rarity of preposition stranding

Daniel Everett dan at daneverett.org
Mon Oct 4 09:40:56 UTC 2010


I was responding to both questions. It isn't that there couldn't be an interesting answer to these questions. But there are two reasons why I am not particularly keen to invest my own time in answering them. First, questions and answers about rarity can lead to circularity. Second, I'd rather look for phenomena relating language and culture, rather than simply structural questions.

Rarity could be a coincidence. Out of all the languages that do, did, or will exist, is this construction really rare? Country music today  is a very simple musical form (not disparaging the masters like Hank Williams, George Jones, and so on). 4/4 beat, with pretty constant melodies, and largely predictable lyrics. It is an outgrowth of a number of different forms fused by American cultural values. But it is so simple, why is it so rare, indeed unique among the world's musical forms? The answer it seems to me is the combination of culture and form. But maybe it will be borrowed and begin to be adapted in and to other musical traditions/grammars - because they have adapted it to their values - and then it will be less rare. I am not saying that all grammar has a cultural explanation, though such explanations might be found more commonly than we realize.  Why do we say 'red, white, and blue' in America and not 'blue, red, and white', for example? Cultural reasons perhaps. I think that Tom Givon's larger account of adpositional, ad-affixal phenomena, plus just preference could account for the rarity of this structure. In fact, if we take Tom's suggestions, then English preposition stranding is a language-specific adaptation of a common process. It is a single off-shoot from a big tree. Language forms could perhaps spread or not spread for the same reason that jokes do or don't.

But the real reason that rarity worries me as an explanans or explanandum is that in the history of linguistics many discussions of it go like this:

1. A is rare.
2. (Therefore) A is marked in some way.

3. A is marked because it is rare and rare because it is marked.

I talked about these things in my keynote address to the Rara and Rarissima conference a few years ago in Leipzig and mention them again in Cognitive Fire. Differences, especially rare ones, are exciting to study, whether coincidental or not, especially rara and rarissima, because they are what make each language distinctive. And, just as importantly,  because they might show that no theory can account for all of human language.

Peter Ladefoged and I made a similar point in our Language paper in the 90s, the Problem of Phonetic Rarities.


On 3 Oct 2010, at 21:25, Matthew S. Dryer wrote:

