In a recent blog post, Geoffrey K. Pullum writes:
Twitter merely coined a verb meaning “send a message via Twitter”, but they didn’t specify what linguists call its subcategorization possibilities. They added the verb to the dictionary, but they didn’t specify its grammar. The verb tweet is gradually developing its own syntax according to what it means and what its users regard as its combinatory possibilities. That is a really interesting, though unintended, large-scale natural experiment in how syntactic change works. And it is running right now, every minute of every day.
If Facebook has, similarly to Twitter, coined a new verb, it’s probably like. Sure, that word existed before, but the way it’s employed on Facebook, it has developed an altogether new meaning. When you “Like” something on Facebook – i.e. you click the “Like” button – you thereby do not “like” it in the old sense – rather, you already liked it before and you now announce this to your friends. In the old sense of the verb, you are the experiencer of an affection. In the new sense, you are an agent, a deliberate performer of an action. The subject of the verb has a different thematic role in each case.
In the German translation of Facebook, the verb gefallen was chosen to translate like. This captures the old sense of like very well, much better than the perhaps more direct translation mögen. On the other hand, gefallen is having difficulties assuming the new sense, that of clicking a button to display one’s approval or enjoyment of something.
I think this is because gefallen assigns the experiencer role to its (dative) object rather than to its subject. For illustration, consider the English verb strike which does the same thing in sentences like “It strikes me that you are losing weight”: the verb is used to describe a situation where somebody experiences something, e.g. the striking perception that another person is losing weight, and the experiencer is described by the object of the sentence, e.g. me. Like is different; here, the experiencer is described by the subject of the sentence. Thus, when translating English like to German gefallen, the syntactic arguments need to be swapped: “I like this” becomes not “Ich gefalle das”, but “Das gefällt mir.”
Why does this prevent gefallen from assuming the new sense of its English counterpart like? Couldn’t “I recently liked that park on Facebook” analoguously translate to “Dieser Park hat mir neulich auf Facebook gefallen”? This sounds very weird to me, and I think the reason is a restriction on the mapping between thematic roles and syntactic arguments. Remember that the syntactic argument that was assigned the experiencer role in the old sense is assigned the agent role in the new sense. And to my knowledge, agent roles are only assigned to subjects in German (and also in English).1
If this were to change by gefallen acquiring a new sense where the dative object fills an agent role, this would pose many syntactic problems. How would you translate “Like this on Facebook!” or “I decided to like the park on Facebook” to German? Imperatives are always addressed to the (invisible) subject, and control always identifies the (invisible) subject of the embedded clause (“to like this park on Facebook”) with the subject or object of the embedding clause (“I decided”). One could construct “Der Park hat sich entschieden, mir auf Facebook zu gefallen”, but that would mean the park decided, not I.
So how do German speakers deal with the challenges posed by grammatically strenuous gefallen being the official translation of Facebook’s ubiquitous like?
First of all, as Facebook’s UI itself is concerned, “Like” and “Unlike” are translated quite freely with “Gefällt mir” (“I like”) and “Gefällt mir nicht mehr” (“I don’t like anymore”). This avoids using a new sense of the verb and rephrases things to use the old sense, by allowing the user to describe herself as an experiencer rather than explicitly offer her the possibility to become an agent. The description does not match reality perfectly, of course, for when I unlike something, that does not imply I don’t like it anymore, it just means I no longer want to commit to that on my Facebook profile.
In status updates about liking pages or links, the issue does not arise, as the English original itself talks about liking in the old sense, as an affectional state rather than an action. After all, it says “Peter Schwarz likes Conny Fuchs’s link”, not “Peter Schwarz liked Conny Fuchs’s link.” Interestingly, the relatively free word order of German makes it possible to keep the structure of such messages using gefällt, putting the dative object before the verb: “Peter Schwarz gefällt Conny Fuchs’ Link.”
Before the verb is even the preferred position for the experiencer in most contexts. Nevertheless, the prototypical main clause word order in German still starts with the subject. And person names are not marked for dative case.2 So the above could also be read with Peter Schwarz as the subject and Conny Fuchs’s link as the object. When Facebook’s “Gefällt mir” button started becoming ubiquitous on the German-speaking Internet, I actually expected people to start using gefallen exactly like like in the new sense, with an agent subject and a theme accusative object.
It is not unusual for German verbs to assume semantic and syntactic argument structures from English counterparts, which is then bemoaned as an Anglicism once it has irreversibly settled in. An example is the verb erinnern (remember). The standard way to say “I remember this” in German is “Ich erinnere mich daran”, with an accusative reflexive pronoun and a prepositional object. However, recently “Ich erinnere das”, with the same structure as in English, is also frequently heard. Or maybe this non-standard usage has been around forever and I’m just interpreting it as a new Anglicism. (See johannes’s comment below.) To come back to my point, no, I haven’t seen or heard gefallen used with the argument structure of like yet. The new sense of like doesn’t seem to have a lexical counterpart in German yet.
