9

If someone could direct me to papers/sites that describe this, and a summary or something, that would be great.

It is just a parameter for languages? What do linguists think so far?

Example: "Which store did you buy the book from?" and "From which store did you buy the book?" are both allowed in English.

I know from Full Interpretation (Chomsky 1993, 1995) that the motivation for Wh movement is a +Wh feature and that locality constraints must be satisfied... perhaps this is related?

hippietrail
  • 14,687
  • 7
  • 61
  • 146
dmonopoly
  • 333
  • 1
  • 10
  • I'll maybe write a more detailed answer later, but here's this in the mean time. P-stranding movement is actually extremely rare cross-linguistically - the norm is to disallow P-stranding. The ban on P-stranding has famously been explained as a sub-case of anti-locality - Klaus Abels' 2003 PhD thesis is the primary reference: http://ling.auf.net/lingbuzz/000049. Anti-locality disallows movement from the complement of a head X to the specifier of that same head; It's a ban on movements that are too short. – P Elliott Dec 01 '13 at 03:20
  • Abels argues that P is a phase-head - which means that movement out of a PP must proceed through the specifier of P. That means that, generally speaking, P-stranding movement will be disallowed, as it has to involve movement from the complement of P to specP. Where P-stranding movement is allowed, as in English, Abels argues that there must be some additional functional structure between the preposition and the moved category, which renders the movement to SpecP long enough such that it no longer violates anti-locality. Alternatively, we could say that P isn't a phase-head in English. – P Elliott Dec 01 '13 at 03:24
  • The Wikipedia article has some info on preposition stranding, with sources mentioned. If you haven't already seen it, here it is: https://en.wikipedia.org/wiki/Preposition_stranding . – Tim Osborne Dec 01 '13 at 04:50
  • 3
    @P. Elliot. I suggest toning down the linguistic jargon that is specific to one particular theory of syntax. My guess is that many readers have no idea what you are talking about. – Tim Osborne Dec 01 '13 at 04:54
  • 3
    It's possible etymology may be to blame for this rare ability of English. It's thought that in PIE or pre-PIE prepositions were originally adverbs, and their objects were in fact the objects of the verb phrase. The same situation is seen (entirely independently) in Totonac, where "prepositions" are in fact bound morphemes of the verb and their objects are free-floating arguments of the verb which participate in the agreement hierarchy. – Justin Olbrantz Dec 01 '13 at 07:56
  • 3
    @TimOsborne I doubt people get asked to tone down math jargon on the math SE. Anyway, since the questioner referred to Chomsky '93/'95, i assumed he would know what i was talking about. – P Elliott Dec 01 '13 at 17:12
  • 1
    @TimOsborne it's also worth mentioning that the wiki article you linked to doesn't actually say anything about why languages differ in this regard - it purely explains what P-stranding is, and gives some data. Since the questioner wanted an account of the difference, some kind of theory-internal explanation was necessary. I don't really see any alternative. – P Elliott Dec 01 '13 at 17:16
  • 2
    @JustinOlbrantz Are you suggesting that this is the situation in English? It seems to me that this is pretty obviously false, since nothing can intervene between a preposition and its complement in English. They clearly form a unit. – P Elliott Dec 01 '13 at 17:17
  • @PElliott "To my knowledge", determiners can intervene between a preposition and its complement :) – cyco130 Dec 01 '13 at 19:14
  • 2
    @cyco130 the determiner is part of the constituent (call it an NP/DP, whatever you want) that the preposition takes as its complement, i.e. John looked [PP at [DP the painting]]. The entire constituent the painting is the complement of the preposition. It's pretty uncontroversial that a preposition and its complement form a unit in English (the exception being complex seperable verbs, such as send out in John sent the letter out), i don't see much point in arguing about this. – P Elliott Dec 01 '13 at 19:32
  • That's one parse. Another is the parse that treats look at as a verb with a transitivizing clitic and the painting as its DO; this is one that's accessible when stranding instead of pied-piping the preposition. I.e, there's no PP constituent that matters, just like there's no [pp of glasses pp] that matters in a pair of glasses. – jlawler Dec 01 '13 at 19:45
  • @jlawler I don't buy that. Note that you can say Who did you talk yesterday with? and it's perfectly grammatical. Given that yesterday intervenes between the verb and the preposition, with can't cliticise to the verb. – P Elliott Dec 01 '13 at 22:26
  • @PElliott: I do not disagree with your analysis that a modern, true preposition cannot normally come before but separated from its object; but why should that rule out the hypothesis that the origin of prepositions is adverbial, and that that's the reason why the same word can act as a true preposition but also as a particle/adverb, and also stranded at the end? I believe this hypothesis has fairly wide support among historical linguists? It is also visible in languages like Greek and the other Germanic languages, and it could also explain "prepositions" prefixed to verbs in Latin etc. – Cerberus Dec 02 '13 at 01:16
