What languages should you learn?

Andrew Sullivan links to a piece advocating for French over Mandarin. Very interesting piece, and the author may be right about the virtues of French, but I wonder if he underestimates Mandarin. He argues that the problem with Mandarin is the writing system:

Fewer and fewer native speakers learn to produce characters in traditional calligraphy. Instead, they write their language the same way we do—with a computer. And not only that, but they use the Roman alphabet to produce Chinese characters: type in wo and Chinese language-support software will offer a menu of characters pronounced wo; the user selects the one desired. (Or if the user types in wo shi zhongguo ren, “I am Chinese”, the software detects the meaning and picks the right characters.) With less and less need to recall the characters cold, the Chinese are forgetting them.


As long as China keeps the character-based system—which will probably be a long time, thanks to cultural attachment and practical concerns alike—Chinese is very unlikely to become a true world language, an auxiliary language like English, the language a Brazilian chemist will publish papers in, hoping that they will be read in Finland and Canada. By all means, if China is your main interest, for business or pleasure, learn Chinese. It is fascinating, and learnable—though Moser’s online essay, “Why Chinese is so damn hard,” might discourage the faint of heart and the short of time.

My thought here is that if the Chinese are failing to learn traditional Chinese writing, and using the Roman alphabet to get a computer to write it for them, that creates good conditions to make a complete switchover to the Roman alphabet. The ability to use a computer to come up with characters other people can’t recognize in print isn’t a terribly useful ability, after all. So there’s that barrier gone.

Mandarin’s big advantage is sheer number of speakers. Check out Wikipedia’s “world language” page–after English, no other language comes close. In fact, there are more native Mandarin speakers than native English speakers. As things stand, this can be discounted somewhat because of China’s poverty. It’s harder to sell stuff to people who are really poor, so poor people matter less from a business-language point of view. I also suspect most tourists, if they’re honest with themselves, aren’t eager to spend a lot of time in the desperately poor Chinese countryside. This will change, though, as China plays catch-up with richer countries.

Now I don’t exactly think Mandarin is the language of the future (as Sullivan puts it), but maybe in the future we’ll all talk like the characters in Firefly. Though for myself, I’m just hoping Korean turns out to be a marketable job skill.

When passing a law is the easy route
Abolitionism vs. reformism
Pimsleur > Rosetta Stone
Slavery abolition and animal rights: the biggest problem
  • Kemist

    In Canada, most especially in areas like Toronto and Vancouver, Mandarin is already a frequent second or third language of choice among high school students.

    There are lots and lots of chinese immigrants in Toronto and Vancouver, and many big companies (in areas like mining and oil) are owned at least partially by chinese interests. This makes learning Madarin or Cantonese worthwhile for business students as well as those who work in healthcare jobs, such as nurses, doctors and caretakers in old people’s homes.

    Many job offers in these area demand proficiency in one of those languages since old patients or residents are first generation immigrants and haven’t learned english, or don’t speak it well enough to express themselves properly.

  • http://www.facebook.com/profile.php?id=611455454 boselecta

    I tried to learn a bit of Scots Gaelic a while back. There is no adult alive who is a native Scots Gaelic speaker who is not also perfectly fluent in English, so its practical usefulness is essentially nil, but it was still fun to try (and I’d be very sad to see the extinction of the language that those wonderful puirt-a-beul songs are in). At least it uses the same alphabet as us already.

  • Jon H

    Korean as a marketable language? The only job I can think of where that would be true is teaching English in Korea.

  • lordshipmayhem

    I’d go for Fortran and Cobol. They may be borderline obsolete, but it’s amazing how many companies still have mission-critical applications that were written in them.

    • Kemist


      Indeedy. The company where I work desperately wants a Cobol programmer to update their obsolete data management system. They’ve been using the same one since the early ’80s.

      • machintelligence

        See the Dilbert cartoon from Nov 4 1997. I believe that was when Bob the dinosaur was introduced. (Go to Dilbert.com and search for “cobol programmer”).

  • timberwoof

    But where are those Mandarin speakers? Although English is not the language spoken by most humans, every human knows someone who speaks it. If I had to teach aliens one language, it would be English. (If I suspected they were up to no good, I’d teach them Monty Python Hungarian Phrasebook English.)

    lordshipmayhem, it’s been said that since Cobol causes brain damage, teaching it should be a capital offense.

