You can get preposition stranding right to start with

« previous post | next post »

John McIntyre notes on his blog You Don't Say that a man named Rod Gelatt, a retired professor of journalism who taught at the Missouri School of Journalism, writes in a letter to the Columbia Missourian newspaper (responding to an article calling for more attention to correcting grammar errors in online content):

in the announcement of the invitation for us to become grammar police, I found two errors: "….who wants to generously point out…" (splitting an infinitive) and "Spell check won't help you when you have the wrong word to start with" (ending sentence with preposition).

I ignore the first point (split infinitives have always been grammatically correct in English; see for example this page). And as for the second, stranded prepositions have also always been grammatical in general, of course; but with respect to Mr Gelatt's example, I wonder what he thought the "correction" would be? The common phrases to start with and to begin with are among the (numerous) cases where stranding the preposition at the end of the phrase is not just permitted in Standard English, it's obligatory.

That is, You can wipe that smile off your face to start with is grammatical in Standard English, but *You can wipe that smile off your face with which to start is not. So Mr Gelatt has it ass-backward: not keeping the preposition at the end here would be the grammatical error.

I agree with John McIntyre that it is a bit scary to think that this man spent a career "standing before the impressionable young" and packing their heads with arrant nonsense that editors like John ultimately have to try and rectify by returning the victims to a state in which they can write their own native language sensibly.

It's another illustration of why I am worried that prescriptivism harms the economy: think of the senselessly wasted thousands of hours each year as dim-witted journalism professors with old-fashioned ideas teach falsehoods about English out of hundred-year-old books of toxic waste (you know which sort of book I mean) so that editorial staff members of newspapers can later spend their expensive time struggling to shake the poor graduates out of their didactogenic misconceptions and get their writing back into a state where it's fit to publish.

Addendum: A particularly dumb comment that originally appeared below said this:

IMHO, the words classified as "prepositions" are, in fact, prepositions ONLY when they are in the "pre-"position in prepositional phrases. Otherwise, they are simply one-word adverbs of location or direction. In verbal forms such as "to start with", the "with" is a meaning-mandated adverb which may occur in any normal adverbial position. If you take this view, the whole question of "stranded prepositions" simply goes away.

I couldn't bear to let such painfully misinformed pomposity ("I[n] M[y] H[umble] O[pinion]", indeed!) remain on the Language Log site; yet it seemed churlish to attack one of our guests. So I deleted the comment, and present its contents anonymized here for correction. It certainly is painfully misinformed, and I need to explain why.

Calling every preposition an "adverb" if it doesn't have an NP complement is the ancient tradition of stupid Latinophilia (notice the assumption that if they are called prepositions they have to be pre-positioned, in accord with the Latin etymology). Sensible grammarians have been trying to do away with it for two hundred years. Jespersen's The Philosophy of Grammar argued for the sensible view, but others preceded him, back to the middle 18th century.

And the idea that "the whole question of 'stranded prepositions' simply goes away" under the ancient stupid view is just a mind-bogglingly awful mistake. Even the proponents of the ancient stupid view agreed that in What can we cut it with? we have a preposition, not an adverb; so stranding remains. The question is whether you are going to deprecate it, and follow the long tradition established by John Dryden, or get in line with all the serious grammatical descriptions of the language, one hundred percent of which agree that stranding is fully grammatical—and much more normal than the rather pompous preposition fronting of With what can we cut it?.

It is clear that this commenter had not been reading Language Log for very long. One place to start reading about the history would be "Hot Dryden-on-Jonson action" by Mark Liberman. But a search on preposition stranding "Langage Log" would bring up a lot more.


  1. Snarky said,

    October 3, 2010 @ 7:34 am

    Yes, split infinitives and stranded prepositions are grammar myths. But I would have argued the second point differently (see In the sentence "Spell check won't help you when you have the wrong word to start with," the preposition is an integral part of the phrasal verb "to start with." So it's not acting as a preposition here, but as part of a verb.

    [No it isn't; your analysis is wrong. This sort of claim, about prepositions closely associated with verbs being "part of the verb", never works. Morphologically, it is hopeless (the preterite tense of throw up is threw up, not *throw upped); and syntactically too there is evidence against it in nearly every case. Take the examples on the page you cite above. You say that look into, meaning "investigate", has into as "part of the verb". But that will not work, because of sentences like The committee is looking, somewhat reluctantly, into the allegation that Professor Fastbuck may have fabricated some of his data (you can't have a two-word adverb phrase in the middle of a verb). You suggest that the in of the expression break in is part of the verb; but that leaves no explanation of sentences like The bars on the window prevent anyone from breaking in or out. The "phrasal verbs" of which you speak are neither verbs nor phrasal. They are sequences consisting of a verb and a preposition (or sometimes two prepositions, as in look up to meaning "respect"). As The Cambridge Grammar of the English Language carefully points out (chapter 4; see pages 273-274), it is impossible to take these sequences as constituents, because sometimes, despite the special meanings attached, they are not contiguous (e.g., in the idiom take [someone] to task). Even in the case of to start with, which is almost an unmodifiable idiom syntactically, it contributes nothing to say that with is part of the verb, which would imply that knowing the verb start was no help in understanding this phrase because it doesn't contain that verb. And incidentally (since a discussion of "particles" starts below) let me point out that with is certainly not a "particle": it doesn't show the freedom of position with respect to the direct object (as in look up something / look something up). —GKP]

