Rosetta Stone: Bad at Languages

A German friend recently tweeted this photo of an ad for Rosetta Stone software on the London Tube:



That’s the song “Let it snow, let it snow, let it snow” with “snow” translated into German, Dutch and Swedish, respectively.

What she means is that they’ve taken the noun form of snow in all three languages, rather than the verb. Which is embarrassing enough for a company trying to sell you language learning software. But it gets better: the press release for the campaign has now been corrected for one word (“sneeuw” > “sneeuwen”), but not the other two:


Stop and think about that for a second: someone at Rosetta Stone (or more likely their “Brand Action agency” MBA) found out that the Dutch word was wrong and went to the trouble of correcting it, but didn’t even check the others.

And here’s the really funny part: Rosetta Stone is far more of a marketing company than an education company, so if anything, you’d expect them to get it right in the ads and wrong in the software. In 2011 they spent more than six times on marketing what they spent on research and development (numbers in 000s):


That’s right: for every time you spend $200 on Rosetta Stone software at the mall, they’re spending about $25 on developing/maintaining it, and over $160 trying to sell it to you. Oh, and another $35 on all the slick packaging. Which, as you may have noticed, is already more than $200. Which is possibly why their stock is down more than 30% since the company went public just a few years ago.

So if you feel like a sucker for buying language software from a company that can only get the word “snow” right one out of three times on the second try, just look at it this way: at least you’re not a shareholder.


Appendix for nerds: why are the verb and noun for snow the same in English, when it’s obviously such a close cognate with the words above? Well, they used to be different in English too. For example, as Thomas Chestre wrote in Sir Launfal:


(“She was as white as a lily in May / Or snow that snoweth on a Winter’s day”)

I don’t know when the verb “snewen” (?) became “snow,” but I think it’s happened with a lot of English noun/verb combinations, since we’ve lost more of the old verb forms than the other Germanic languages above.

[EDIT: I originally misattributed that quote to Chaucer, see correction from Michael W in the comments]

46 thoughts on “Rosetta Stone: Bad at Languages”

    1. They already have. There are many great courses and other material. You just need to find what suits you, stick to it and combine a few things together. For a lot of good advice, go to or there is a lot of learning blogs etc. Just seek advice from people who USE the courses, NOT SELL them.

    2. They already have. Babbel, Busuu and LiveMocha all offer web-based language learning programs that are cheaper and constantly updated. They’re all doing great work.

    3. I have a language learning system that rivals Rosetta Stone and most others. It’s all based on 35 years of research, development and teaching experience. I’m getting my website made to get the word out there.

  1. That last letter in “snewen” is the thorn, pronounced th. “Or snow that sneweth yn wynterys day”. So it’s closer to modern English than it looks.

    This bit may be apochryphal, but I understand that this is where we get “ye olde pub”, etc. Lacking a thorn in their French-made type settings, printers would substitute ‘y’. So “ye olde pub” should still be pronounced “the old pub”.

    1. Yep, I was referring more to the vowel shift in the stem though, snEweth > snOweth/snOws

      That’s really interesting about “ye,” never thought about that before. I hope it’s true.

      1. Yes, that’s true. See Wikipedia’s article on “Thorn (letter)”, specifically sections “Middle and Early Modern English” ( —

        The modern digraph th began to grow in popularity during the 14th century; at the same time, the shape of thorn grew less distinctive, with the letter losing its ascender (becoming similar in appearance to the old wynn (Ƿ, ƿ), which had fallen out of use by 1300) and, in some hands, such as that of the scribe of the unique mid-15th century manuscript of The Boke of Margery Kempe, ultimately becoming indistinguishable from the letter Y. By this stage th was predominant, however, and the usage of thorn was largely restricted to certain common words and abbreviations. In William Caxton’s pioneering printed English, it is rare except in an abbreviated the, written with a thorn and a superscript E. This was the longest-lived usage, though the substitution of Y for thorn soon became ubiquitous, leading to the common ‘ye’ as in ‘Ye Olde Curiositie Shoppe’. One major reason for this is that Y existed in the printer’s type fonts that were imported from Germany or Italy, and thorn did not. The first printing of the King James Version of the Bible in 1611 used the Y form of thorn with a superscript E in places such as Job 1:9, John 15:1, and Romans 15:29. It also used a similar form with a superscript T, which was an abbreviated that, in places such as 2 Corinthians 13:7. All were replaced in later printings by the or that, respectively.