  • Mr.Kosta

    Ich komme aus Spanien, und ich spreche Englisch und ich lerne Deutsch gerade.

    (Translation: I’m from Spain and I speak English and I’m studying German)

  • http://kagerato.net kagerato

    Chinese hanzi (and their counterpart Japanese kanji) are entered phonetically, so at least that much is factually correct. The reason is awfully simple: can you design (let alone build) a practical keyboard with two to three thousand keys? No one’s managed it yet.

    I’m sure people have thought of this before, but you might be able to circumvent the issue if you worked with hanzi using breakdown by radical. (Radicals are basically sub-elements of the character. At the simplest level you’re talking about an individual stroke or line.) This would be an especially good system for entering characters you know only by their shape. (If you want to enter them now, you basically use a paint-like program to draw the character and then it will be looked up in a dictionary based upon your stroke order, direction, similarity, and so forth.)

    I doubt such a system would be strictly faster than phonetics, and it conveys a penalty against the illiterate, so I cannot recommend it.

    In any case, thinking we’ll be free of ideograms anytime soon is just wish fulfillment fantasy. The only feasible way to do it is by government dictate, such as when the Korea’s King Sejong decided to replace hanja with the alphabetic Hangul script. That was over five centuries ago, and a very different time and place.

    It’s also important to note that any advantage you might gain with such a move not only takes centuries to play out completely, but would also probably be out-weighed by simply promoting bilingualism of a alphabetic language. In the latter case, you clearly gain in competitiveness and don’t have to deal with nearly as much criticism of sacrificing ‘traditional’ culture, whatever that is really meant to be.

    • Michael

      Taiwan uses the bopomofo system to enter characters. It does not use the latin character set but rather characters derived from hanzi similar to how katagana and hiragana were derived in Japanese (but without the use as a character set in and of themselves).

      The mainland uses Pinyin and other places sometimes use Wade-Giles.

      I use Pinyin (with tones) and sometimes I use a radical method to enter characters if I do not know the pronounciation. I have also used a stroke order method and a stroke count method to enter the character I want when I know it and the computer fails to pick the right character out (usually because I put in the wrong Pinyin – but not always).

      Any romanization of a character is however, much less meaningful than the actual character. The characters carry meaning that is not represented by pronounciation alone, and ofter a character may have multiple pronouciations and multiple meanings. Romanizations are by and large a quick cheat and are not up to the heavy lifting of the Chinese language.

  • Monimonika

    This may not be too much of a problem in French (I have no clue!) but in English there is a particular problem that is made worse by the advancement of computer typing. Spelling.

    If not knowing how to hand-write a Chinese character but being able to read it is considered a problem, then not being able to spell English words correctly without spell-check is just as much a problem.

    Again, I have no clue if French does or does not suffer a similar problem. I doubt it does, though, given that English spelling is the result of mixing languages and mix-&-matching phonetic spellings with differing pronunciations over time.

    …guess it just goes to show that languages evolved over generations rather than were intelligently designed.

    • Ysanne

      French spelling has very clear rules.
      They’re a bit less straightforward than those of languages with very simple phonetic spelling, e.g. Hungarian or German, with one specific sound for each letter, and only a few combinations that stand for special (but intuitive) sounds different from what pronouncing the letters in order would produce. But once you learn these rules, you can read pretty much everything without having to think and fiddle, and have a good shot at writing most things you hear correctly.

    • lordshipmayhem

      Yes, British spelling can be quite strange.

      I’m going to have some tuna phyche for lunch now.

  • piero

    Let’s see:

    I think everyone should learn Italian, just because it is beautiful. It has a very simple phonetic structure (consonant-vowel-consonant-vowel) which makes it eminently suitable for singing. How could you hold a note if the word you are singing ends in “t”, for example? That’s one of the main reasons why opera was mainly an Italian creation. It has no harsh sounds either, like “j” in Spanish or “ch” in Gaelic.