  2. Nick Z said,

    October 3, 2010 @ 7:45 am

    Rod Gelatt's "correction" might be 'When you have the wrong word to start with, spell check won't help you' (note that Mr Gelatt objects to ending a sentence with a preposition, not a phrase). [I did him the favor of ignoring the usual error of thinking that a stranded preposition is always at the end of a sentence; he clearly means that stranded prepositions are errors, and he would take What are you looking at, asshole? to be just as culpable as What are you looking at?. So I was kindly trying to avoid the red herring of whether "end of sentence" describes the phenomenon correctly (it does not). But I see that no good deed goes unpunished. —GKP]
    Or it might be to compose a different sentence altogether. Changing the sentence to the clearly ungrammatical '*You can wipe that smile off your face with which to start' is not the only option, despite the mistaken premise of No Stranded Prepositions. [Sigh… The old "you could rephrase it completely" nonsense is surely the last resort of the idiot prescriptivist caught in a position of having needlessly banned the best way to express a particular thought. —GKP]

  3. aqilluqqaaq said,

    October 3, 2010 @ 8:32 am

    Ok, I’ll bite – in what sense is ‘with’ a preposition here?

  4. J. Martin said,

    October 3, 2010 @ 8:59 am

    @Snarky, @aquilluqaaq look here:

    “However, you might ask, doesn’t it still end with a preposition? Well, yes and no. It ends in a word that is classed as a preposition by The Cambridge Grammar, which takes what I consider the right view. But it’s a preposition that does not take an object. For that reason it is irrelevant. In fact the traditional view (which has a somewhat fetishistic attachment to the Latin meaning of pre-) refuses to call it a preposition because it is not before a noun phrase.” G.K.P.

  5. Joe said,

    October 3, 2010 @ 9:35 am

    I would be interested in an analysis of "to start with."I can see the relationship between "you can wipe that smile off your face to start with" and "you can start with (?) by (?) wiping that smile off your face." But there is an obvious difference in meaning between "spell check won't help you when you have the wrong word to start with" and "spell check won't help you when you have started with the wrong word." I imagine it has just become an idiom, but I haven't quite worked out the relationship between what I imagine to be a to-infinitival clause and the main one.

  6. George said,

    October 3, 2010 @ 10:01 am

    @Snarky and GKP: According to the "Oxford Concise Dictionary of Linguistics," a phrasal verb is, ". . . 2. Specifically of a unit in English which is formed from a verb with the addition of a preposition or adverb which can variously precede or follow an object."

    I am not sure that 'start with' passes this test. Example:

    1. She started the lecture with phrasal verbs.
    *2. She started with the lecture phrasal verbs.

    This might be contrasted with another example:

    3. She turned the light on.
    4. She turned on the light.

    [Quite right: to start with doesn't fit the definition of the misbegotten term "phrasal verb". And I never said it did, of course. In fact The Cambridge Grammar of the English Language doesn't use that term at all, on the grounds that it is misleading, badly defined, and unneeded. —GKP]

  7. Philip Spaelti said,

    October 3, 2010 @ 10:42 am

    @George: what you have is not the phrase "to start with". This phrase must be in the infinitive. In your example "with phrasal verbs" is a prepositional phrase, and in such cases there of course cannot be another intervening noun.

  8. Chris Buckey said,

    October 3, 2010 @ 10:46 am

    I hesitate to suggest this, but you should really consider writing your own style guide to shunt That Whose Name I Shall Not Speak into well-deserved oblivion.

    I would certainly pre-order a copy or twelve.

  9. Philip Spaelti said,

    October 3, 2010 @ 10:47 am

    Another striking thing about the phrase "to start with" is that it can be preceded by "right". "Right" is typically a test for prepositions but here it cannot directly precede "with". It must precede the whole phrase.

    [Not every preposition allows the modifier right. And certainly not necessarily in fixed-form idioms like this. —GKP]

  10. Eli said,

    October 3, 2010 @ 10:51 am


    Don't VP-topped constituents have to include their objects (since their are below them in the tree)? That would be an argument for verbish-ness of phrasal verbs. Maybe it counts against them being named "phrasal", but at the very least if the P is not "part" of the verb then it is obligatory or semantically necessary given a certain meaning.

    Although it has just struck me that you may be arguing against a word that sounds like a preposition being part of the V itself. (I guess it's a little hard to tell.)

  11. Bloix said,

    October 3, 2010 @ 10:58 am

    He doesn't want to generously point out. He generously wants to point out. Perhaps he wants to vigorously point out. Perhaps he generously wants to vigorously point out. But he doesn't want to generously point out.

    [The original sentence was apparently "If you are an exceptionally eagle-eyed reader who wants to generously point out more than one error per story, that's great…". You're saying that generously seems not to make sense as a modifier of point out. I think that's correct; the writer did not, apparently, mean "to point out in a generous way". So there actually is an error here. And as so often happens, the lunkhead tradition of prescriptivist poppycock fails to nail it. The error is not the placing of an adverb between to and the head verb of the infinitival complement that it introduces; the error is putting the wrong adverb beside the verb. Mr Gelatt actually had an error in his net, but he didn't recognize it, and he let it slip away over the side of the boat. You know, if prescriptivists were as bad at getting sex right as they are at getting grammar right, they would die out within a generation. —GKP]

  12. Joe said,

    October 3, 2010 @ 11:01 am

    We are not talking about garden variety "start" since "with" can be stranded: who should we start with? It's the phrase "to start with" that is at issue (and I agree that with is a preposition). If I wasn't clear before, what I find odd in "when you have the wrong word to start with" is how it differs from, say, "if you have the wrong tool to do the job," where "tool" is the understood subject of the clause. Of course, it must have to do with the preposition, since "the wrong tool to do the job with" patterns the same way as "to start with". Anyway, just thinking aloud and more thinking about the historical development than anything else.