        — and the immediately following, “Modern English” —

        Thorn in the form of a “Y” survives to this day in pseudo-archaic usages, particularly the stock prefix “Ye olde”. The definite article spelled with “Y” for thorn is often jocularly or mistakenly pronounced /jiː/ or mistaken for the archaic nominative case of the second person plural familiar, “ye”.
        A handwritten form of thorn that was similar to the letter “y” in appearance with a small “e” written above it as an abbreviation for “the” was common in early Modern English. This can still be seen in reprints of the 1611 edition of the King James Version of the Bible in places such as Romans 15:29, or in the Mayflower Compact. The word was never pronounced with a “y” sound, even when so written.

      2. This vowel shift isn’t present in modern Dutch or modern Swedish either. The only reason the verbs are different from the nouns in those cases is because those languages add a suffix to mark the infinitive of the verb (“-en” in Dutch, “-a” in Swedish), whereas an English infinitive does not have any special suffix.

      3. I know I have read snarky comments (not here, of course) on several occasions that the “ye” in fakey Olde English pub signs should actually be pronounced “the.”

      4. It’s not that there’s been a “vowel shift” in the verb “to snow”, per se. What it is is that vowel quality used to change to indicate grammatical function (ablaut and umlaut). That grammatical feature has slowly been erradicated from English, surviving now in only a handful of verbs, and only to distinguish present from past simple: eg A gift, I give and I have given all have an “ih” vowel sound, but I “gave” has a completely different one.

    2. Close, but no cigar. The similarity is not with ‘thorn’ (the letter representing the sound at the end of ‘with’), but with ‘eth’ (representing the sound beginning the word ‘the’). It looked like a “lazy d”with a line through the top. That line across the top creates something resembling the top of the letter ‘y’, and if you shrink the loop at the bottom of the “lazy d”, you something that looks a lot like a modern ‘y’.

  2. The OED says:

    Forms: OE sniwan (sniu-, sniuw-), ME snywe(n, ME sniuw-, ME sneuw-, ME–15 snewe.
    Etymology: Old English sníwan , = West Frisian snije (sneie , snīe), North Frisian snī, sneie, snaie, Middle Dutch sniwen, snien, Middle Low German snîghen, snygen, Old High German snîwan (Middle High German snîwen, snîgen, snîen, German schneien, dialect schneuen, schneiben, etc.) < *snigwan– , related by ablaut to snow n.1

    1. intr. To snow.
    c725 Corpus Gloss. (Hessels) N 117 Ninguit, sniuwið.
    c900 tr. Bede Eccl. Hist. ii. xiii, [If] hit rine & sniwe & styrme ute.
    a1000 Epist. Alex. in Cockayne Narrat. Angl. (1861) 23 Ða cwom þær micel snaw and swa miclum sniwde swelce micel flys feoll.
    a1250 Owl & Nightingale 620 His hou [= hue] neuer ne uorlost, Wan hit sniuw [v.r. snywe] ne wan hit frost.
    a1325 Orfeo 245 Þei it comenci to snewe and frese.
    a1400 K. Alis. (W.) 6450 Whan hit snywith [Laud MS snoweþ], other rayneth.
    a1400 Launfal 293 Sche was as whyt as..snow that sneweth yn wynterys day.
    1530 J. Palsgrave Lesclarcissement 130 Il neige, it sneweth.
    1746 Exmoor Scolding (ed. 3) i. 8 Whan [it] snewth, or blunketh, or doveth, or in scatty Weather.
    c1405 (1387–95) Chaucer Canterbury Tales Prol. (Hengwrt) (2003) l. 347 It snewed in his hous of mete and drynke.

    2. trans. To sprinkle like snow.
    c1440 Pallad. on Husb. xi. 332 On kadis thre Of wyn a certeyn of this flouris snewe.

    1. Another interesting thing from the OED is that when the word “snow” came to be used as a verb, for a time people conjugated it with a past tense of “snew” and a past participle “snown”, based on “blow”/”blew”/”blown”.

  3. I don’t think your interpretation of Rosetta’s financial figures is entirely fair — most companies spend more on sales and less on R&D than you expect. Apple, for example, spends about $2.50 on R&D for every $100 you spend on the latest iDevice (for the three months ended September 29, 2012, see

    On the other hand, I can see that languages are apparently not their strong suit (:-).

    1. You’re right. I was more focused on the marketing costs than the R&D. Apple doesn’t break out SG&A in that income statement, but the total amount is only 6.5% of sales (10/156) and for RST it’s 83% (161+62)/268.