> Dan,
> Fritz actually asked two questions and it's not clear which of them you are
> referring to.  
> One was "Why is preposition stranding (or adposition stranding) of the English
> sort so rare?"
> The other was "If it's so rare, why is it so productive in English?"
> If your comment relates to the second of these questions, then you may be right.
> For example, another relatively unusual feature of English (though not quite so
> unusual) is having a glide r.  While it makes sense to ask why it is uncommon,
> it's not so obvious that there is any interesting answer to the question why
> English has a glide r when it is uncommon crosslinguistically.
> On the other hand, sometimes, there are interesting answers to these questions. 
> Some -though not many - languages have roughly equal amounts of prepositions and
> postpositions and there is often a clear answer to why this has happened: the
> language is VO with Genitive-Noun order and the prepositions arose from verbs
> while the postpositions arose from nouns in genitive-noun constructions.
> It's not clear to me whether or not there is any interesting answer to the
> question why English allows preposition stranding (although I shared some
> speculations separately with Fritz), whether it is like glide r in English, or
> the example discussed in the preceding paragraph.
> However, if your comment was about the first question ("Why is preposition
> stranding (or adposition stranding) of the English sort so rare?"), then I am
> more puzzled, since this question is no different in its logic from any question
> of the sort "Why do most languages have such-and-such a property?"  Saying that
> adposition stranding is rare is equivalent to saying that most languages don't
> allow adposition stranding.  Perhaps you aren't interested in explanations for
> universals (absolute or statistical), which is fine with me: I personally devote
> far more time to finding out what IS typical in language than in attempting to
> explain such generalizations.
> (I don't have access right now to a way to hearing things on the internet, so no
> way at the moment to listen to what you say at the link you provided.)
> Matthew
> On Sat 10/02/10  1:12 PM , Daniel Everett dan at daneverett.org sent:
>> One could approach the question, Fritz, from a different angle, i.e. why
>> wouldn't English's pattern be rare? There are various things to accomplish
>> and different peoples accomplish them in different ways. If you assume that
>> the grammatical structure is basic, then, sure, we might wonder why it
>> isn't found more places.
>> But one could also ask why country music isn't found in Africa.
>> Dan
>> On 2 Oct 2010, at 13:08, Frederick J Newmeyer wrote:
>>> Tom,
>>> Even if you are right that it is correct to
>> class together the phenomena that you call attention to, there is nothing
>> that you wrote that begins to explain why the English/Scandinavian pattern
>> ('Who did you talk to?') is so rare crosslinguistically. And that, after
>> all, was my question. And nothing (I think) that explains why, despite its
>> typological rarity and therefore possible 'nonfunctionality', English has
>> steadily expanded its stranded preposition possibilities over the
>> centuries, from topicalizations ('John, I would never talk to') to
>> wh-questions ('Who did you talk to') to passives ('John was talked
>> to').> 
>>> Its fine, I suppose, if you want to expand the
>> notion 'preposition stranding' to deal with words like 'intend' and
>> 'whereof', but that does not move us forward on explaining the
>> crosslinguistic rarity of the pattern:> 
>>> question-word (did) subject V P?
>>> Best,
>>> --fritz 
>>> Frederick J. Newmeyer
>>> Professor Emeritus, University of
>> Washington> Adjunct Professor, University of British
>> Columbia and Simon Fraser University> [for my postal address, please contact me by
>> e-mail]> 
>>> On Fri, 1 Oct 2010, Tom Givon wrote:
>>>> Au contraire, cher ami. There IS a very
>> general mechanism of zeoing of >> argument; in the 1960's terminology either
>> by 'movement' or by 'deletion. In >> the case of embedded REL-clauses, we tend to
>> see it as 'deletion. In the case >> of AGT-deletion passives, too, maybe. In the
>> case of WH-questions, we call it >> 'movement'. In the case of promotion to DO
>> (as in Rwanda), maybe 'movement' >> again. The common denominator is that the
>> noun that carried the ad-position >> is now missing from its 'normal'
>> (high-frequency) position, and what shall we >> do with the  poor beached-whale
> adposition?
>> It carries vital information >> about GRs or SR's, so we can just pitch it
>> (we pitch the noun because of >> predictability, but the adposition is less
>> predictable).>> 
>>>> In the case of English WH-question, you go
>> back to 18th Century written >> English, you find preposition migrating to
>> the WH-word. And this already >> appears at the same period (& earlier)
>> with REL-clause subordinators such as >> 'whereof', 'whereas', 'whereat',
> 'wherefor',
>> 'whereto', 'wherein' etc. Also, >> incidentally, in non-embedded
>> referring/anaphoric expressions such as >> 'thereof', 'thereby', 'thereat',
> 'thereto',
>> 'therein', 'therefor', etc. It is >> fairly clear, further, that the use of the
>> 'where-PREP' pattern in English >> REL-clauses hitched a ride on the earlier
>> use in WH-questions. Such >> hitchiking, precisely in this direction, is
>> extremely common, and has cropped >> up later on in English again (in spoke
>> German 'with /wo/, in Greek with >> /pou/, in Kriop with /w(h)e(re)', in  spoken