  13. Jonathan Mayhew said,

    October 3, 2010 @ 11:45 am

    An idiomatic phrase, a chunk of language that native speakers use all the time, cannot be ungrammatical.* You can't say "to start with" in this particular sense in any other way. Putting the preposition there is obligatory.

    *Leaving aside idioms viewed as dialectal, and ungrammatical in other dialects.

  14. I.D. Mercer said,

    October 3, 2010 @ 12:47 pm

    I honestly wonder if phobics of preposition-at-the-end would "correct" the sentence

    What was the name of the space shuttle Christa McAuliffe was on?


    What was the name of the space shuttle on which Christa McAuliffe was?

  15. figleaf said,

    October 3, 2010 @ 1:29 pm

    Hmm… Based on the context and his rhetoric ("…the invitation for us to become grammar police") I wouldn't jump to the conclusion that Rod Gelatt was being fusty. I read it as something closer to "those who live in glass houses shouldn't throw the Chicago Manual of Style."

    Even without as generous a reading I think it might also be significant that Gelatt was a professor of journalism rather than a linguistics professor or even an English professor. Journalism, it just occurred to me, is artificially stylized in the manner of many forms of poetry… limericks, for instance. With that understanding one could say "it's not journalism if you split an infinitive," but only in the same way one could say it's not iambic pentameter if there aren't five pairs of syllables per line.

    We would all correctly laugh and point if someone claimed language must follow the rules of galloping meter and rhyme to be correct. Yet we're repeatedly expected to dogtrot inside the conventions of journalistic style. The difference, I think, is that unlike poets, journalists have access to ink by the barrel… and enough other things on their mind to mistake their conventions (which might be logical for hierarchically structured, space-constrained, and deadline-conscious editorial production) for "all language."

    Could that be why so many grammar police seem to be current or retired journalists and their editors?


  16. The Ridger said,

    October 3, 2010 @ 1:39 pm

    Genuinely curious here: What's wrong with calling things like this "particles"? I believe they're related to German's separable prefixes, only permanently separated (einsteigen – ich steige ein; get in – I get in) but I can see that calling them "permanently separated prefixes" is a bit silly. But insisting on preposition seems stubbornly unhelpful. My students seem content with "particle", and they stop looking for objects.

    [If you like, use "particle" as a one-word substitute for "light preposition that doesn't have its own object and is allowed to occur before as well as after the object of its governing verb". But the reason for not treating Particle as a separate part of speech is that almost all the so-called particles are identical in both form and meaning with an uncontroversial preposition: in, on, up, down, round, through, to… there are lots. And even words like back, which occur as particles but never have their own object, pass tests for prepositionhood like taking right as modifier (We gave the money right back; We had to give right back all the money that they lent us). It's not that particles don't have special properties. It's like auxiliaries and verbs: the auxiliaries are a distinguished subset of the verbs with special syntactic properties, and the particles are a distinguished subset of the prepositions with special syntactic properties. However, returning to our topic, with definitely isn't one. Particles can come either before or after the direct object of the verb. No sign of that for with, here or elsewhere. —GKP]

  17. NW said,

    October 3, 2010 @ 1:42 pm

    I suppose journalists are a bit prone to talking codswallop about language. But then there are three grades of people: ordinary people; journalists; and linguists. Grades 2 and 3 habitually think about language. Only grade 3, linguists, have to get it right or they're not doing their job; but journalists can just write like ordinary people and still do their job properly. So if they learn wrong grammar in a wrong belief that it matters that way, this doesn't lose them their jobs. And as they have a public platform, their irrelevant beliefs about language are magnified: like those people Truss and Heffer.

  18. NW said,

    October 3, 2010 @ 1:47 pm

    @ The Ridger

    Calling prepositions in some circumstances particles is multiplying entities unnecessarily, but who cares how many classes of words there are? So that's just stamp-collecting. Call them particles if it makes things easier.

    The real problem is that particles are obviously of different parts of speech, almost all of them prepositions, but there are a few others: verbs (let go the rope, let the rope go) and adjectives (let loose the tiger, let the tiger loose, throw open the window, throw the window open). Particle is a perfectly good term for a word that can optionally appear before or after an object. But they all fall neatly into the traditional categories, and while some are adjectives, most are obviously intransitive prepositions in addition to being particles.

  19. Joe (same as one above) said,

    October 3, 2010 @ 2:21 pm

    Sorry, I'm still a bit stuck trying to figure out the analysis of "to start with" in this context but I know where I went wrong. Although it is rare, it is possible to front the preposition in some contexts of "to start with": "If someone had never read Lewis before . . . there would be no better book with which to start" (BNC). Here, however, the infinitival is an integrated relative clause (as it is in my earlier examples of "the wrong tool to do the job with"). In "have the wrong word to start with," "to start with" isn't an integrated relative clause, so it must be adjunct (and then there is no surprise why "with" must be stranded). In some contexts, the adjunct means something like "from the start" (the context the professor commented upon), in others, it means, "for a start" (as in GKP's "wipe that smile off your face to start with" ). I might be completely wrong here, but my sense is that in the professor's example (i.e., the "from the start" interpretation), "to start with" is VP oriented, but in GKP's example (the "for a start" interpretation), it is clause oriented.

