In a related but different question to Indo-European prepositions: whence did they come?, why do just about all modern Indo-European languages have prepositions rather than postpositions?

PIE is believed to have been an SOV language, and like textbook examples of SOV languages it had relational postpositions rather than prepositions. Yet every modern Indo-European language I know of uses prepositions primarily (one or two have postpositions as well, but in such languages prepositions are a clear majority). Even Latin, which looks to have a nominal word order of SOV (though the inflectional case system allowed for large variation in word order) has exclusively prepositions (excluding the comitative pseudo-case where previously postpositional "cum" - which, I may add, is a preposition in Latin - was reduced to a suffix).

How is it that just about every Indo-European language converted to prepositions (presumably) in parallel, when such a change contradicts typical word order laws?

  • 1
    Latin also has the suffix -tenus, and semi-prepositions gratia and causa, and probably a few more that I can't think of—but, certainly prepositions are the overwhelming majority. As to why prepositions may be preferred: they can be more convenient when marking large noun phrases. They turned the lavish gifts, provided by Cleopatra in twenty golden boats, steered by beautiful maidens, away. ... – Cerberus Sep 2 '12 at 20:34
  • 1
    ... Here the particle-preposition would make the sentence much easier to parse if it had gone before the lavish gifts, because it is an essential part of the construction, which we prefer at an earlier position, since it is needed to interpret the whole sentence. The house [my father's second wife, who left Athens around the turn of the century]-of — here I have simulated putting the true preposition of after the noun phrase it governs, which results in a sentence that is equally difficult to parse until you reach the end. ... – Cerberus Sep 2 '12 at 20:36
  • ... Something like this could have been a part of why we have prepositions and not postpositions. Cases, which often have functions similar to prepositions, do not have this problem. – Cerberus Sep 2 '12 at 20:39
  • 1
    I'm not sure either part of this sentence is true: PIE is believed to have been an SOV language, and like textbook examples of SOV languages it had relational postpositions rather than prepositions. Many people think PIE had neither prepositions nor postpositions (see this recent LL post, which partly answers your question about modern IE languages, too). And given that word order is quite free in several early IE languages, it's not so clear that PIE had a fixed basic word order of any kind. – TKR Jan 15 '14 at 22:21
  • 1
    I was under the impression that Indic languages had postpositions. I would also think that there are more Indic languages extant today than non-Indic Indoeuropean languages. – hippietrail Feb 19 '14 at 6:47

Most prepositions are originally, most likely, nouns of their own, in different fossizilied case forms that preceeded or followed the corresponding noun. For example, Latin pro, de are IIRC believed to stem from ablative forms prod, ded. Coincidentally, they also require ablative. So, a preposition was originally some sort of abstract apposition to the noun being modified.

So the answer why we have prepositions instead of postpositions is probably the same answer why PIE's appositions usually place more abstract entities before less abstract entities, e.g. mother Earth, my friend Alice etc. In the same way, I'd think, pro me (< prod med) would mean "front-ABL me-ABL", just two nouns in apposition.

Then it developed into full-fledged prepositions...

  • PIE had no ablative. – Anixx Mar 15 '14 at 20:39
  • 2
    @Anixx I've never seen a reconstruction of PIE without an ablative. What do you call the form that's ancestral to Sanskrit ašvāt, Old Latin equōd, etc., if not ablative? – TKR Mar 21 '14 at 17:10
  • @TKR You are correct, I confused it. – Anixx Mar 22 '14 at 5:24
  • Assuming they were originally nouns: in Latin mecum was acceptable, in Armenian im het is literally my with. (And in both languages it would still be mother earth.) – Adam Bittlingmayer Jul 30 '16 at 6:47
  • As far as origin as nouns, I think the origins are in fact mixed. For those that take genitive, it makes sense. But in 20 years ago or 20 years back, where ago and back start to function like postpositions, it would be misleading to see back as a noun even though there happens to be a n back too. – Adam Bittlingmayer Jul 30 '16 at 6:54

I can hardly improve on this observation from the English Wikipedia article Preposition and postposition:

Whether a language has primarily prepositions or postpositions is seen as an aspect of its typological classification, and tends to correlate with other properties related to head directionality.

Since an adposition is regarded as the head of its phrase, prepositional phrases are head-initial (or right-branching), while postpositional phrases are head-final (or left-branching).

There is a tendency for languages that feature postpositions also to have other head-final features, such as verbs that follow their objects; and for languages that feature prepositions to have other head-initial features, such as verbs that precede their objects.

This is only a tendency, however; an example of a language that behaves differently is Latin, which employs mostly prepositions, even though it typically places verbs after their objects.