    I find Latin poetry hauntingly beautiful. You can hear a good reading of Catullus here:

    I’m not particularly fond of French: too many letters are wasted to express what amounts to a mere phoneme, too many letters are silent. For example “eau”, “Jacques”, “jeunesse”.

    Spanish has the great advantage of being phonetically fool-proof: you cannot mispronounce a Spanish word, because ortographic rules are strict and there are no exceptions. So even if you’ve never heard a Spanish word before, you can always read it correctly. Compare that with “Fotheringay”, for instance, which is pronounced “Fungy”. Crazy!

    Despite its monstrously inadequate spelling, English has several advantages: its verb structure is extremely simple, and it has an amazingly vast and precise vocabulary. In Spanish or Italian there is no way to distinguish between “jump”, “skip” and “hop”.

    Which language is worth learning? Who knows. It is not a linguistic problem, but a political one. English is widely known just because it so happened that the British empire was immediatly replaced by the American empire. Latin was spread by the Roman empire. Esperanto was a failure because it had no army behind it. The next universal language will be the one spoken by those who win the third world war.

    • Ysanne

      It is not a linguistic problem, but a political one.

      True, however it does help that basic English grammar is simple to the point of non-existence. :-) Plus, when you’re starting from some European language, you already know a lot of words, just because English stole from so many other languages.
      Even with very basic and slightly wrong pronunciation, a beginner can express simple things understandably, which is a huge advantage over tonal languages. And if you don’t know how to spell something, you at least have a shot of spelling it phonetically so a reader can figure out what you meant. Conversely, you can at least have an educated guess what a written word might sound like.

      No chance of any of these things in a tonal language with somewhat complicated grammar and little pictures to encode whole words.

      Being idiot-proof for learners is not what makes a language useful, but it sure helps when you’re trying to teach it to non-geniuses. :-)

      • piero

        Yes, tonal languages are a royal pain in the arse. An a pain in the ears too, at least to mine. Chinese music was severy hampered in its development precisely by the tonal nature of the language.

        • http://kagerato.net kagerato

          I’ve never cared for the sound of Chinese, spoken or sung. Japanese, on the other hand, sounds like poetry in motion. It has a phonetic structure similar to the one you were describing for Italian above: each syllable is either one of the five vowels or the combination of a consonant and a vowel. The only exception is syllable ‘n’, a consonant used in words like “senpai” (upperclassman) and “senritsu” (melody).

          Vowels in Japanese can also be held for twice their normal length, which increases variation. An example would be ‘oozora’ (large/broad/wide sky). In music, there are certain words where the long vowel will often be pronounced according to its constituent parts for effect. One example off the top of my head is ‘eien’ (eternity), which in speech is normally said ‘EEH-EHN’ [long e, short e, syllabic n], but the alternate pronunciation ‘EH-E-EHN’ [short e, short i, short e, syllabic n] is also considered correct.

          I’m not really convinced about whether the structure of the language affects developments in music, though. That seems to be more a matter of priorities and taste than anything scientifically demonstrable.

          • Monimonika

            A quick google search turned up this preliminary paper asking, “To what extent does speech melody influence the composition of song melody in tone languages?”:


            “Singing in a Tone Language: Shona” by Murray Schellenberg of the University of British Columbia.

            The paper notes that there are basically 3 theories of how tone languages affect music melody:

            1) Speech melody determines sung melody
            2) Speech melody and sung melody are unrelated
            3) The middle ground

            and that numerous past studies looking into various tone languages are all over the place in their conclusions of which theory most applies.

            Theory #1 has support in a study of 4 contemporary Cantonese songs and another study of Cantonese opera.

            Theory #3 has this in support (sorry for the long quote, you can skip to the bolded bit at the bottom):

            Wee (2007) offers a different interpretation of the middle ground. He proposes that speech melody and musical melody correspond in positions of metrical prominence; that is, that syllables on the most prominent beats of a bar will match for lexical tone and melody (taken as a pitch relative to both preceding and following pitches). He supports this claim with an analysis of 10 randomly chosen folk songs in Mandarin; finding conformity to be 97.2% (649 syllables out of a total 668).Richards (1972) makes a somewhat similar observation in Hausa. He notes that much of the song is based on two “melodic sentences” (p. 140) which he categorizes as A and B. These two sentences appear with variations throughout the song. He does not make it clear in his transcription just what the relationship between these two musical phrases is but he shows that the vast majority of parallelism between lexical tone and song melody occurs at the beginning of the A-type melodic sentences. The general claim of this interpretation is that lexical tone is only marked at certain, strategic points in the music and the rest of the tones are assumed from context.