    The OED gives this as the earliest reference to this sense of "to start with" as

    1866 MRS. OLIPHANT Agnes I. xxii. 280 Her mind..was of a much higher order than his to start with.

    (and I think this could be ambiguois between either a VP orientation or a clause orientation).

    I'm still a bit curious about how this developed, not so much in the semantics (which is obvious) but the syntax. But I am sorry for commenting so much on it. And, of course, I'm sure this was GKP's point the entire time, and my brain has been too fogged over to grasp it.

  20. Anya Lunden said,

    October 3, 2010 @ 3:35 pm

    I am with you in the response to Snarky in which you pointed out that the analysis of object-less prepositions as verb particles of some sort doesn't work. But I wonder what you think of the case I know of where the whole unit does seem to act like a verb: Some such sequences can take the suffix "-able" which attaches to verbs to form adjectives, e.g. fix-up-able. Whereas *upable is hopeless (unless it's the verb "up").

  21. George said,

    October 3, 2010 @ 3:37 pm

    @NW: "I'm still a bit curious about how this developed, not so much in the semantics (which is obvious) but the syntax."

    Could these things (paritcles, prepositions, light prepositions, whatever) develop through ellipsis?
    1. Start X with Y,
    2. Start X with (Y),
    3. Start (X) with (Y)
    4. Finally becoming a fixed VP: Start with.

  22. Coby Lubliner said,

    October 3, 2010 @ 3:46 pm

    GKP: round is not a proposition in American English. [It is found, but I agree that around is more frequent. —GKP]
    Also, I don't get the idea that modifiability by right is the litmus test of a preposition. Would you consider proper in a right proper gentleman to be a preposition? [No. But that is archaic or regional; nobody says right proper in contemporary speech. You're quoting from Dickens or something. —GKP]How about away in right away? I also feel that right in an expression like I'll be right with you modifies be, not with. [Right has various uses; but it doesn't occur as a pre-head modifier inside a phrase in the contemporary language unless the head of that phrase is a preposition. —GKP]

    My problem with this whole discussion is what seems to me the basic premise of CGEL, which is to treat grammar as a quasi-scientific discipline, with strict definitions and all that apparatus, rather than what it historically has been, a methodology for a learning a written language. All the grammars of Antiquity and the Middle Ages were meant as aids in the learning of "dead" literary languages such as Sanskrit, Homeric Greek, classical Latin, Koranic Arabic and Biblical Hebrew. The first modern grammar, that of Castilian by Nebrija, was meant to (1) "fix" the standard so that future readers would understand the chronicles of Queen Isabella, (2) teach the concepts of grammar in order to facilitate the study of Latin. To these two aims, Bishop Hernando de Talavera added a third: to help the kingdom's non-Castilian-speaking subjects to learn the language. Now, I find that the CGEL approach obscures rather than clarifies grammatical questions that might matter to learners of English. [I can't answer all of this. But let me just say that CGEL is not a book for non-English-speakers to learn English from; it's a book to enable specialists to understand the structure better so that they might be better equipped to design courses, and books to teach the language from. —GKP]

    Finally, with regard to phrasal verbs: it seems to me (as someone who knew German before learning English) that the concept is influenced by its analogue in other Germanic languages. To build on The Ridger's comment above: in German ich gehe aus corresponds precisely, word for word (even, in this case, etymon for etymon) to I go out. But in the infinitive, participles and gerund, the "particle" (preposition or adverb, as the case may be) merges as a prefix with the basic verb: aus(zu)gehen, ausgehend, ausgegangen, and this is why ausgehen is regarded as a verb on its own. In the finite forms, on the other hand, all kinds of stuff can go between gehe and aus, e.g. ich gehe heute Abend zu Fuß aus. [Fair enough; but I confess I can't see the relevance of any of this. The topic, originally, was whether a preposition is stranded in to begin with, wasn't it? —GKP]

  23. George said,

    October 3, 2010 @ 4:01 pm

    @Coby: "My problem with this whole discussion is what seems to me the basic premise of CGEL, which is to treat grammar as a quasi-scientific discipline, with strict definitions and all that apparatus, rather than what it historically has been, a methodology for a learning a written language."

    Wow! This ought to stir up some conversation. Maybe, we should start with defining grammar. Linguists use the term to mean the rules or structure of a language in the way it is spoken, not the way someone thinks it should be spoken.

  24. Qov said,

    October 3, 2010 @ 4:27 pm

    Prescriptive rules like "don't start a sentence with a conjunction" and "don't talk to strangers" are useful guides for people too inexperienced to make their own judgments about forming sentences and social relationships, but it's hard to make your way in the world if you abide by them, and censure others who don't, all the days of your life.

  25. áine ní dhonnchadha said,

    October 3, 2010 @ 4:50 pm

    Topic snark: this is why I invented the term "overhung" to mock pedant friends of mine at university. Nothing worse for them to hear when they are all suffering from too much alcohol the night before than my chipper voice castigating them for drinking so much it left them "overhung". They'd wince, then they'd try to figure out how to say "hung over" without a hanging preposition and they'd wince again.

  26. Jonathan Mayhew said,

    October 3, 2010 @ 5:23 pm

    I've been mulling this over a bit more. In my view, "to start with" is an idiomatic phrase meaning "in the first place." It is different from the use of "start with" in sentences like "Let's start with the easy cases." You could say, "Let's do the easy cases to start with," but not "*Let's do with the easy cases to start" or "*I don't have a million dollars with which to start." The first is ungrammatical and the second means something else. In other words, in the idiomatic use of that phrase, preposition "stranding" is obligatory because the preposition with doesn't go with (anything else).

