Syntax random

[bradrn]

A further question: under what conditions can the subject be deleted?

I think it can always be deleted:

Reading to yourself is a good habit.
Reading to himself is how Felix gets to sleep.
She's always in the back row reading to herself.

Oh, that’s true; I’d forgotten that English nominalisations always allow subject deletion. (I was actually asking the question more generally, not just about nominalisations, but I think your answer applies more generally as well.)

[dɮ the phoneme]

Let me know if this question is better placed in another thread, but does anyone know of any pervasively head-final languages with wh-movement? Presumably if the wh-phase moves to the CP head as in English, you'd get rightward wh-movement, right? But the WALS chapter on wh-movement says this is extremely rare or not attested. What's going on with the asymmetry?

[Moose-tache]

Let me know if this question is better placed in another thread, but does anyone know of any pervasively head-final languages with wh-movement? Presumably if the wh-phase moves to the CP head as in English, you'd get rightward wh-movement, right? But the WALS chapter on wh-movement says this is extremely rare or not attested. What's going on with the asymmetry?

Wh-fronting in languages like English is a form of topical fronting, not moving the Wh-word into clause head position (though they are often the same). Languages like Japanese, Korean, and Choctaw all allow Wh-fronting, usually with similar rules to topical fronting, but I think full-on OSV sentences would probably only sound natural in very specific circumstances.

[dɮ the phoneme]

Right, wh-words end up in Spec-CP rather than the CP head, or something like that? My syntax knowledge is very rusty. I'm working on a head-final language at the moment, with relative clauses preceding the head noun, and I want to use European style relative pronouns that indicate the relativized position via case. So I'd like to end up with a construction like "I saw whom boy" for the "the boy I saw", etc. Is something like this attested at all?

[bradrn]

Today I saw an advertisement with an interesting line: ‘[Companyname] online delivers to you’. This is obviously an instance of noun incorporation (which clearly is more productive than I thought in English). However, what I find interesting here is the word order. English is an SVO language, so I might naïvely expect an incorporated object to go after the verb — but here it is the opposite order. Does anyone have an explanation (diachronic or otherwise) for why this happens? And are there any other incorporating SVO languages which show this seemingly counterintuitive word order?

[zompist]

Today I saw an advertisement with an interesting line: ‘[Companyname] online delivers to you’. This is obviously an instance of noun incorporation (which clearly is more productive than I thought in English). However, what I find interesting here is the word order. English is an SVO language, so I might naïvely expect an incorporated object to go after the verb — but here it is the opposite order. Does anyone have an explanation (diachronic or otherwise) for why this happens?

Why would this be noun incorporation? "Online" is an adverb. I think the unmarked order for this sentence would be "<Company> delivers online to you" or "<Company> delivers to you online." It's a bit weird to me to put "online" before the verb... but that's a perfectly fine place to put an adverb. Try replacing it with one of: always, happily, promptly, quickly, electronically.

But you also want to know about noun incorporation. So far as I can see, NV is the normal order for this in English: head-hunt, babysit, house-break, mind-read, ice-skate, pub-crawl, cherry-pick.

[bradrn]

Today I saw an advertisement with an interesting line: ‘[Companyname] online delivers to you’. This is obviously an instance of noun incorporation (which clearly is more productive than I thought in English). However, what I find interesting here is the word order. English is an SVO language, so I might naïvely expect an incorporated object to go after the verb — but here it is the opposite order. Does anyone have an explanation (diachronic or otherwise) for why this happens?

Why would this be noun incorporation? "Online" is an adverb. I think the unmarked order for this sentence would be "<Company> delivers online to you" or "<Company> delivers to you online." It's a bit weird to me to put "online" before the verb... but that's a perfectly fine place to put an adverb. Try replacing it with one of: always, happily, promptly, quickly, electronically.

Ah, that is true, sorry. (I typically use ‘online’ as an adjective or noun, rather than an adverb.)

But you also want to know about noun incorporation. So far as I can see, NV is the normal order for this in English: head-hunt, babysit, house-break, mind-read, ice-skate, pub-crawl, cherry-pick.

Yes, so my question still applies: why does English have OV incorporation when word order is SVO? And are there any other languages which do so as well?

[zompist]

But you also want to know about noun incorporation. So far as I can see, NV is the normal order for this in English: head-hunt, babysit, house-break, mind-read, ice-skate, pub-crawl, cherry-pick.

Yes, so my question still applies: why does English have OV incorporation when word order is SVO? And are there any other languages which do so as well?

My guess is that it leaves the verb at the end, where it can do verby things. "She babysat twice last week" is a lot more natural than "She sitbabied twice last week."

[bradrn]

But you also want to know about noun incorporation. So far as I can see, NV is the normal order for this in English: head-hunt, babysit, house-break, mind-read, ice-skate, pub-crawl, cherry-pick.

Yes, so my question still applies: why does English have OV incorporation when word order is SVO? And are there any other languages which do so as well?

My guess is that it leaves the verb at the end, where it can do verby things. "She babysat twice last week" is a lot more natural than "She sitbabied twice last week."

