> When we say, "It's raining," to what does "it" refer?

Let me guess. Some teacher in grade school taught you that all pronouns must have an antecedent, and you can't figure out how there can be an antecedent in It's raining.

Well, you're right. You can't find an antecedent in It's raining. What Miss Fidditch should have told you is that referential pronouns have to have an antecedent. But not all pronoun usage is referential. Pronouns are not always atomic, meaningful words like book and keep; quite often they're ionized for use as pieces of grammar, like the -er or the -s in bookkeepers.

All the common contractions with subject it are really grammar markings:

             it's  [meaning 'it is']     Progressive or Passive
                   [meaning 'it has']    Perfect
             it'd  [meaning 'it would']  Conditional
                   [meaning 'it had']    Pluperfect
             it'll [meaning 'it will']   Future

And those are only the ones with funny spellings; there are plenty of other special-purpose its that we don't bother to spell, and therefore don't pay any attention to, like the little chirp that constitutes the second syllable of Check it out in ordinary speech.

English is the kind of language (an "analytic" or "positional" language) that has most of its grammar concerned with word order, which word to use, prepositions and pronouns and "little words" and how they're organized. In a word, Syntax.

Languages farther toward the "synthetic" end of the analytic-synthetic axis,

(e.g, German, Spanish, Latin, Homeric Greek, Sanskrit, Navaho, Lushootseed, and Inuit,
to list a few synthetic languages, in increasing order of syntheticness)
tend to use more and more affixes and other form-changing kinds of grammar (in a word, Morphology), and syntax is less and less important in them. Morphology and Syntax are, respectively, the internal and the external economy of words, and together they make up Grammar.

OK, so what about It's raining? It's a kind of construction called a Dummy it. That is, the it has no meaning whatsoever (you're far from the first to be puzzled by it) and is used strictly as a placeholder, like the dummy hand in bridge, or the zero on 101. Why do anything that bizarre? Well, see, English Syntax has this Rule that says -- in ponderous and self-enforcing tones -- Thou Shalt Have A Subject In Every Finite Sentence. And thou must, indeed.

Ordinarily, this Subject is some noun or referential pronoun that is the salient agent or experiencer or patient of whatever the verbal predicate refers to:

    the writer in He wrote it
    the hearer in He heard it
    the hurter in He hurt it.

This falls down when dealing with some predicates. Rain is a superb example. Rain is a thing and an event, both nouny and verby. In using it as a verb, we must endow it with a Subject. But it's already its own subject, self-activating.

Solution: a Dummy Subject, obeying the letter but not the spirit of The Rule. One of several: this particular construction is called Weather it, to distinguish it from Distance it in

It's a long way to Schenectady
or Extraposition it in
It's important to me that she be present.
which is related by a syntactic rule called Extraposition to
That she be present is important to me.
And there are other Dummies as well, like the there in
There is a unicorn in the garden.
or the it in
Why don't we do it in the road?
Dummy words, placeholders, are fairly common in analytical languages.

These are special constructions, of which there are many thousands in English, and in every other language. This kind of information generally comes as a surprise, though, to many people whose grammar schooling took place in the U.S. The educational approach to English grammar that we appear to be stuck with in American English, at least, has been heavily influenced by Latin, where virtually all pronouns were referential, and where there was lots more morphology, and therefore a word-based grammatical tradition was reasonable. In English, though, you have to use a construction-based grammar to get decent descriptions and explanations.

Since almost nobody studies Latin any more (and those who become our teachers, and those who train them, are no more likely than others), what's left in the schools is a catechism of shibboleths and jargon that nobody claims to understand. And nobody does. Perhaps in another century or two, the school system will accelerate to match the glacial pace of language change, and children will actually be taught English grammar instead of picking it up on the streets.

But I'm not holding my breath.

-- followup - 2

>>OK, so what about "It's raining"?  It's a kind of construction
>>called a Dummy 'it'.  That is, the 'it' has no meaning whatsoever

>So, how about

> " 's raining"?  ("Is raining": all the words really needed to express
>the meaning of this sentence)?

'S raining /sre'nIng/ is one way of many that we can pronounce It's raining. The beginning of a sentence, especially if it's predictable and meaning-free, is a frequent place for sound loss in pronunciation. Others of this type include:

    Never see that again.           [I'll deleted]
    Daughter's on the phone, Bob.   [Your deleted]
    Never been there.               [I've deleted]
or, for truly egregious examples, the phrases MacDonald's employees use to mimic saying Can I help you?: [kyE~wpy@], or Is there anything else?: [zE~i0Ews].

By the way, you're probably not right to identify 's raining with Is raining. That's an /s/, not a /z/; that is, you wouldn't say /zre'nIng/. But is has a /z/, not an /s/, and it shows up in fast speech rules when it really means is. For instance, if you meant to say He's reading and you were in a hurry, you'd say /zri'dIng/, not /sri'dIng/. So the contraction is just that -- a shortening of a predictable sentence part. What else is new?