    [This seems like sound observation. I never mentioned particles originally; I merely said that with here is a preposition, and in this idiom stranding is your only option. We seem to be in full agreement. —GKP]

  27. Steve Harris said,

    October 3, 2010 @ 10:56 pm

    To start with, I'll consider whether "with" has any object, anywhere, in "to start with".

    To start with a simpler idea, I'd have had to phrase that sentence differently, one which illustrated a specific object for "with". Instead, I decided to go into the deep end first.

    So what did I start with, in that first sentence? I started with the notion that "I'll consider whether 'with' has any object, anywhere, in 'to start with'". But the word "notion" nowhere occurs in that sentence; indeed, I might just as well have said, "To begin". This supports the analysis that " to start with" is just an idiomatic synonym for "to begin". But I like the idea of seeing this as having come from something like this:

    To start with the first thing I'm thinking of, I will begin by saying XYZ

    (where XYZ is an independent clause, such as "I'll consider whether …")

    Obviously, it's much short simply to say, "To start with, XYZ"; but I wonder why the "with" has been retained? Why didn't it ground down to "To start, XYZ"?

    It sounds to me like a cooking-recipe approach to self-described discourse: To start with my first topic, XYZ. To mix in my second subject, PQR. And to top things off with my last thought, ABC.

    To finish off, I'll just stop.

  28. John Cowan said,

    October 3, 2010 @ 11:39 pm

    "Look at" seems to have a few verbal properties: in particular, it can be used in the imperative, Lookit! (though not in writing).

  29. WindowlessMonad said,

    October 4, 2010 @ 4:22 am

    I've been waiting for someone more edudite than me to take issue with 'grammar as a quasi-scientific discipline' …Quasi!?!? As opposed to treating it as a matter of intelligent design, perhaps?

    @Coby Lubliner: There's nothing quasi about grammar as science. Its propositions are well-defined, testable, and — if you care to do look at the emprirical evidence (as working grammarians, unlike their critics, actually do) — refutable. Stand still a moment, would you, while I beat you about the head with my very hard-back copy of Logik der Forschung.

  30. George said,

    October 4, 2010 @ 7:18 am

    @Steve Harris: "To start with, I'll consider whether "with" has any object, anywhere, in "to start with".

    Good question. How about –

    "To start with (the issue of the object of the preposition 'with'), I'll consider whether 'with' has any object, anywhere."


    "To start with (the issue of misspellings), spell check won't help you when you have the wrong word."

  31. Rodger C said,

    October 4, 2010 @ 8:58 am

    Here's my conjecture as to the genesis of this phrase:

    "To start with the toves, they're slithy."
    "To start with … [pause for emphasis] the toves, they're slithy."
    "To start with, the toves are slithy."
    "The toves are slithy, to start with."

  32. George said,

    October 4, 2010 @ 9:51 am

    Is preposition stranding found in other languages?

    I can think of no examples from Arabic except where there is clear conversational ellipsis.

    (GKP: I swear that I and all the members of my family accept preposition stranding as acceptable English grammar.)

  33. John Ward said,

    October 4, 2010 @ 10:04 am

    I realize this is mostly a matter of definition, and admittedly, I'm a sucker for etymology, but I'm one of those who feels a little strange saying that "with" in "to start with" is a preposition.

    You might think we're too hung up on the Latinate meaning of "preposition," but arguably, it's others who are too unwilling to give up the word "preposition."

    The thing is, why does "preposition" (or "postposition") have to be a distinct part of speech? It sounds to me like "preposition" describes a syntactic relationship between at least two words. If people weren't so attached to the name, they'd just call them all "particles" (or some better name), and talk about "preposition" or "postposition" as the construction where they head up something else on the left or on the right.

    So what I'd like to say is that "with" in "to start with" is a particle, and particles can often be prepositions in English, but, in this case, it isn't a preposition of any other word.

    Or am I missing something?

    [Well, you are certainly missing the several times above I have stated clearly that with is NOT a particle under the familiar definition: it does not have freedom of position with respect to the direct object of the governing verb. I don't know how many more times I am going to have to say this; the task seems endless. —GKP]

  34. dwmacg said,

    October 4, 2010 @ 10:27 am

    @John Ward,

    Would you say that "at" is a preposition in (b), but not in (a)? And if so, what does that get you?

    (a) What are you looking at?
    (b) I'm looking at you.

  35. Greg Morrow said,

    October 4, 2010 @ 10:39 am

    "Spell check won't help you when you have the wrong word with which to start" is grammatical in my version of English, albeit formal, stuffy, and only possible in the written language. [And it has a different meaning. —GKP]

    In other words, I agree with Jonathan Mayhew; there is an idiom "to start with", with complement-less and obligatorily stranded "with", which is distinct from the verb "start" with a complement prepositional phrase "with X", where the prepositional phrase is capable of being split and moved around like other such.

  36. Cy said,

    October 4, 2010 @ 11:05 am

    I don't know if you're still reading these (since the next posting seems to suggest it's a little tiring), but I've always wondered something. I'm not stating an opinion here, I'm actually asking this as a non-rhetorical question – if some linguist was documenting a spoken-only, non-orthographic language that also happened to be English, and he came upon these "phrasal verbs," where sometimes they're split down the middle, sometimes they have adverbial phrases wedged in between the two verb parts, strange direct-object behavior – would any linguist be tempted to classify this as infixing of some sort? Would anyone ever actually describe them as verbs that are separable, and if not, why? I can't claim that I know of any languages that allow phrases to divide their verbs, but is there a nice, concise reason why this wouldn't be so? Or am I just attaching too much meaning to traditional grammatical classifications? I mentioned orthography to suggest that the field researcher may not have the insight into the fact that these "particles" are "the same" as the prepositions in common use – so he would only hear /drap.ɪɾ.af/ or / and not see it as 'drop it off', but rather a verb with a direct object in the middle (in this instance).

