rarity of preposition stranding

E.G. eitan.eg at gmail.com
Fri Oct 1 21:11:05 UTC 2010

Hi all,

First to add another language, which does it very sporadically: Modern
Hebrew, in phrases like 'lalexet im ve-lehargiS bli' (to_go with and-to_feel
without'), taruc im targiS bli (run with feel without), the latter said
about running shoes that allegedly make the user feel barefoot. However,
it's pretty restricted, and you wouldn't normally say anything equivalent to
"Who did you talk to?"

A big problem, it seems to me, is that 'adposition' isn't a very well
defined term. In fact, a lot of books gloss over the definition, saying
things like 'a preposition is the head of a PP,' which is tautological (or
wrong, depending on your point of view). If you look at descriptive
grammars, you'll find the same element described as an adposition, a case
marker, a 'class three clitic,' a 'rank 4 suffix' and so on. This
terminological problem seems to be especially acute with respect to elements
that are sometimes called postpositions. There's a good discussion of the
problems in Pietro Bortone's "Greek Prepositions: From Antiquity to the
Present" (Oxford, 2010) –– in fact, a very useful typological study. It is
worth noting that there are a lot of descriptive categories one could set up
between lexical item and case-affix, and elements somewhere between the two
could often be called adpositions. Bortone notes colloquial Turkish
comitative -le/-la, which is usually described as a postposition but seems
to act more like a case ending.

This terminological and conceptual vagueness leads to a situation in which
phenomena similar to the one you describe would not be called preposition
stranding, as the 'free morpheme' might be called a 'resumptive adverb' or
something else. This is the case for Ancient Egyptian, in which one can find

a. bw nty Hm=f im=f (place REL majesty=3sg.m in=3sg.m) 'the place in which
His Majesty is' (lit. place that his Majesty in it)
b. bw nty Hm=f im=ø (place REL majesty=3sg.m in) 'the place in which his
majesty is' (lit. place that his Majesty is in)

In the second example, im is considered to be a resumptive adverb by
traditional grammars, although it would probably answer to the notion of a
stranded preposition.

So the point to be made here is that the phenomenon might be better attested
than it seems but obscured by the diversity of descriptive grammatical
terminology. Probably a language won't have been considered as having
'preposition stranding' unless it's been described in a generative

Another point is that for early Indo-European languages, it's not so obvious
that Prof. Newmeyer's question is the right one to ask. For example, I would
have a look at Silvia Luraghi's very interesting "On the meaning of
prepositions and cases" (John Benjamins, 2003). She discusses Homeric Greek,
in which the 'proper' prepositions can occur as 1. preposition, 2. free
adverbs, and 3. preverbs (which is what Prof. Givón was alluding to, if I'm
not mistaken). She considers the problem of categorial assignment ('are
these elements adverbs or prepositions?') to be a pseudo-problem. Bortone
adduces examples like en used adverbially meaning 'inside' -- but to a
generative perspective, wouldn't this look like stranding, e.g., "in (it")?

I haven't had a chance to read Claude Hagège's "Adpositions" (Oxford, 2010)
properly yet, but it seems that he considers stranding to be the result of
the non-tonicity of adpositions.

In any event, since adpositions aren't well-defined as a concept for
cross-linguistic comparison, i.e., not well distinguished from other kinds
of elements, and the descriptive terminology used for relators in different
languages – and often for the same language –  tends to vary extensively, it
is hard to know whether this observation about the rarity of adposition
stranding is even right.

Best wishes,

On 1 October 2010 21:13, Frederick J Newmeyer <fjn at u.washington.edu> wrote:

> On Fri, 1 Oct 2010, Tom Givon wrote:
>  Well, I DID mean massive. I'm not as well-versed in Germanic, tho I see it
>> there too (Bernd Heine could tell you aplenty). So just think Latin for a
>> sec: Pre-tend, ex-tend, in-tend, con-tend; per-tain; con-tain, re-tain,
>> su(b)-stain, main-tain, ob-tain; re-pulse, ex-pulse, im-pulse, com-pulse;
>> re-ject, e(x)-ject, in-ject,  ob-ject; con-ject(ure); con-struct,
>> in-strtuct, de-struct, re-struct(ure);  etc. ect. ect.  There's a whole page
>> of those in my Syntax vol. I (2001), one of the early chapters, mostly
>> talking about the metaphoric etymology, which we know well. (George made a
>> lot of hay off this, claiming that metaphors never die, they just go & get
>> reified in some lexical Heave...). But we also know a lot (well, some of us
>> do, maybe) about the diachronic-syntax pathways that lead to such
>> 'stranding', & how it connect to the type of ad-position, earlier vs. later
>> WO, zero-anaphora of both types, the availability of other clitic-trapping
>> word-types, ets. All that is needed is widening our typological--and
>> diachronic, really the same thing--horizons just a little bit and what seems
>> to you so exceptional reveals itself to be rather massive. Best,  TG
> I see. Then we mean something very different by 'preposition stranding'.
> Let me rephrase my question:
> Does anybody have an explanation for why constructions of the following
> form are so rare crosslinguistically:
> question-word (did) subject V P?
> ...where 'question-word' is a free morpheme and understood as the object of
> P.
> Such constructions are extremely rare in I-E and crosslinguistically, as
> far as I know.
> --fritz

Eitan Grossman
Martin Buber Society of Fellows
Hebrew University of Jerusalem

More information about the Funknet mailing list