            But, yeah, it’s pretty much all dependent on a combination of what language and type of songs are studied, as well as how stringent the study criteria is.

  • http://www.facebook.com/gojaejin jeremygoard

    @Jon H:

    Korean as a marketable language? The only job I can think of where that would be true is teaching English in Korea.

    LOL. I’m an English teacher in Korea with intermediate Korean, and if anything it’s the job (chatting with teachers in the lunchroom, or with parents) that helps my Korean, not the Korean that helped me get the job. It’s my fervent hope that Korea will soon grow into the kind of country able to demand at least basic competence in its language, but as things are most teachers don’t speak a lick and most schools don’t expect it.

    That said, Korea is a very nice place to live, and I envision transitioning into a respectable job where the language really is a requirement.


    Which language is worth learning? Who knows. It is not a linguistic problem, but a political one.

    There is a correlation, though. Languages of empire tend to simplify grammatically, due to the influence of adult learners over their history. Old English got simplified by vikings, simplified again by French speakers, and today has more nonnative than native speakers, who are in the process of shaping a simplified “world English”. Mandarin, Spanish and French have undergone similar processes. So has Persian, which is much simpler than the closely related Pashto. Other things being equal, the more practical languages are also going to be a bit easier.

    • piero


      I’m not sure I agree. Latin was surely simplified, bur after the Roman empire fell. Italian, Spanish and French did away with declensions and replaced them with prespositions (though German, for reasons unkopwn to me, kept the declensions).

      Besides, I’m not sure that prepositions are such an improvement. In Latin, word order is immaterial, whereas in modern European languages it determines meaning: “Man bites dog” and “Dog bites man” have totally different meanings: in Latin you cannot pull that trick: “canem vir mordet” and “canis virum mordet” keep the word order, but the meaning changes according to the declensions.

      English has not, in my view, been simplified by its diffusion. If anything, it’s become more complicated, at least in its spelling. A native Spanish speaker can read Chaucer fairly correctly, but not any sample of modern English.

      I still love English, though; it is extremely flexible, it has the largest vocabulary of any language I know, and when spoken by a professional speaker it sounds melodious and rich.

      • Ysanne

        Besides, I’m not sure that prepositions are such an improvement. In Latin, word order is immaterial, whereas in modern European languages it determines meaning: “Man bites dog” and “Dog bites man” have totally different meanings: in Latin you cannot pull that trick: “canem vir mordet” and “canis virum mordet” keep the word order, but the meaning changes according to the declensions.

        Ummm… that’s not “modern European languages”.
        That’s some, such as English and French, with their absence of declination: This forces you to use word order to distinguish subject from object, since there are no other clues (such as obligatory case-specific articles, case-specific word endings, etc).
        While a non-standard word order may sound a bit Yoda-like in the other languages I know, it has its perfectly acceptable uses in placing emphasis on different aspects of an action.
        In your example, both “Der Hund beisst den Mann” and “Den Mann beisst der Hund” mean the same in German: The dog bites the man. The difference is that the 2nd sentence makes a point of the man getting bitten as opposed to, say, some woman. Note the definite articles “der” and “den”: They signify subject and object, and without them (or their indefinite counterparts) the sentence would become ambiguous pidgin German.
        In Hungarian, you’d say “A kutya megharapja a ferfit.” or “A ferfit harapja meg a kutya.” Here “ferfi” means man, and a “t” is added to the end to signify “something is being done to him”.

      • J. Goard

        English has not, in my view, been simplified by its diffusion. If anything, it’s become more complicated, at least in its spelling.

        My bad. My comments should be taken as being exclusively about the spoken language, as is usually the case with linguists unless we specifically indicate otherwise. Most of the aforementioned language-simplifying adult learners were illiterate or nearly so, whether we’re talking about English, Spanish or Mandarin.