  37. Ellen said,

    October 4, 2010 @ 11:59 am

    While that sentence (in Greg's post just above) is grammatical for me, it doesn't mean the same thing for me as the original sentence.

  38. Ray Dillinger said,

    October 4, 2010 @ 12:46 pm

    Please do not disparage prescriptive texts as a whole.

    When prescriptive texts are accurate, they are valuable to students seeking to master a prestige dialect. I agree with you that disparaging the split infinitive and the trailing preposition are inaccurate statements about English. But accurate prescriptive texts have great value for people like myself, whose careers and futures did depend on mastery of a prestige or predominant dialect. I should not have been taken seriously by my peers had I arrived here still speaking the patois of my youth. The pedagogic approach that helped me to overcome that barrier was unashamedly prescriptive, and I am grateful for it.

    Prescriptive texts don't help us much in dealing with English as it actually occurs on webpages, blogs, journalistic articles, and the other corpora that we're working with now. What we really need for that is descriptive texts. But the descriptive texts available are utterly inadequate for our needs.

    Our company writes and maintains software that attempts to render free English text into a reasonably coherent knowledge base that can be queried with database tools. The problem is that descriptive texts written by linguists are generally aimed at humans, and ignore most or all of the minor distinctions and subclasses that are really, truly, necessary in automatically parsing English and getting sensible results.

    We had to classify "To Start With" as a "Split-3 Postpositional Delimiter" — not because that's a category recognized by any English Textbook, but because "Split-3" describes to our software which class of separability rules it follows (and what form of each word is required in which separation scenarios) and "Postpositional Delimiter" was linguistic mumbo-jumbo that seems to remind us of its function without clashing with any of the other linguistic mumbo-jumbo we've had to use for other classes of word and phrase.

    I would pay a very large sum of money for a descriptive grammar of English that were even 95% complete with respect to our needs. It simply doesn't exist, and we're investing dozens or scores of man-years to build it from scratch – sadly without help from academic descriptive linguists who seem too focused on pedagogic needs to even consider industrial needs.

    [First: I'm not opposed to prescriptivism as such, but to stupid prescriptivism. Second, while I have done very little that could be called pedagogically oriented, I did work for several years at Hewlett-Packard Laboratories on a natural language processing project, so I know the kind of problem you're talking about. —GKP]

  39. The Ridger said,

    October 4, 2010 @ 1:37 pm

    So, "intransitive preposition" is the name for these? I suppose that works, though redefining a part of speech seems no better than renaming it.

    I shall endeavor to convince my students that prepositions don't need objects, though I was taught they did (and that's why the PP is the only Phrase that MUST have two words in it). Times change, and so do analyses.

  40. Xmun said,

    October 4, 2010 @ 2:50 pm

    Re: "split infinitives have always been grammatically correct in English"

    I know this is often said, but I've yet to find one in the first three volumes (all I have yet read) of Gibbon's History of the Decline and Fall of the Roman Empire. If I find one in the remaining three volumes I'll let you all know. And if anyone can show me I missed one or more of them in the first three volumes, please let me know.

    [Some authors have deliberately avoided them. Gibbon may be one. You're free to avoid them if you wish. Just don't call them errors when other people put the adverb where they want to put it. —GKP]

  41. Faldone said,

    October 4, 2010 @ 2:55 pm

    I dunno. In my very non-professional opinion the examples I've seen to establish the existence of intransitive prepositions are just transitive prepositions with the object understood. But what do I know? I'm just this guy.

  42. Jerry Friedman said,

    October 4, 2010 @ 2:56 pm

    @Xmun: I wonder about that too. Split infinitives were extremely rare in the 16th and 17th centuries. There's only one definite one in Shakespeare, for example, and as it's inverted word order to get the rhyme and meter to come out, I'm not sure it's evidence on grammaticality. With something like the split infinitive, which became popular, then almost disappeared, then became popular again, what's the criterion for grammaticality?

    (I and all members of my family consider split infinitives 100% grammatical during our lifetimes, which I trust will continue, as long as I don't reveal that I like Lowth's self-referential comment on stranded prepositions.)

  43. tudza said,

    October 4, 2010 @ 2:59 pm

    Since you have closed the comments for your most recent posting, where you complain about people posting the same comments that have been seen in the past, I will post my comment to that here.

    I know all about the pseudo-Chuchill saying but I am likely to use it on occasion anyway, not because I believe I am quoting Winston Churchill, but because I find it terribly amusing by itself. You should reserve your door-to-door vendetta for those that actually claim Churchill as the source.

  44. Greg Morrow said,

    October 4, 2010 @ 3:09 pm

    GKP's insertion in my post above is not something I agree with entirely*. That is, in that context, I believe that I am perceiving "to start with" as a superposition of analysis. On one side, you have the idiom which means "firstly", or "as a preliminary matter", in which "with" cannot be separated and cannot take a complement. On the other side, you have the verb "start" + "with X" formation, meaning more or less "to begin by first considering X".