>> Hebrew, etc.) These are >> extremely mundane facts, Matt, readily
>> available, both in  old texts and in >> the lit. (Bern Heine wrote about it in 2007
>> & 20089, inter alia). All it >> takes is looking, and I suppose seeing. So
>> just buy some old books & start >> reading them. Sure, there's a lot of
>> complexity in those pathways. But still, >> with all this diversity, there are some
>> clear central MECHANISMS of >> emergence, not only a collection of surface
>> patterns. After all, Fritz didn't >> only ask if the patterns are rare (he, I
>> think naively, assumed that). He >> also wanted to know--or so I gave him credit
>> for (stranded PREP again, >> dammit!)--why. A collection of facts is
>> decidedly not an answer to a WHY >> question, but qat best the reason for asking
>> it.  TG>> 
>>>> ==============
>>>> Matthew S. Dryer wrote:
>>>>> Not so fast, Tom.  It is certainly true
>> in principle that one can often >>> better
>>>>> understand why a rare phenomenon is rare
>> by getting a better understanding >>> of
>>>>> more common related phenomenon.  But I
>> see nothing in your two emails that >>> sheds
>>>>> any light on why the English-type of
>> adposition stranding is so rare or how >>> any
>>>>> of the literature on the related
>> phenomena you discuss sheds any light on >>> this
>>>>> question.  Unless you can do that, I see
>> no reason why it would be worth >>> Fritz
>>>>> looking at these related phenomena, to
>> help answer his question.>>> 
>>>>> Matthew
>>>>> On Fri 10/01/10  5:40 PM , Tom Givon tgivon at uor
>> egon.edu sent:>>> 
>>>>>> Copy of note to Fritz:
>>>>>> From where I sit, it is all
>> connected, both synchronically (similar >>>> pattern) and diachronically
>> (patterns mutating into other patterns). There >>>> are grammatical
> constructions that
>> act as context for the original>>>> 'stranding'; then you have various
>> next-steps, eventually to (in some >>>> cases) full lexicalization (as in
>> Latin or Germanic). So in Rama, the >>>> exact same configuration as in IE
>> exists, but it is a much earlier stage, >>>> so I can see the early 'trapping'
>> process a bit more clearly. In>>>> Latin & Old Gothic it's already
>> too advanced, hard to see the variational >>>> steps any longer, it is largely 
>> already lexicalized. In Rama you can see >>>> just the beginning of
>> lexicalization, a few compound>>>> verbs.
>>>>>> In Klamath or Numic you can see much
>> more, a host of it, tho you can still >>>> see the nominal or verbal etymology
>> of the ad-positions. In Latin>>>> they LOOK like they should be
>> de-verbal, as in Rama, but the etymology is >>>> not quite as clean, too much time
>> has pass. So you still have the verb >>>> 'ex-it' on 'en-ter', but it's harder
>> to find the verb 'con' or 'sur' or >>>> 'per'; tho in Spanish 'sub-ir' is
>> still a verb meaning 'go down/under'.>>>> 
>>>>>> But In Bantu the grammatical process
>> is much more advanced that in Rama,>>>> it gotten into  REL clsauses,
>> passives, and other derivatives from them.>>>> And there's a considerable amount of
>> lexicalization, mostly in set-phrases >>>> (typical early stage) such as
>> 'excuse me', 'thank you' 'how>>>> are you' & more. In all these
>> cases, you can see the role of >>>> zero-arguments right there (missing
>> AGT-of-passive, zero coreferent inside >>>> the REL-clause. My las supervised
>> African dissertation, a grammar>>>> of Lunda (Boniface Kawasha, ca.
>> 2002, U. Oregon) has tons of that in >>>> REL-clauses, it is like the
>> promotion-to-DO in Rwanda (Kimenyi 1976), but >>>> only in REL clauses, not main
>> clauses. I flashed on this when I did>>>> my dissertation on Bemba (1969).
>> Then, my supervisor, Paul Schachter, said >>>> "you've got too much in it
>> already, I don't want to read a whole>>>> grammar". So eventually I
>> dumped two boxes of data. Sic transit.>>>> 
>>>>>> You gotta open up your
>> classification schemata just a little bit, Fritz.>>>> Otherwise you'll keep
> missing the
>> real goodies, where the explanations of >>>> typological differences lie--usually
>> in plain site.  TG>>>> 
>>>>>> ======================
>>>>>> Angus B. Grieve-Smith
>> wrote:>>>> 
>>>>>>> On Fri, October 1, 2010 12:16
>> pm, Frederick J>>>>> 
>>>>>> Newmeyer wrote:>      
>>>>>>>> Dear Funknetters,
>>>>>>>> Does anybody know of a
>> functional>>>>>> 
>>>>>> explanation (published or not) for
>> why>> preposition stranding is so rare >>>> in the
>>>>>> languages of the world? (I
>> am>> referring to constructions such as 'Who >>>> did
>>>>>> you talk to?', 'Mary was>>
>> talked to', etc.) As far as I know, it>>>> exists only in Germanic,
>> marginally>> in French, and possibly in some >>>> Niger-Congo
>>>>>> languages. There are a
>> number>> of functionally-oriented accounts of>>>> P-stranding in English, but I
>> wonder>> if anybody has taken on the >>>> question of its
>>>>>> rarity crosslinguistically.>> 
>>>>>>> In order to have preposition
>> stranding, you need>>>>> 
>>>>>> prepositions, right?  So> the
>> only way we can answer the question of how>>>> rare languages with> preposition
>> stranding are is by getting a rough>>>> sense of the proportion of>
>> languages with prepositions they represent. >>>> Mr.
>>>>>> Givon mentioned a bunch> of
>> languages with them, but is there a>>>> comprehensive list in some
>> typology> text somewhere?>>>> 
>>>>>>> I also wanted a clarification
>> from Mr. Newmeyer:>>>>> 
>>>>>> your category of> preposition
>> stranding includes (1) but not (2),>>>> right?>
>>>>>>> 1) Who are you going
>> with?>>>>> 2) Are you coming
>> with?>>>>> 

More information about the Funknet mailing list