I’m not convinced: just because the noun is at the end of the verb complex, that doesn’t mean the noun needs to be the inflected part! A V+[N+inflection] construction like ‘I skate-iced’ is certainly unnatural, but a [V+inflection]+N construction like ‘I skated-ice’ is no less natural than the N+[V+inflection] order of ‘I ice-skated’.

[zompist]

My guess is that it leaves the verb at the end, where it can do verby things. "She babysat twice last week" is a lot more natural than "She sitbabied twice last week."

I’m not convinced: just because the noun is at the end of the verb complex, that doesn’t mean the noun needs to be the inflected part! A V+[N+inflection] construction like ‘I skate-iced’ is certainly unnatural, but a [V+inflection]+N construction like ‘I skated-ice’ is no less natural than the N+[V+inflection] order of ‘I ice-skated’.

Well, no, because English does not allow inflections in the middle of a word.

[Richard W]

Yes, so my question still applies: why does English have OV incorporation when word order is SVO? And are there any other languages which do so as well?

Because it's GN and Adj+N.

[bradrn]

Yes, so my question still applies: why does English have OV incorporation when word order is SVO? And are there any other languages which do so as well?

Because it's GN and Adj+N.

So are you saying that the order of a verb and its incorporated noun is more closely correlated to the noun/modifier order than the V/O order? (I assume ‘GN’ here is ‘genitive+noun’.) That does sounds reasonable — but on the other hand, it’s not too hard to find counterexamples. (e.g. Skou has ‘noun adjunct’ constructions which are syntactically very similar to noun incorporation, in which the noun is placed before the verb, but Skou also has postposed adjectives.)

My guess is that it leaves the verb at the end, where it can do verby things. "She babysat twice last week" is a lot more natural than "She sitbabied twice last week."

I’m not convinced: just because the noun is at the end of the verb complex, that doesn’t mean the noun needs to be the inflected part! A V+[N+inflection] construction like ‘I skate-iced’ is certainly unnatural, but a [V+inflection]+N construction like ‘I skated-ice’ is no less natural than the N+[V+inflection] order of ‘I ice-skated’.

Well, no, because English does not allow inflections in the middle of a word.

But how do we know that it’s one word? And besides, which is the relevant definition of word here anyway? (Besides, I believe there do exist languages which allow inflections in the middle of words, though I can’t quite remember which ones.)

[KathTheDragon]

I'd wager the OV order is historical, as a retention from the verb-final PIE. O-V compounds thus mirror Adv-V compounds in reflecting a straight univerbation of a verb phrase.

[Richard W]

Well, no, because English does not allow inflections in the middle of a word.

But how do we know that it’s one word? And besides, which is the relevant definition of word here anyway? (Besides, I believe there do exist languages which allow inflections in the middle of words, though I can’t quite remember which ones.)

Of course, English does have a few mutation plurals, such as tooth/teeth, indivisible forms such as paterfamilias/patresfamilias, menage a trois/menages a trois and Knight Templar/Knights Templar, and flexible compounds such as Lords Temporal and Lords Spiritual (conjoinable as Lords Spiritial and Temporal). Phrasal verbs may be examples in some lects, though in mine, the ready incorporation of objects argues against their being single words. Agent nouns are a bit awkward or even debatable, like colloquial eater upper and fully standard washers up.

[Richard W]

I'd wager the OV order is historical, as a retention from the verb-final PIE. O-V compounds thus mirror Adv-V compounds in reflecting a straight univerbation of a verb phrase.

G+N and Adj+N seems to fit with SOV, though there is a recent claim that the correlation is merely a Eurasiatic inheritance / areal feature. However, to ice-skate feels more like a back formation from ice skater and ice skating, where G+N comes into play. Progressive forms feel less unnatural - as though they should be analysed as containing gerunds or participles rather than simply being verbal inflections. In this particular example, ice is not the object of to skate. Also, O-V ordering doesn't look like a O-V univerbation - IE O-V compounds are primarily nominal (or rather adjectival); one needs other elements to make a verb. Or am I overlooking clear counter examples?

[zompist]

Well, no, because English does not allow inflections in the middle of a word.

But how do we know that it’s one word? And besides, which is the relevant definition of word here anyway?

Calling it noun incorporation already assumes that it's one word!

Phonologically, these words have just one primary stress. Syntactically, nothing can come between the components.

Wordishness is a continuum, and many of these are not quite full words; this is reflected in the fact that some require a hyphen, and verbal inflection can sound a bit weird. (This seems most evident with nonce coinages. I could get away with "I'm going to lexicon-tinker all night", but maybe not "I lexicon-tinkered all night".)

(We can also contrast these with verb-particle compounds, which are lexical entries but resist word status. They are still separable: "I could eat that plate of cookies right up." Logically, we could fully compound and transfer the inflection to the particle— "I eatupped the cookies"— but we just don't. Note that we can do it if neither component is a verb: "I one-upped my brother.")

Besides, I believe there do exist languages which allow inflections in the middle of words, though I can’t quite remember which ones.)

