the romanized rendering Xi Jinping rather than 习近平/習近平, and most younger South Koreans would not know how to read the characters in Sino-Korean, so 시진핑 Si Jinping is easier than 습근평 Seup Geunpyeong. Ethnic Koreans in China on the other hand never stopped writing Chinese names according to Sino-Korean readings, as they are of course intimately familiar with Chinese characters.

    I haven’t seen any indication that this 1986 decision was political in nature or if there was any pressure from the Republic of China (South Korea hadn’t normalized relations with the People’s Republic yet at the time). But I have read that in recent decades, North Korea’s treatment of Chinese names have changed from Sino-Korean readings to transcribing the Standard Mandarin pronunciation and back again according to the nature of Pyongyang’s relationship with Beijing at the time.

    For Japanese the situation must be different because Japanese speakers of course continue to use kanji and would find it much easier to use Sino-Japanese readings for Chinese and Korean names than for Koreans to use Sino-Korean readings for Chinese or Japanese names. So it definitely sounds plausible there was some external pressure to push Japanese towards phonetic renderings of native names.

    One of the gripes that Koreans have is that during the Japanese colonial rule, native Korean place names were replaced by names written in Chinese characters for the ease of administration in Japanese. Some of the native names have now been revived, including of course 서울 Seoul which used to be known as 京城 Keijō in Japanese or Gyeongseong in Sino-Korean under Japanese rule.

    I know that South Korea wanted Seoul to be written in Chinese as 首爾/首尔 Shǒu’ěr instead of the traditional 漢城/汉城 Hànchéng. The latter corresponds to Sino-Korean 한성 Hanseong which is confusingly also in use in Korean as an alternative name for the city, leading Chinese translations to conflate Seoul National University and Hansung University for example. 首爾/首尔 has now largely been adopted by Chinese speakers when referring to the modern city, I believe, though there must still be some resistance to it.

    I’m not aware of South Korea making similar requests to Japan regarding the rendering of its place names, though it certainly sounds like something that could have happened. As I mentioned in a previous comment, the Seoul Metropolitan Government has published a list of standard renderings of Seoul place names in katakana. This doesn’t mean that it requested Japan to follow its practice, but if it is using such renderings in the official signage, then it will of course influence what Japanese speakers will feel to be the preferred practice.

    One major practical reason for using katakana for the Seoul Metro station names is that not all of them are Sino-Korean; some are native, like 뚝섬 Ttukseom, or hybrid, like 학여울 (鶴여울) Hangnyeoul. I see that in Chinese these names are translated as 纛島 Dàodǎo and 鶴灘 Hètān respectively, in the latter case using the original Chinese characters that the Korean name is based on. But in Japanese there is less of a reason to pronounce 纛島 and 鶴灘 as if they were Japanese names (Tokutō and Kakutan, I guess). So katakana トゥクソム Tukusomu and ハンニョウル Hannyouru it is. Maybe it would be fun to have hybrid translations like Tokushima or Kakuyouru, but it couldn’t be done systematically.

  39. Text-to-speech is techy posturing. It seems like public transportation systems, especially trains, have for a long time needed to give the image of being technologically cutting-edge.

  40. The whole time I lived in Boston-Cambridge, around the turn of the millennium, the newer Red Line trains had artificial voices doing the announcements. They were perfectly pronounced, of course, but recognizably not quite human. The system was also clearly not amenable to modification; when they added an announcement for “Mass Eye and Ear Infirmary” to the arrival notice for “Charles/MGH; Mass General Hospital,” it was recorded by a female employee with a strong Boston accent, making a grating contrast with the mellifluous artificial voice. I had a friend who interned for the MBTA one summer, who had heard stories at work about the voice system, but unfortunately, some of them were so obviously false that I had to conclude that everything he had to say about the matter must be considered unreliable.

  41. I forgot to point out that some Seoul Metro station names are neither Sino-Korean or native Korean but of course loaned from other languages, principally English. 디지털미디어시티 Dijiteol Midieo Siti, “Digital Media City”, is an example. In Chinese this is translated as 數碼媒體城/数码媒体城 Shùmǎ Méitǐ Chéng while Japanese renders it デジタルメディアシティ Dezitaru Media Shiti as expected from the original English.

  42. Lars (the original one) says:

    @Jonseong Park: There are cases where the text-to-speech algorithm just gets it WRONG, and I don’t think those will influence how people speak — what comes out is simply not Danish — and then there are cases where the ‘proper’ stress is not as expected from the spelling, at least for people of my generation, and there the algorithm usually sides with the young and the provincials. But they are already a majority, so maybe hearing it in the bus will speed up the process but it would happen anyway.

    (There are a few streets where it’s not just stress placement that is idiosyncratic relative to their spelling (devoicing, cluster simplification, vowel shortening, fancified spelling) but I don’t remember hearing them in a bus so I don’t know what happens there).

    It’s all just shibboleths to let the proper people recognize each other, so I won’t be mourning their passing.

  43. The change to automated voices using text-to-speech engines irks me because I often listen to the pronunciations of the station names used in the announcements as a guide to which pronunciations are correct or at least acceptable. I didn’t know that Michel-Ange was pronounced as [mikɛlɑ̃ːʒ] in French until I heard the announcements for Michel-Ange – Auteuil in the Paris metro. Singapore’s MRT is great for listening to the announcements in English, Mandarin Chinese, Malay, and Tamil, with the station names having diverse linguistic origins, like Tampines [ˈtæmpəniːz] and Outram [ˈuːtrəm].