        As far as the specifics of quantifying morphosyntactic and semantic complexity, well, that’s a substantial topic for a truly interdisciplinary conference – I don’t have the math to address it the way I’d like. If you haven’t read John McWhorter’s recent books, that should give you some idea of where I’m coming from.

  • ‘Tis Himself, OM

    (Or if the user types in wo shi zhongguo ren

    Your hovercraft is full of eels?

  • timberwoof

    How could you hold a note if the word you are singing ends in “t”, for example?

    You stretch out the vowel for the duration of the note. If there is a word following, you tack the T onto the beginning of that word; otherwise you just make it at the end of the note. Pay attention the next time a good choir sings “good will to men”. You’ll hear they actually sing “goo dwill to men”. (That’s the example I learned after I was roped into a chorale.)

    Italian might be so well-suited for song that any old composer can do it, but a good lyricist can write beautiful songs in rougher languages such as German. J. S. Bach, W. A. Mozart and Schiller/Beethoven are existence proofs.

    I could be persuaded to learn Italian. Is there good fast Internet available in any of the the Cinqua Terra?

    • piero

      Yes, you are in fact right, and I’m wrong. I was just watching Marita Solberg’s performance of Solveig’s Song on YouTube, sung in Norwegian, and it was simply ecstasy. Also, some English choral music is magnificent. Nad Bach… well, Bach is Bach.

      Dere’s no need for me to persuade you to learn Italian. Vee have vays of making you speek, you know?

  • Pen

    The real advantage of hanzi is that once you learn it, it’s good for all the languages you want. I can read a few hundred characters and know what they mean, even though I don’t know how they are pronounced in whatever local language is using them.

    • http://kagerato.net kagerato

      True, the meaning of hanzi are largely conserved amongst languages that use them. Their shape isn’t necessarily exactly the same, though. This drove some Japanese scholars up a tree when the Unicode consortium decided not to specify separate code points for hanzi with common origin. (This was called Han unification.)

      In practice, the real issue is that Unicode doesn’t provide a mechanism in its encodings for identifying the language, and so instead it has to be provided out-of-band by something else — for example, the HTML document. You can solve the slightly varying shape issue easily by having different fonts that are specialized for the language, but you do need to know what language the text really is. To some extent, this is solved these days by algorithmic detection — for example, if you see hiragana you can normally make a safe guess that you’re looking at Japanese. However, auto-detection is not perfect and does not solve a lot of the edge cases with academic documents, cross-language comparison, and other bizarre multilingual cases you would never think of if you hadn’t seen them in real life.

      The problem becomes even more complicated once you start considering simplified versus traditional Chinese. Due to the fact that there are cases where multiple traditional hanzi were all mapped down onto a single character, and also cases were traditional characters ultimately became unused, the two do not have a one-to-one mapping. Han unification wasn’t able to deal with this issue, and so there are often multiple code points for two characters which ultimately mean the same thing, but are merely written differently. You can start to understand why the Japanese scholars were so frustrated; from their perspective, this looks like special treatment even though the cases are not exactly analogous.

      From the technical side, it all looks like a train wreck caused by trying to force the use of UTF-16, a 16-bit encoding. (It’s not strictly 16 bits per character, unfortunately, another striking flaw.) If the original development had standardized on a minimum of 24 or 32 bits, there would have been a massive space of directly addressable free code points such that giving every language its own “duplicates” is no issue at all. Sure, they could have done this with UTF-16 too through the use of the surrogate pair system to access the “high” code points outside the basic multilingual plane, but that is an very messy (inelegant) solution. It’s very clear that one or more people making the key decisions in Unicode’s development had their heart set on identifying all the world’s characters in just 16 bits, and that was a large, easily foreseen mistake.

      Well, most of that was very tangential to your point, but perhaps someone will find it interesting.

      • Monimonika

        Well, I found what you typed interesting. (^_^)

        Is this also the reason why mojibake happens?

  • http://ostatic.com/member/MollyGuy Discount Bats

    2.IT abilities: understanding of what techniques can and cannot do.

  • http://rocket-italian.spruz.com/disable EBONY

    Woah this blog is wonderful i love studying your posts. Stay up the good paintings! You realize, many persons are looking around for this information, you could aid them greatly.