  • @JustinOlbrantz So that is a nice, unified theory that could explain why "prepositions" occur in non-prepositional positions in many (all?) Indo-European languages. Perhaps prepositions do not have an adverbial (pre-)history in Chinese? – Cerberus Dec 02 '13 at 01:18
  • @Cerberus, "Perhaps prepositions do not have an adverbial (pre-)history in Chinese?". I think that's safe to way. Prepositions are historically (and are still?) verbs in Chinese. – dainichi Dec 02 '13 at 01:42
  • @dainichi: Ah! Then that might explain it. – Cerberus Dec 02 '13 at 01:54
  • @PElliott: Don't worry, you don't have to buy that; it's free, and available whether you buy it or not. As for intervenementations, I don't know of any prohibitions; and I don't find Who did you talk yesterday with? "perfectly grammatical", either. So no problemo; we speak different idiolects, which is not surprising, since everybody else does, too. – jlawler Dec 02 '13 at 02:51
  • 1
    It's much improved if yesterday receives focal stress. Also, @JustinOlbrantz i can see the appeal, but unfortunately most Indo-European languages don't allow preposition stranding. English is an unusual case. The norm, e.g. in German, is for it to be disallowed. – P Elliott Dec 02 '13 at 03:44
  • The re-analysis account of P-stranding which some of you are alluding to here was proposed by Weinberg & Hornstein in '81 (here: http://www.jstor.org/discover/10.2307/4178205?uid=3738032&uid=2&uid=4&sid=21103032723557, behind a paywall unfortunately). See page 247 of Abels' thesis, which i linked to earlier, for what i take to be some definitive arguments against it. – P Elliott Dec 02 '13 at 04:02
  • @jlawler Here's another, more acceptable-sounding counter-example which might be more convincing to you: Who did Sam talk to Harry about?. Note the lack of adjacency between the stranded preposition and the verb. A cliticization analysis predicts this to be bad. – P Elliott Dec 02 '13 at 04:04
  • Your imaginary cliticization analysis predicts that; not me. Adjacency doesn't always matter, like I said. Talk about certainly allows intervenementation, at least with a dative PP; though note that the P is still necessary; it looks like an incipient 3-place phrasal verb rather than a clitic. English prefers shorter clitics than about; if you're thinking of German as some standard here, about might be trennbare Nachsilben. – jlawler Dec 02 '13 at 14:40
  • @jlawler I was doing my best to try to make the cliticization account falsifiable. Doesn't seem like such a leap to assume that adjacency is necessary for cliticization - that's pretty much definitional of a clitic. I highly doubt that ueber is a bare suffix, but i don't see how that's relevant. German is indeed representative of most Indo-European langs in the sense that it doesn't allow P-stranding. – P Elliott Dec 02 '13 at 15:07
  • But it does have trennbare Vorsilben, just as English has phrasal verbs, so adjacency and cliticization are independent -- though not orthogonal -- variables. By the way, don't bother to make up "accounts" on my behalf, please; I've gotten in the habit of doing my own accounting. – jlawler Dec 02 '13 at 15:30
  • Oh, i was never under the impression that you came up with the cliticization analysis in this comments thread! It's been proposed elsewhere. I was entertaining what i take to be a strong version of that hypothesis as a hypothetical, i never attributed it directly to you. – P Elliott Dec 02 '13 at 15:56
  • @dmonopoly Are you somehow unsatisfied with the current answer to your question? If so, it could be nice to explain why. If you are satisfied, it is also nice to accept the answer: the SE websites work much better in this way. – Olivier Dec 12 '13 at 22:14
  • Just needed to do some more research work! (I also am aware of the strangeness of how many comments there are... on other SE sites I've never seen such a long comment thread. Usually it gets blocked off or you're suggested to move to a chatroom. Hopefully users of ling SE will give answers to questions more instead of just posting comments in the future...) Anyway, thanks for the answer! – dmonopoly Dec 14 '13 at 07:00
  • I realise this discussion is quite old, but since Chinese diachronic syntax was mentioned, I should mention this: Some coverbs ('prepositions') COULD be stranded in Archaic Chinese RCs, including 以 and 與. – WavesWashSands Nov 21 '16 at 08:13
  • Actually I should add: Archaic Chinese also had wh-movement, and also stranded coverbs: 何以戰 (canonically 以X戰), 王誰與為不善？. The loss of wh-movement probably resulted in the impossibility of stranding here. As for RCs, 所 began to combine with 之N/者 in LAC, and this is now compulsory: nowadays object RCs must occur with 的(N) (which is etymologically derived from 之 and 者), whereas 所 has become optional. It's no longer the old RC construction; 所 is only left as a vestige of the old construction, used in formal speech. So in the Chinese case, the constructions involving stranding were lost. – WavesWashSands Nov 21 '16 at 08:44

3 Answers3

3

It is genetic. Prepositions in Indo-European languages come from adverb-like particles which themselves often come from some sort of noun in a specific case. This adverb could be mostly anywhere in the sentence given the relatively free word order of PIE.