    TTS systems don’t handle proper names very well, especially those with irregular pronunciations. I’ll know not to trust the TTS announcements on the Copenhagen buses, then. I don’t really remember if I took buses in Copenhagen during my last visit, and the previous visit when I do remember taking a bus was almost twenty years ago. But I would have been listening to the announcements for clues on how to pronounce these tricky Danish names.

  44. TTS systems don’t handle proper names very well, especially those with irregular pronunciations.

    Exactly, and you’d think that would in and of itself preclude their use in transportation systems where proper names are a vital element (for Pete’s sake).

  45. I just traveled from Indiana to SE KY listening to a TTS system that pronounced “Hindman” as “Hinman” and “Mountain Pkwy” as “Mountain P’kwee.”

  46. I’m not a violent man, but I’m starting to have regrettable fantasies about what should happen to the people who put such systems in place.

  47. Ben Tolley says:

    In the last few years, new buses with recorded announcements were introduced on some routes around the Black Country. The original announcements were by a female speaker with a good Black Country accent. For some reason, these deteriorated over time (I’m really not sure how they managed that – after a couple of years, you’d just get half the name of the stop, or there would be a sound, but it would be incomprehensible. It can’t have been physical damage to the speakers, because some would be perfectly clear).

    Then we got new ones, also recorded, but this time by in an RP accent. That in itself was rather a shame. But the new announcements, instead of giving just the name of the stop, included the the number of the bus route and the final destination: [route number] to [final destination]. The number, to and the destination must have been recorded separately; with the resulting odd timing and stress and an unreduced to, on my usual bus on the way to work, there was a cut-glass voice apparently going, ‘One, two, Dudley! ‘, which made me dissolve in hysterical laughter the first time I heard it.

  48. John Cowan says:

    NYC transit (subways, buses, commuter trains) also records snippets and glues them together, so the pronunciations are right but the prosody is often wrong: “There’s a train leaving at. Twelve. Twenty. P.M. On track. Five”. On the other hand, “Please stand clear of the closing doors” is recorded as a block, and sounds fine, though it uses artificial stress: “PLEASE stand CLEAR of the CLOsing DOORS”. (Which by the way is a perfectly good Germanic alliterative line, even if only one of the stressed words is Germanic.)

    The voices are done by voice actors, radio announcers, and random MTA employees on a volunteer basis. The MTA looks for people who can speak clearly, but a lightweight NYC accent is not a problem: indeed, it may be an advantage. On the other hand, at least one announcer is English, and learned his American accent by listening to the radio.

    Women’s voices are used for advisories; men’s voices for commands (which are usually “Please” + imperative). Tests show this to be the most effective scheme.

  49. AJP Crown says:

    I had a friend who interned for the MBTA one summer, who had heard stories at work about the voice system, but unfortunately, some of them were so obviously false that I had to conclude that everything he had to say about the matter must be considered unreliable.
    Sounds like the margin isn’t big enough for the proof, Brett. Give us one story. I like that Boston accent.

  50. AJP Crown says:

    Women’s voices are used for advisories; men’s voices for commands (which are usually “Please” + imperative). Tests show this to be the most effective scheme.
    Ooooh. Reinforcing stereotypes. I’m surprised they can get away with that in New York.

  51. John Cowan says:

    New Yorkers have the greatest respect for what works.

  52. Yes, if you want adherence to principle go to Portland.

  53. Lars (the original one) says:

    The only one that I can’t stop noticing is when the otherwise pretty good Danish voice on Google Maps navigation needs to pronounce ◯4s — (motor) ring four south. Somebody conflated ◯ with O or didn’t keep the OCR from doing it, as the case may be, and then another rule cuts in that should have been special cased away. The result is O fire sekunder instead of the Ring fire syd that you are trying to get onto

    So now I run with Waze with the Mexican male voice, just for fun. At least he knows better than trying to pronounce road names.

  54. David Marjanović says:

    “PLEASE stand CLEAR of the CLOsing DOORS”.

    Best sung to the tune of Il est né, le divin enfant.

  55. Andrew Dunbar says:

    I came eager to re-hear the announcements from Tbilisi and Taipei after very different lengths of time away… Oh well

  56. You can find announcements from many subway systems on the internet. Here is a video of the Taipei Metro announcements. I remember enjoying listening to the announcements in Mandarin, English, Taiwanese Hokkien, and Hakka on the Taipei Metro and buses.

  57. @Lars Mathiesen:

    I’m still not sure I have completely understood when to use llevar or traer, though.

    Could you give an example of a confusing case? The way I see it, the contrast between llevar and traer is a pretty straightforward one of motion away from the speaker vs motion towards the speaker; you don’t get cases where the deictic centre is shifted towards the agent instead of the speaker, as in this instance of “he brought some friends along [to the recruiting station]”. In Spanish that would only be a felicitous phrasing if the author of the phrase had been in the recruiting station at the time of events or of composition.

  58. Lars Mathiesen says:

    @Alon,

    The most salient distinction in Danish is between hente = ‘fetch’ and tage med = ‘take/bring’ (formerly bringe) — the point is whether you have to go somewhere to get hold of whatever, not whether it will end up where the speech act happens. And then, as you say, the viewpoint in Spanish is the speaker’s, not the actor’s, that probably confuses me as well.

  59. I have no Danish, but if those work roughly like Swedish hämta and ta med, there will be lots of cases where an idiomatic Spanish translation would not use either llevar or traer: “på svenska apotek kan du hämta läkemedel trygg” would be best translated with “recoger”

  60. Lars Mathiesen says:

    Thanks, Alon.

  61. Trond Engen says:

    Alon: “på svenska apotek kan du hämta läkemedel trygg”

    “tryggt”

    (Neuter as adverbal form.)

    Edit: Double g in Swedish.

Speak Your Mind

*