    *This is hubris, for Professor Pullum thinks about this stuff for a living, and I don't. Thus call the Kindly Ones and thus am I damned.

    My reason for asserting this superposition is 1. "word" is nearby and can serve sensibly as X in the latter interpretation; 2. when the former interpretation is suppressed, e.g. when the desired register deprecates stranded prepositions, I actually can and do readily switch to the latter.

    My understanding is that this sort of simultaneous meaning/analysis overlap is one of the engines of language change. I'd presume it's exactly how idiomatic "to start with" arose to start with. E.g. "We'll start with X -> we'll choose X to start with -> we'll do X to start with".

  45. George said,

    October 4, 2010 @ 3:13 pm

    Radford ("Transformational Grammar," Cambridge University Press, 1988) calls these prepositions and a constituent of a 'phrasal verb.' He does an exhaustive analysis to show they are a constituent of the verb.

    Tallerman ("Understanding Syntax," Arnold Publishers, 1998) has a similar analysis and also uses the term 'phrasal verb.'

    [Yes, lots of syntax texts give an analysis where look up the number has a complex verb [V [V look ] [P up ] ]. That doesn't mean they're right. —GKP]

  46. Arvind said,

    October 4, 2010 @ 3:38 pm

    Also on the subject of the pseudo-Churchill saying, here's the Hansard transcript of Glenvil Hall's telling of the Churchill anecdote:

    I find it interesting that Glenvil Hall doesn't expressly say that the note was in response to a grammatical correction. This just might suggest that the anecdote was well known at the time.

  47. Picky said,

    October 4, 2010 @ 3:48 pm

    Incidentally, re the "Churchill" quote, the Glenvil Hall passage Ben Zimmer mentioned was this (according to Hansard 18 March 1948):

    I agree with him [Edward Keeling MP], however, that some words are frequently used in the wrong sense. One of the words to which he drew our attention, namely "evacuate," is one of those words. You cannot strictly "evacuate" a person. You "evacuate" a place; and yet today, almost everybody, and, I freely admit it, Government Departments, use the word "evacuate" when speaking of a person. Words such as that, I agree with him, ought not to be used, for English is a beautiful language, and those who work and speak for the Government, of whatever complexion that Government may be, should try to use English in its proper sense. Nevertheless, frequently we do get phrases and sentences which may not be strictly grammatical, but which do convey exactly what the person means. Of course, the right hon. Member for Woodford (Mr. Churchill) is a past master in using words in their proper sequence and to mean exactly what he wishes to convey. But even he at times will speak ungrammatically, and yet in no uncertain way convey what he wishes to make plain. There is on the record in Whitehall a paper with a comment of his scrawled in the margin, "This is all nonsense up with which I will not put," That might be phrased differently, but it could not be phrased more forcibly and the person who saw it could be under no misapprehension as to what the right hon. Gentleman wished to convey.

  48. John Ward said,

    October 4, 2010 @ 5:25 pm


    (a) What are you looking at?
    (b) I'm looking at you.

    I would call "at" a particle in both sentences. But I would describe its position in (b) as a preposition of "you," and I would describe its position in (a) as a discontinuous postposition (or however you prefer) of "what."

    Clarity. The terminology on this whole point is pretty confused already. In leftheaded languages, we have a word class called prepositions, in rightheaded languages, we have a word class called postpositions, and then there's languages with mixed headedness like Chinese and German where the same particles can be used both prepositionally and postpositionally with slightly different meanings. If this same part of speech can also be used intransitively, that's just one more argument for finally avoiding the term "preposition," as a part of speech, anyway, just as the CGEL has already avoids "conjunction."

    In my own experience, books on Korean are rather confused, I've noticed, with the same phenomena being called "postpositions," "case endings," "particles," and even "prepositions" in different books. There's also the problem of people saying "preposition" when they mean "locational phrase."

  49. Nicholas Lawrence said,

    October 4, 2010 @ 6:09 pm

    In standard Latin pax tecum, is cum a preposition or a particle?

  50. J. Goard said,

    October 4, 2010 @ 8:52 pm

    @John Ward:

    In my experience with other Korean learners (mostly non-linguists), what they are most likely to call "postpositions" are not particles per se (the topic or nominative particle a "postposition", really??), but rather translation equivalents of English prepositions, which typically means a locational noun with a locative particle, e.g.:

    가게 뒤-에
    store rear-LOC
    'behind the store'

    상자 안-에
    box interior-LOC
    'in the box'

    For me, it's still somewhat hard to really grok the nouniness of these locational noun compounds, for example to use them in the nominative (where English would use an "expletive" it or there):

    교실 안-이 조용해-요
    classroom interior-NOM quiet-POL
    'It's quiet in(side) the classroom.'
    (Or, if you prefer Google's translation: 'Ophthamology Department of Silence')

  51. John Walden said,

    October 5, 2010 @ 3:17 am

    Though stupid Latinophilia, the particle as preposition vs particle as adverb is how ESL teachers explain, to themselves if not to students, the word order when the direct object is a pronoun: 'Look for it' is not 'Look it for' because 'for' is a preposition, most if not all the time, like 'into' and 'of' (for the purists these are not PVs). By the same token, 'I'm trying to get this difficulty through to you' is not 'I'm trying to get through this difficulty to you'; the first 'through' is an adverb. It's not 'through' anything in particular while the second is.

    In this analysis 'look up to' is a three parter consisting of verb-adverb-preposition. According to all this, that 'up' is not a preposition: 'up' what?