In English, many of these adverbs never fully transitioned into pure prepositions as they can stand on their own:

The treasure is 2 meters below/above the surface. (preposition)

The treasure is somewhere below/above. (adverb)

I came through hell. (preposition)

I came through. (adverb)

Fire in the hole (preposition)

The spy is in. (adverb)

Furthemore, these adverb like particles not only became prepositions in the languages (i.e. noun modifiers), but they also became verbal prefixes. Again, in English the move to the verbal prefixes did not happen all too much (another example is German where there are still verbs with separable prefixes).

The possibilities of development can be observed in ancient Greek:

Oreos bainó kata. - mountain-GEN go-1SG down

Kata oreos bainó - Down mountain-GEN go-1SG

Oreos katabainó. - mountain-GEN down-go-1SG

In some languages, it can even combine the two, adverb became a verbal prefix, but the verb still requires a noun with the same preposition - e.g. in Czech you may have both:

Přejdu přes řeku - Over-go-1SG over river-ACC

Přejdu řeku - Over-go-1SG river-ACC

So basically, I believe it is safe to say that it is mostly due to chance that in Germanic languages the grammaticalisation went more slowly (to the of not being finished in English) than in other IE languages.

Eleshar
  • 2,363
  • 15
  • 21
3

I write this as an answer because the comment thread is too long already for (what I take to be) the spirit of SE sites. The language used by the author of the question (parameter, wh-movement and the two references to Chomsky) strongly suggests that he wants (or at least is comfortable with) an answer within minimalist linguistic theory. Within this framework, the answer given by P Elliott in comments seems to me to be quite perfect. The main points, as I see them, are the following.

  • As P Elliott said, prepositions stranding is a rare phenomenon cross-linguistically, with just a handful of examples outside the Germanic family.

  • Answering it is just a parameter would be theoretically very unsatisfactory. At the very least, one would want to this parameter to impact different parts of syntax, so that one could make experimental predictions of the form If a language tolerates preposition stranding, then.... Ideally, one would also want these correlations to be built on theoretical grounds which are independently motivated.

  • K.Abels' thesis 1 does just that. As P Elliott said, the core of the thesis is the isolation of a condition on movement (the anti-locality condition) which is certainly independently motivated in the derivation by phases conception of minimalist syntax. This condition entails that preposition stranding is possible only when PP is not a phase, something that can be investigated experimentally independently. For instance, though it might not be explicitly stated in his work, it follows from E.Reuland's analysis of binding theory that there should be a correlation between languages allowing preposition stranding and languages allowing binding of a reflexive in a PP, and this is indeed not a trivially false prediction, as is shown by the comparison below. In (297) of 1, another (highly non-trivial) prediction is also offered and argued to follow from the antilocality condition: a language allows clitic pronouns as complement of a preposition if and only if it has clitic pronouns and allows preposition stranding.

  • Another well-known experimental prediction relating preposition stranding to other syntactic phenomena is the so-called Merchant's generalization: a language allows preposition stranding if and only if it allows preposition stranding in elliptical wh-construction (so called sluicing constructions) 2

Jacob often thinks about him (him≠Jacob).

Nathan pense souvent à lui (lui=Nathan or lui≠Nathan).

Sources:

Successive Cyclicity, Anti-Locality, and Adposition Stranding

The Syntax of Silence: Sluicing, Islands, and the Theory of Ellipsis

Olivier
  • 1,896
  • 13
  • 14
  • 1
    @PElliott Well, it is only natural I should acknowledge you, as you did all the work. But I come from a SE site which really frowns upon extended comment discussions to questions with no answer. – Olivier Dec 03 '13 at 09:32
0

I figured I'd give an answer with respect to the languages mentioned

With regards to questions, Chinese is wh in situ, and consequently has no need for coverb stranding - contrast this with the translation (sorry if you hate this construction):

(1) 你在哪裡度假?
    Ni    zai nali  du-jia?
    You   at  where spend-holidays?
    Where did you spend you holidays at?

As for RCs, Chinese already has another [+case] strategy, the resumptive pronoun, so by paradigmatic economy, coverb stranding is not needed:

(2) 我用它搜尋網頁的工具
    Wo  yong ta  souxun wangye  de  gongju
    1sg use  3sg search webpage LIG tool
    the tool which I search for webpages with

By contrast, English doesn't generally allow resumptive pronouns in simple RCs like these:

(3a) the tool which I search for webpages with (*it)
WavesWashSands
  • 3,451
  • 16
  • 31