Instead, gefallen is used with a slightly different sense in the context of Facebook – “Mir gefällt das” not meaning that I like it but that I made liking it part of my Facebook profile – but with the same argument structure (experiencer subject, theme dative object, no agent), still describing a state, not an action. As long as the context does not absolutely require an agent, gefallen in this sense is used rather consistently, as some of tonight’s first Google hits for „auf facebook gefallen“ show:
- Wenn dir beispielsweise etwas wie ein Buch, ein Film oder jemand wie ein Sportler gefällt, wird diese Verbindung genauso Teil deines Profils wie dies der Fall ist, wenn dir Seiten auf Facebook gefallen. (Facebook Help Center)
- High Live würde sich freuen, wenn noch mehr Leuten „High Live“ auf Facebook „gefallen“ würde… (High Live’s page)
- Die Seiten oder Produkte, die Mitgliedern auf Facebook gefallen, generieren automatisch entsprechende Vorschläge auf Amazon. (Social Media Pro)
- Wir haben für euch auf Facebook eine einige Seiten erstellt, die für Spieler gedacht sind, denen unsere Seiten zu Blizzard, Diablo, StarCraft und Warcraft auf Facebook gefallen. (BlizzCon 2010)
There are some bewitchingly creative variations:
- Gaggle ist auf diversen Plattformen präsent, über YouTube gibt es Gaggle-Videos, über Facebook werden neueste Nachrichten aus dem Gaggle-Kosmos ausgetauscht, 1253 Personen gefällt das. (Die Zeit)
- 313’000 Personen finden auf Facebook gefallen an Swarovski. (fuellhaas.com)
And the ultimate solution to the problem that the one who clicks “Like” is not a subject in such constructions has been found by YouTube user Clixoom. He uses a “causative” construction with lassen, enabling him to use gefallen in an imperative statement:
- lasst euch Clixoom auf FACEBOOK „gefallen“
This is also a creative new use of the phrase “sich etwas gefallen lassen”. The above invitation to like Clixoom on Facebook can also be read as a self-mocking invitation to put up with Clixoom on Facebook.
1 If you don’t count “logical subject” phrases in passive as objects of the verb, which you shouldn’t.
2 In Russian they are, and Facebook doesn’t seem to know how to do that yet even for Russian names written in Cyrillic, which is kind of lame and leads to lots of grammatically incorrect status updates, since the Russian like, нравиться, works pretty much exactly like gefallen.
Nice post. One little correction, though, even if you already anticipated it:
„das erinnere ich“ has been around for longer, and it is actually the standard construction in Low German. Now you could argue with some justification that Low German and High German are in fact different languages. But then I would assume that the simpler direct object construction has actually started its campaign from there, possibly winning against a standard construction weakened by English influence.
That’s a pity – I’ve had the exact same idea about „sich etwas gefallen lassen“ while reading your blog entry. Hmpf!
But I have got another supposition. Since I am no member of the facebook community, I follow your description of the original phenomenon:
When you “Like” something on Facebook – i.e. you click the “Like” button – you thereby do not “like” it in the old sense – rather, you already liked it before and you now announce this to your friends.
Nobody gains control over their experience of liking or disliking. It’s just that they can „make this happen“ in the sense that it becomes visible. So, in a way, this new sense of the word „to like“ is a causative. To get this translated properly, we should have a closer look at the „gefallen“ counterpart of german facebook. As you mentioned, the person clicking the button is a dative object here. So creating a causative from this, we could omit mentioning this object all the time – the subject of the new causative and the dative object would be identical anyway.
The solution would hence look like this:
Immer mehr Leute gefällen Texttheater bei Facebook.
Elegantly, this new verb would coincide with the noncausative version in the third person present. Ke gefällt das. would actually mean Ke lässt sich das gefallen.
@Kükenschublade: Bravo, I like this!
@David: Because it would ruin my point. :P Joking aside, to be more precise, I think the “logical subject” can’t be counted among the syntactic argument structure that a verb specifies lexically, because passive and the modified argument structure it requires follows rules that hold for all verbs.
This really is an argument. If it is not clear whether to count something as, for example, a complement or an adjunct, it is of course sensible to make the choice that best allows for generalisations like „only the subject can be assigned the role ‚agent'“. But the footnote seemed to suggest that you had independent arguments.
The argument that „passive and the modified argument structure it requires follows rules that hold for all verbs“ seems to presuppose that the lexicon (whatever that may be) is not a decent place in which to put the necessary generalisation. Because, if it is, it is still possible to postulate, for example, that every base form of a transitive verb (keeping things simple here) yields a passive participle selecting for a subject corresponding to the original object and optionally selecting for a suitable prepositional phrase, corresponding to the original subject. That is how things are or were handled in GB and HPSG literature, where the participles were derived by lexical rules. In this setup, the prepositional phrase would be an object of the verb, an element of the argument structure, optionally realised. So the argument structure of the passive participles itself would be where the relevant generalisation is expressed by means of the lexical rule.
Another approach, the spirit of which appeals to me much more, is to locate the needed generalisations about passive in the passive auxiliary (which would only be an obligatory dummy in the lexical rule approach). In this case the auxiliary would select for a participle and identify its accusative object with its own subject and it could also be specified to optionally select for a prepositional phrase that semantically corresponds to the original subject. But if one tends to favour this approach, a true problem with the object status of the PP arises, as witnessed by the following sentence (which I do not consider prettiest German, but grammatical enough to count as evidence).
(1) Von Karin gebacken wurden der Käsekuchen und die Himbeertorte
With [von Karin gebacken] occupying the Vorfeld, the expression should be considered a constituent. But selection for [von Karin] would have to originate from wurden. How come, then, that gebacken can form a constituent with a complement of a verb of which it is itself a complement? No matter how one would want to combine [von Karin] with [gebacken] in the first place, there would still be need for a very unwelcome case of nonlocal subcategorisation. There seems to be no easy way around the problem that would allow for constituenthood of [von Karin gebacken] and selection for the PP by wurden at the same time. Hence this setup seems to require the PP to be an adjunct. This is the only decisive argument concerning the status of the PP that I know of, and it depends on an earlier choice for a particular kind of analysis for passive. But it would of course make the suggested generalisation about the subject-agent relation possible.
Sorry if that is too far off topic.
Auch sehr schön: „Leute, deren Internet fast nur noch in Facebook stattfindet, können Gsallbahdr Zwei dort gern haben.“