    Pretty basic stuff and mostly terminological bickering. Where it's especially moot is 'Put your hat on' because suddenly it's a preposition if we add 'your head' but apparently not if we don't, although we know damn well that it's 'your head' all along.

    'Preposition' is not a helpful word because of its etymology. Peeople who think that prepositions should at least sometimes be found in a pre position have to jump through flaming hoops to explain what part of speech 'ago' is.

  52. Svlad Cjelli said,

    October 5, 2010 @ 9:19 am

    Don't "to start" and "for a start" work? Is "with" essential?

  53. Faldone said,

    October 5, 2010 @ 9:45 am

    John Walden: Where it's especially moot is 'Put your hat on' because suddenly it's a preposition if we add 'your head' but apparently not if we don't, although we know damn well that it's 'your head' all along.

    This is exactly the sort of understood object that I was talking about above. It's like enjoy. Enjoy doesn't suddenly become an intransitive verb when the waitress says, "Enjoy" after laying a plate of food on the table in front of you.

  54. Adam said,

    October 5, 2010 @ 11:46 am

    I don't have a copy of the CGEL handy, but I'm curious as to whether the classification of "up" as a preposition in both of these is based on semantics as well as syntax.

    (1) He climbed up.
    (2) He climbed up the ladder.

    While "up" and "up the ladder" both modify "climbed" in semantically similar ways, to me (I was taught a bit of HPSG and similar formalisms) the two kinds of "up" look syntactically quite different. In (1), "up" takes no arguments and is a modifier with the target "climbed". In (2), "up" takes an NP argument (I know, some prepositions take an NP, some take a clause, and some take either) and the resulting phrase modifies "climbed".

    I also don't get the argument that "it is impossible to take these sequences as constituents, because sometimes … they are not contiguous". Parsing with discontinuous constituents has been around in NLP for a while, and it seems to be a good way to handle a PP consisting of a stranded preposition and its argument.

  55. Jerry Friedman said,

    October 5, 2010 @ 8:12 pm

    @Svlad Cjelli: "Spell check won't help you when you have the wrong word to start with" becomes ungrammatical (in my idea of American English) if you replace to start with with either to start or for a start. The sense of the original is something like when you started with the wrong word.

  56. George said,

    October 5, 2010 @ 9:03 pm

    @Adam: I am not sure that 'climbed up' is a phrasal verb or 'up' is an intransitive preposition.

    I think your example (1) has an implied object that is not explicitly stated because the reader or interlocutor know it from the context. Just stated alone, "He climbed up," prompts the question, "climbed up what?"

  57. Picky said,

    October 6, 2010 @ 3:36 am

    @Faldone: but in Put your trousers on, Put your bra on, Put your knickers on, Put your tie on – is there some identified, named, elided part of the body, or are we just using a modern equivalent of "don"?

  58. Faldone said,

    October 6, 2010 @ 7:08 am

    @Picky: I suppose if you were to put your bra on your head as often as you would on a more obvious part of your anatomy then it might be necessary to explicitly state where it is you're putting it. And, if the AHD is to be trusted, don is a contraction of do on so I'm not sure I see a difference between on as a transitive preposition with an implied object in the one case and the other.

  59. Picky said,

    October 6, 2010 @ 7:45 am

    @Faldone: I think you're ducking it. Put your trousers on … your legs ? Put your bra on … your breasts? Put your knickers on … your backside? Put your tie on … your … dunno. They don't work, do they? There isn't a missing noun.

  60. George said,

    October 6, 2010 @ 8:13 am

    @Picky and Faldone: FWIW, "put on" passes one of the tests of phrasal verbs according to Tallerman ("Understanding Syntax," 1998), "The preposition can alternately follow the direct object NP."

    1. She put on her hat.
    2. She put her hat on.

  61. Faldone said,

    October 6, 2010 @ 9:59 am

    So, you tell your two-year old to put his hat on and he puts it on his feet. You're not going to say, "On your head!"? This is part of the problem faced by those who think about this a lot. We can't have phrasal verbs because the boundary between them and simple verb-preposition structure is too fuzzy. I maintain that there are phrasal verbs where the particle is free to wander where it may and others where the particle is tightly bound to the verb. My criterion for telling a phrasal verb is whether the meaning of the verb and the meaning of the word that might be a preposition have been changed drastically in meaning.

    I ran into the burning building.

    I was probably actually running and I ended up inside the building. Verb/preposition.

    I ran into an old friend.

    I may well have been sitting on a park bench and I certainly did not end up inside my friend. Phrasal verb.

  62. Ray Dillinger said,

    October 6, 2010 @ 1:06 pm

    But the meaning of "run into" changes again if the person involved was driving a car at the time. In that case when he says

    I ran into the burning building.
    I ran into an old friend.

    He was not actually running, and he means that the car nominally under his control collided with the building or the old friend. Here "run into" is phrasal verb in either case.

  63. Andy Lee said,

    August 29, 2011 @ 12:57 pm

    Seems to me the idiom "to start with" can be read two ways here:

    1. "You can start with the action of wiping that smile off your face", which we'd normally say more simply as "You can start by wiping that smile off your face". There is no noun in the original sentence that represents the action of wiping, which is why it can't be "fixed" by the usual "with which" substitution.

    2. "I hereby start with the suggestion that you wipe that smile off your face (and I have further suggestions that I have not yet stated)." In this reading, the sentence itself is the object of "with". "To start with" is like "To be blunt…" or "Not to beat a dead horse by commenting on this article ten months later…".

RSS feed for comments on this post