      Still not totally fair to RST in some ways, since software development is more front-end / R&D heavy than hardware, and I didn’t look at all their old numbers. But the main point is that products like this are far more sold than bought, more akin to timeshares than iPhones in that sense. So you’d expect them to be a little more careful about marketing.

  4. I never understood why an A1 language learner needed to know how to say “The boy is under the plane.” As for the Y/th thing, it’s definitely true. The gravestones at the colonial graveyard near my hometown are pretty consistent about using those Ys (as well as long S).

  5. I think the marketing orientation of the company is a good explanation of why they are so sloppy on accuracy; the whole mindset of marketing people (whom I have to deal with constantly at work) is that factual accuracy is close to irrelevant compared to making the desired impression on the target demographic.

    1. Ohh yeah. I was senior linguist for Dragon Systems (Dragon NaturallySpeaking…) all through the nineties. The sales staff always used to play down the need to train the program and the user (i.e., learn how the program works).

  6. The American Heritage PIE dictionary says:

    *sneigʷh- Snow; to snow. 1. Suffixed o-grade form *snoigʷh-o- ‘snow’, from OE snāw, from Gmc *snaiwaz ‘snow’. 2. Zero-grade form *snigʷh- in neve, nival, niveous, from Latin nix (stem niv-) ‘snow’. [Pokorny *sneigʷh- 974.]

  7. The problem with this entire line of ridicule is that the ad is a joke. It’s not meant to be a model of the correct mixing of Germanic languages. It’s a harmless macaronic that works on simple word substitution. To fit with the song’s meter, it needs monosyllables. To allude to language learning materials, it needs graphics that look like vocabulary cards. To be comprehensible to its audience, it needs words that evoke the English word ‘snow’ without demanding too much effort from the reader.

    I don’t get the idea that mixing English with German, Swedish and Dutch is an acceptable conceit, but using nouns for verbs is an incongruity too far. ‘Let it Schnee’ is wrong, all wrong – but ‘Let it schneien’, that would be fine? It’s bilingual word-play, from start to finish. They shouldn’t even have ‘fixed’ the Dutch.

    1. Well, I see what you mean, though I don’t quite agree. Yes, to me the partial correction was worse than the original ad. And if they were in any business other than language instruction, maybe it wouldn’t have jumped out as much. But even the original is worse than you’re making it sound.

      The thing is, mixing English with those particular three languages is more than an “acceptable conceit,” it’s done all the time, and *especially* in a cute/joking way. Remember VW’s “Fahrvergnügen” slogan in the US? I just read a pulpy old American sci-fi novel that was full of henchmen-type supporting characters with names like Schmutzig and Dreck. And here in Germany there are lots of shop names and advertising slogans that rely on English-German puns and mashups.

      So yes, it’s wordplay, but that doesn’t mean anything goes — there’s an art to bilingual wordplay like any other wordplay, and it’s totally possible to get it wrong. So, to directly answer your question: my German is far from perfect, but yes, “let it schneien” sounds like a coherent joke to me, while “let it Schnee” just sounds like a mistake. And a lot of native speakers seem to agree.

      1. a lot of native speakers seem to agree

        I speak German, but not Swedish or Dutch, and “Let it Schnee” does trigger a double-take that “Let it schneien” doesn’t. But where I part company with apparently everyone but Harry Campbell, including the perpetrators themselves, is that I don’t see any basis for calling it objectively wrong. They’ve pulled a stunt that is partly linguistic, but also visual, under various constraints of audience accessibility, and I don’t know what rules exist that it can be said to offend against. I’ll take your word for it that German-English code mixing happens a lot, but surely it doesn’t happen so frequently that there are unchallengeable conventions. (The speech of bilinguals isn’t relevant: the ad isn’t an attempt to emulate particular kinds of speaker.) The problem seems to be that the ad’s creators are playing a different game than nearly everyone on this blog assumes they’re playing. The unarticulated rule the critics are assuming is that, when you code-mix, you must use the same category of L2 word as the L1 word you’re substituting, much as bilinguals do. If you don’t do that, the argument seems to be, we’re entitled to assume you don’t know the L2.

        But what I take the ad people to be doing is more like a bilingual variant of the rebus. In a rebus, you associate the pictograph with a word and put into the text the pronunciation of that word, leaving the meaning behind. In this ad, they take the sound value of an orthographic L2 word and leave the grammatical category behind. That’s not wrong, it’s just not the game you thought they were playing.