This isn't relevant to whether English can do it!

(Other languages allow VO compounds, notably Mandarin. OV order certainly isn't a universal!)

Richard's examples (like ménages à trois) are more interesting, though I'd point out that these are either foreign borrowings, or imitations of French. "Passersby" is attested, but there's a tendency to regularize it to "passerbys". So, like so many things in English, the prohibition isn't absolute. But overwhelmingly, English doesn't like inflections within a compound, so I think it's not surprising at all that the end of the verb, where inflections go, is preserved. The same thing happens with noun compounds: madman > madmen.

I'd wager the OV order is historical, as a retention from the verb-final PIE.

All of the words I've found seem pretty new. E.g. "babysit", surely one of the more established of these terms, is mid-20th century. I don't see how Old English habits would be relevant. But I'd love to see the oldest citation for such terms.

[zompist]

Sorry, one more thought. My impression is that we use O+V compounds as nouns far more readily than as verbs: "man-eating tiger" is fine, so is "man-eater", but "*The tiger man-ate all week" is not. It wouldn't surprise me that most or all of the actual verbs are back-formations from these.

[Richard W]

"Passersby" is attested, but there's a tendency to regularize it to "passerbys".

I'm not sure I've come across the latter - though I might instantly have corrected it to passersby. Although we have pushups (though I'm not happy with the spelling of that word), I have hangers-on.

I'd wager the OV order is historical, as a retention from the verb-final PIE.

All of the words I've found seem pretty new. E.g. "babysit", surely one of the more established of these terms, is mid-20th century. I don't see how Old English habits would be relevant. But I'd love to see the oldest citation for such terms.

Well, you can't get any older than Old English and still call it English

I'll open the bidding with 17th century kidnap, though Onions derives it from 17th century kidnapper in line with my view that it's a back-formation. Incidentally, another back-formation route is We've been blackerry picking, which could be interpreted simply as We've been picking blackberries rather than implying going somewhere and returning. [Postscript: I suspect I'm not checking for more recent posts between reading to the end of the thread and starting to compose a reply - I sometimes have to research before replying.]

English has had to handfast since the 14th century, but before that we only have the past participle going back to the 12th century and presumed to be borrowed from Old Norse handfesta.

[Richard W]

Brothers-in-law isn't based on French.

Hasn't, hadn't, doesn't, didn't. On the other hand, don and doff go stem-adverb-inflection.

[bradrn]

Well, no, because English does not allow inflections in the middle of a word.

But how do we know that it’s one word? And besides, which is the relevant definition of word here anyway?

Calling it noun incorporation already assumes that it's one word!

Huh? No it doesn’t! There’s quite a few isolating languages with noun incorporation. (Niuean comes to mind; pity I can’t find a reference grammar to figure out whether its NI is within a single word or spans multiple words.)

Phonologically, these words have just one primary stress. Syntactically, nothing can come between the components.

That phonological criterion at least is debatable: it’s easy to apply with ice-skate, but I’d argue that e.g. mountain-climb has two stresses.

Wordishness is a continuum, and many of these are not quite full words; this is reflected in the fact that some require a hyphen, and verbal inflection can sound a bit weird. (This seems most evident with nonce coinages. I could get away with "I'm going to lexicon-tinker all night", but maybe not "I lexicon-tinkered all night".)

Personally, I’d say that this has more to do with the productivity of incorporation (or a lack thereof) than wordhood: incorporated predicates are relatively unproductive, so resist new coinages.

(We can also contrast these with verb-particle compounds, which are lexical entries but resist word status. They are still separable: "I could eat that plate of cookies right up." Logically, we could fully compound and transfer the inflection to the particle— "I eatupped the cookies"— but we just don't. Note that we can do it if neither component is a verb: "I one-upped my brother.")

Interesting… I hadn’t noticed the different order. I might conjecture that it comes from the influence of prepositions: I [ate up] the plate of cookies can easily be reanalysed as I ate [up the plate of cookies]. (I find it interesting to compare this to other IE branches, e.g. Indo-Iranian, where incorporation consistently has a modifier+modified order for both noun and particle compounds… maybe this has something to do with their SOV word order?)

Besides, I believe there do exist languages which allow inflections in the middle of words, though I can’t quite remember which ones.)

This isn't relevant to whether English can do it!

…

Richard's examples (like ménages à trois) are more interesting, though I'd point out that these are either foreign borrowings, or imitations of French. "Passersby" is attested, but there's a tendency to regularize it to "passerbys". So, like so many things in English, the prohibition isn't absolute. But overwhelmingly, English doesn't like inflections within a compound, so I think it's not surprising at all that the end of the verb, where inflections go, is preserved. The same thing happens with noun compounds: madman > madmen.

I did find myself wondering how you came to the conclusion that ‘English does not allow inflections in the middle of a word’, but I had forgot about this sort of reanalysis, which is indeed strong evidence that English hates middle-of-the-word inflections. (Even middle-of-the-phrase inflections are undesirable, with the best examples being French loans: attorneys general → attorney generals, sergeants major → sergeant majors.)