June 14, 2007

Now it's time to play our game

So, yesterday, I asked which of the following four instructions was not like the others:

English: Work into lather with a little water.
German: Mit etwas Wasser aufschäumen.
French: Faire mousser avec un peu d'eau.
Spanish: Producir espuma con un poco de agua.

As one would expect from the sophisticated beings that are you LLog regulars, many people had interesting thoughts about what might make one of these sentences stand out from the crowd. I had one particular, very syntactic, dimension of difference in mind, which a lot of commenters also spotted, but there were (of course) several other possibilities mooted.

Some people pointed out ways in which one or the other language's lexical choices stood out. For instance, peter berry writes:

All except the German have "a little" translated directly.... If the translators had done the same for German it would be "ein bisschen", but instead we have "etwas".

And terrycollman notes:

... the English one uses "lather", not "foam" - a lather, surely, is thicker and heavier than a foam ...

Both perfectly true, and both touching on the thorny problem of lexical translation equivalents: A given word in language A in context a and b may not have a direct translation in language B at all, or may be idiomatic only in context a but not context b. Or language A might split a given semantic field into two subcategories where language B only bothers with a name for the supercategory.

Another kind of difference, more syntactic but also not what I had in mind, noted by jack lecou, among other, is that the word order of the German example stands out. The PP Mit etwas Wasser comes first, and the verb aufschäumen at the end; in all the other languages, the verb comes first, a phrase related to the foam next, and the PP last.

One pretty subtle semantic difference that had definitely not occurred to me, noted by miked (who remarks, "IANALinguist", but who could be), is that, in the English sentence,

...whether the lather already exists could be ambiguous (ie, mix product into an existing lather by using water (say as a solvent) versus using just this product and some water to create a lather.

Given the English sentence out of context, one can imagine that there might have been previous instructions to create an independent lather, e.g. with some soap, and then sprinkle the powder into it and mix it in, incidentally using a bit of water. This interpretation isn't possible for the French or the German, since the 'foam' bit is contained within verbs in both these two cases, and definite reference is famously not possible within lexical items; nor is it available for the Spanish since the instruction is not ambiguous between a producing-foam interpretation and a do-something-to-existing-foam interpretation, because it explicitly uses a verb meaning 'produce'.

People also noted two different kinds of morphological differences. An anonymous commenter thought the stand-out language was:

German, for instructing you to "foam" rather than to "make to foam" "produce foam" or "work into lather."

Andrew referred to the hardworking German verb aufschäumen as a "synthetic causative with an incorporated result", but don't mind him; he's been as overexposed to linguistic terminology as I have.1

The second-most-identified difference is also morphological, and, maddeningly, caught me a bit off guard. In every language except English, the verb form is not imperative, but infinitive! I'd noticed that for the French and Spanish, enough to change 'imperative' to 'instruction' in the LL version of my original post (but, sadly, not in the Heideas version), but didn't think too much about it. It should have been reflected in my glosses, though. If I'd been glossing the English, the verb would have come out something like, 'work.IMP', and, properly glossed, the French, German and Spanish verbs should have been 'make.INF', 'foam.INF' and 'produce.INF', respectively. (Indeed, the French should have been 'make.INF foam.INF', definitely not, 'make to.foam', which implies that the faire and the mousser are not the in the same form.) Since the bare infinitive and the imperative are indistinguishable in English, though, the distinction was not reflected in the glosses. My bad!

Anyway, that's a really obvious way in which the English stands out from the other languages. And it's really quite interesting! I had never really consciously recognized this use of the infinitive in French, though I've read enough French product directions and recipes. And it's certainly interesting that it's common to French, Spanish and German but not English! Commenter Sus speculates:

I think it's because we have the "du vs Sie" problem. If we were to use the imperative in manuals and instructions, we'd constantly have the trouble of having to decide the age group and status of our customers... But I'm just thinking aloud here.

Certainly the availability of grammaticized formal vs. informal forms of address is one way in which French, Spanish and German pattern together and differ from English, so this makes sense as a possible explanation, though I would have thought the natural thing to do in anonymous instructions would be to just use the formal form. A Spanish-speaking anonymous commenter notes:

Recipes and product instructions normally use the infinitive (producir espuma) or a passive voice (prod˙zcase espuma). The imperative (produzca espuma) seems very intrusive to me, and definitely feels like a literal translation from American English.

I'll have to check with a buddy who wrote her dissertation on imperatives to find out if this difference has been discussed in the literature.

The difference I was actually fixated on, though, was the one picked up on by the first commenter (and many subsequent ones), gregates, who wrote:

It seems to me it's the Spanish variant which is different, on the grounds that the word for "foam" is the direct object of the verb, whereas in the other three the object is left unspecified, with the context filling in the product itself as what one is to make into foam (up-foam?). So the other three tell you what to do with the product, whereas the Spanish version seems to leave the product out of it.

Well spotted! (If only the LL would fund a prize... but I'm afraid the whole discretionary budget is gone on the sherry in the senior writers' lounge. Some kind of Oxbridge thing. You could have one of my old conference name-tag holders, if you want.)

In fact, in each of the the first three sentences, there's an implicit direct object, however they might otherwise differ from language to language. It would be grammatical to insert a direct object noun phrase -- 'this powder', 'it', whatever -- into the sentence in the appropriate spot in all three cases. In the English case, it'd be the object of 'work'; in the German, it'd be the object of the causative verb 'aufschäumen'; in the French, it would be the structural object of the whole clause but the semantic subject of the intransitive 'mousser'. In the Spanish, however, no additional direct object is possible; 'espuma' is the direct object of 'producir'. The reader has to infer that they're supposed to use the powder to produce the foam; syntactically, there's no spot for it in the clause.

The issue of objects and their potential absence has been extensively discussed before here at LL in the past, e.g. here, here, here, and here. This 'recipe' scenario is one of the most well-known object-drop contexts, and clearly has its object-eliminating effect in at least three languages in which object drop is normally not too common. It would be interesting to know if this context can eliminate objects universally. Particularly, I'm curious about what happens in this context in polysynthetic languages, where object marking is generally obligatory. I bet you need some object marker even in this context. I'll see if I can find out and report back.

The reason this set of sentences caught my eye in the first place was that several brands of syntactic/lexical semantic analysis (including my own) would assign a broadly uniform syntactic structure to the English, German and French sentences, emphasizing their common causative and resultative natures. It would differ in its surface particulars from language to language, but not much in essentials. As I ran my eye over the different instructions with a particular kind of analysis in mind, comfortably going, mmm, yep, oh, sure!, running into a different structure for the same message brought me up a bit short. That Spanish sentence would get a different kind of analysis.


1 Commenter Sus provides some nice examples indicating that the auf in aufschäumen is one of them there nifty separable prefixes, being pulled apart from the schäumen in the finite forms Ich schäume auf, 'I foam up', du schäumst auf, 'You foam up'...

Posted by Heidi Harley at June 14, 2007 06:37 PM