        And I think it matters if you treat this as solely a linguistic text, neglecting the visual aspect. It’s not just a linguistic mash-up, it’s a genre mash-up. The ‘snow’ words are meant to resemble what I suppose must be a typical ‘word-card’ in the software, with both a word and a depiction. So after each ‘Let it’ you’re not just switching language, you’re switching genre, or voice, with an attention-grabbing discontinuity. (I imagine a TV version of the ad might intercut the English song with German/Swedish/Dutch voices, perhaps spoken; maybe the voices of the software itself, for a similar effect.) The point is, if you reduce the whole ad to just the words “Let it Schnee Let it sneeuw” etc., you’re missing important cues for how the ad is meant to be taken – i.e., partly as an allusion to the software – and it reads more strangely than it would if you were reading the words in the context of the ad layout.

        In a nutshell, I think they were attempting to be more playful than they’re being given credit for, and it obviously backfired for some portion of their audience.

      2. The unarticulated rule the critics are assuming is that, when you code-mix, you must use the same category of L2 word as the L1 word you’re substituting, much as bilinguals do. If you don’t do that, the argument seems to be, we’re entitled to assume you don’t know the L2.

        But what I take the ad people to be doing is more like a bilingual variant of the rebus. In a rebus, you associate the pictograph with a word and put into the text the pronunciation of that word, leaving the meaning behind. In this ad, they take the sound value of an orthographic L2 word and leave the grammatical category behind. That’s not wrong, it’s just not the game you thought they were playing.

        That’s very well put. I guess it’s subjective which framing dominates for any given viewer, but I think you’ve given the best possible defense of the original ad, and one that I hadn’t fully thought through.

      3. very well put

        Not that well! This…

        In this ad, they take the sound value of an orthographic L2 word and leave the grammatical category behind.

        …should be this:

        In this ad, they take the sound value of the English translation of an orthographic L2 word and leave the grammatical category behind.

  8. If you don’t mind a newcomer to your blog, please permit the following post—something of an “official” response from us here at Rosetta Stone:

    In a word, we’re ashamed. We tried to capture the spirit and meter of a popular Christmas tune and, regrettably, our enthusiasm for spreading marketing cheer outpaced our respect for linguistic accuracy. We green-lighted an ad before its time. The fact is, we have a stringent pedagogical approval process at Rosetta Stone, and we missed an important check-point here. There’s no excuse. The ads have been recalled. We assure you that from here on out, no one at Rosetta Stone–including marketing–will be taking shortcuts. We’re sure that this post will invite more thoughtful (even heated) criticism, and we hope you’ll understand if we don’t engage further in the dialogue for the moment—we have important work to do on the home-front. Thank you for keeping us in check and have a great holiday. (Hey, maybe we’ll try ‘Silent nuit, holy Nacht’….)

  9. Reblogged this on Sundry Times and commented:
    From a faux pas to a clarification and an apology by way of a fascinating etymological wander. I think I forgive them because sometimes the act of doing something is bound up in the being of it. To put it another way: let it be snow.

  10. It’s usually fair to attribute all Middle English quotes to Chaucer, but that line is Thomas Chestre’s. It appears in the citation Language Hat posted (Chestre wrote Launfal). I can’t find anything like that in the Canterbury Tales, and presumably neither could the OED.

    But since that seems to be an image of a scanned book, I’m curious if the misattribution came from there or elsewhere.

    1. Thanks for the catch! I’ve corrected it above.

      The image is from Google Books and when I find it again I’ll post the source, but I’m sure the error was mine.

  11. The ad shows pretty well what’s wrong with RS. You can’t learn to say “let it snow” by learning the words let, it and snow.

    I don’t think it’s a multilingual rebus. That’s simply a silly contribution to all the language myths out there. English native speakers already tend to translate this sort of thing literally, no need to encourage this even more. That’s not the way to show that language learning can be easy.

  12. Listen, silly rabbits: What Rosetta Stone really meant to say was “Let it Kokain! Let it cocaïne! Let it kokain!”

    But while we’re on the subject of rebus,

    “That’s not wrong, it’s just not the game you thought they were playing.”

    I don’t know. I think you’re being a little generous with that interpretation. If a rebus were intended, a better one would be


    which would have the simple word substitutions, language learning allusions and audience accessibility you mentioned while succeeding at artful multilingual wordplay (and keeping with the Christmas song theme).

Leave a Reply

Your email address will not be published. Required fields are marked *