Branching direction itself is likewise just an inconsistent tendency with multiple dimensions - variant, emphasis and of course time in history, as these features evolve. Left-branching approaching consistency is an areal feature found mostly in South, Central and East Asia or languages like Turkish that migrated from there relatively recently.

Thus this has little to do with SVO/SOV directly or Indo-European languages specifically. Plenty of IE languages use or used postpositions or both pre- and postpositions, and plenty of IE languages have or had left-branching or just relatively free word order, and some major neighbouring non-IE languages like Arabic and Hebrew use mostly prepositions.

  • 1
    Latin is also ambivalent about gapping. The famous first sentence of Caesar's Bello Gallico exhibits gapping twice, once to the left and once to the right. English gaps to the right, as a right-branching language ought, and Japanese to the left, as a left-branching language ought. But Latin swings both ways. I suspect it's in transition from SOV to SVO; the older postpositions have become auxiliaries or affixes, and the upstart prepositions are horning in. And they kept horning in after the fall of the case markers until they took up all the slack. – jlawler Jul 29 '16 at 14:31

Yet every modern Indo-European language I know of uses prepositions primarily (one or two have postpositions as well, but in such languages prepositions are a clear majority)

I think you are focusing too much on European IE languages here: if you look at the more eastern branches (Armenian, Indic, Iranian), postpositions are quite common, and may even predominate over prepositions.

For example, in Armenian, the basic way to say "in", "for", "upon", etc. involves an oblique case-form of the noun followed by a postposition. Although Farsi (Persian) prefers prepositions, most other Iranian languages don't, as far as I know.

The preference for prepositions in European languages may simply be an old areal feature, to which there are still many exceptions (cf. German Schau mich an! "Look at me!", literally "look me at").

  • 2
    The German example does not illustrate a postposition, it illustrates the seperable verb anschauen. – jk - Reinstate Monica Jul 30 '16 at 19:31
  • That may be how some grammar books describe it, but how is "postposition" not also a valid description in this case? As a preposition an means "at", and it also (to my knowledge) means "at" in the context of Schau mich an. If it is performing its prepositional function after a noun/pronoun, this seems to imply that it is functioning postpositionally. – user8017 Jul 30 '16 at 21:54
  • 2
    There can be other phrases intervening between mich and an, e.g., *Schau mich morgen im Fernsehen an*–not exactly the behaviour of a postposition, but typical for a separable verb. – jk - Reinstate Monica Jul 31 '16 at 14:55
  • If an is not strictly a postposition in that case, then it does at least seem to be an adverbial use of an (not tightly connected to any noun), so it remains an "exception" in the way I described above, where a word normally described as a "preposition" is not actually preposed to the noun. – user8017 Jul 31 '16 at 20:09
  • 1
    Whatever you call it, it is definitely not a postposition. The Stuttgat-Tübingen POS tagset classifies it as a partikel: PTKVZ - Abgetrennter Verbzusatz – jk - Reinstate Monica Aug 1 '16 at 9:17

i think, probably position of pre/post-positions related to phrase with what they govern, was one of classification criteria when classifying some languages as indoeuropean. if so, for that, the requested language, strongly as defined in the question, cannot be.

"indirect answer", "flexible answering":

maybe, indoeuropean languages are chosen as indoeuropean languages not only because of lexicon like swadesh list, but for common grammar (prefixes, cases, their positions), in that case, if they rechose the languages only by "simple lexicon", who knows, maybe uralic languages should be accounted as indoeuropean languages?

uralic languages are not indo-european, but have many common roots, but have agglutinative grammatic morphemes like in altaic languages.

uralic languages are accounted as relative language (what is proper term for it?) for indo-european, see this tree: http://mentalfloss.com/article/59665/feast-your-eyes-beautiful-linguistic-family-tree .

i have copy-pasted second part of answer with small changes from my answer Is there any agglutinative Indo-European language? .

  • 2
    Too much of speculation without supporting facts or references. – jk - Reinstate Monica Jul 27 '16 at 9:52
  • @jknappen but i do not put "speculation" in place of proved thing. i say "who knows, maybe" – qdinar Jul 27 '16 at 10:43
  • 1
    Grammatical morphemes are actually a better indicator of genetic relatedness than lexicon, so there's a good reason for not grouping languages based on lexicon only. – TKR Jul 27 '16 at 22:55
  • @TKR ok but, i see now my answer has got one more minus, and it has -3 now, and i am afraid that somebody understand your comment as a strong counter-argument, and i say: question itself requires searching of indoeuropeic languages with other grammar. – qdinar Jul 28 '16 at 14:21
  • 1
    That would be true if having prepositions vs. postpositions was one of the criteria linguists use for identifying Indo-European languages, but it really isn't. – TKR Jul 29 '16 at 19:05

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.