--- 3
>> By the way, you're probably not right to identify " 's raining" with
>> "Is raining".  That's an /s/, not a /z/; that is, you wouldn't say
>> /zre'nIng/.

> Now, who was it who wrote the following just the other day?

>JL> English speakers are still getting used to calling it "Sri Lanka"
>JL> instead of "Ceylon".  Anyway, /sr-/ is not an acceptable English
>JL> initial consonant cluster.

>'Sreally interestin'.
'Deed it 'tis. Both statements are true -- details follow, as per implicated request -- but they have different contexts.

In the context of the set of English words (the "lexicon"), /sr-/ is a prohibited initial cluster. That is, no English words begin with /sr-/, though lots start with /sl-/, which is extremely similar. This is called "lexical phonology", or "word-formation rules", or the like.

In the context of English speech, however, it is a very obvious fact, due to the stress-timed nature of English, and its habit of reducing unstressed syllables under the timing pressure, that English speakers habitually pronounce many clusters in phrases that would be forbidden in a single word. This is called "fast speech rules" or "phrasal phonology" or "assimilation" or the like. The two kinds of phonological rules are quite different. Lexical phonology is much more categorical than phrasal is, and so there are many more and much stricter rules about word and syllable formation than there are about phrase or clause formation.

[This difference between the Phonology of words vs phrases parallels the differences between the Grammar of words vs phrases, and the Semantics of words vs phrases, by the way -- Language is full of symmetric patterns like that.]

I can't say _____ really means I can't say ___ in a word. When I go to teach American English speakers how to pronounce the unEnglish sounds of foreign languages, I'm usually able to find an English phrase that they can chant normally until it zaps their semantic interpreter and they can hear it as if it were a foreign word and they're pronouncing it correctly. If you think you're pronouncing a phrase, you see, you don't pay careful attention to the pronunciation -- you just do it.

For one very detailed instance: lots of American English speakers claim that They Just Can't Say the German initial cluster /cv-/, as in the word zwei ('two'). Now, it's true enough that [tsv-] is a forbidden initial cluster in English words, and if the student insists on doing their talking and learning exclusively at (what they (think they) know of) the word level, they're usually sunk anyway. But if they don't have imaginations that are atrophied from too many years absorbing what passes for Reality in the educational system, I usually urge them to try this:

Suppose you were having a discussion about what color something was.
[For non-US Englishers, this may sound weird, but such conversations are common here, especially between men and women, because men and women tend to differ in their usage of English color vocabularies -- though nobody takes such differences very seriously -- and so these episodes allow relatively safe flirting]
And suppose the offending object is colored in some shade at the non-red end of the visible spectrum.
  So somebody says    It's purple
  and somebody says   No it's lilac
  and you think to yourself
             no that's the wrong flower.
             this looks exactly like the color of a violet.
             not like lilac.
  and you say          No it's violet
                           ^^^^^^----------- and there it is: [tsvai].

Now, if you're speaking natively and naturally, and you're concentrating on something besides how you would spell what you're saying, you have a certain likelihood of dropping the /I/ part of it's, especially since it's unstressed and located between two strongly stressed syllables (/no/ and /vai/). The English tendency there is to reduce that syllable somehow, and vowel deletion is one common way.

You might or might not say it this way in this conversation, but since this conversation is imaginary anyway, let's imagine it as if you did, OK? So, try saying it out loud. Repeatedly.

Try pronouncing [no tsvai] as part of how you'd say No, it's violet in this imaginary conversation. Then ignore the last syllable and concentrate on saying the rest over and over again until your English understander gets tired of understanding what [no tsvai] means every time you say it, and goes off and starts thinking about something else more interesting, and you can put the repetition on automatic pilot. Then you can hear all of the [no tsvai] as just sounds.

Then you do the magic part: begin to imagine you, speaking German. Feel free to imagine any conversation partners you choose to accompany you -- this is all imaginary, of course. Now imagine you listening to the [tsvai] you're saying as if it were a German phrase, which might be spelled Noh, zwei -- never mind the meaning, you don't speak German yet, meanings come later. But I guarantee you some German speaker might say it, and they'd say it pretty close to the way you're saying it (we can work on your vowels later).

Then you can learn to hear it in German phonemes, as in

     /cvai/   [tsvai] 'two'
where the German /c/ phoneme (spelled ) is pronounced [ts]. This is completely analogous to the way the English /c^/ phoneme (spelled ) is pronounced [tS] in
     /ic^/   [itS] 'each'

Then, suddenly, it dawns on you that you're not just imagining hearing this, and that you're not imagining saying it. In fact, you yourself are speaking a word in German with German sounds, and hearing it in German with German sounds, and it sounds right. And it feels good to be able to do it. Keep on doing it as long as you like.

-- OK, done for now? You can practice this again any time.

This kind of practicing and guided pronouncing is what underlies Language Laboratories in educational institutions. Anybody can learn to pronounce words in any language; you just have to learn how to hear the sounds as Phones [fo:~nz] instead of Phonemes /fónìmz/. After you do that, you can learn how to rearrange the phones into a different language's phonemic pattern, and pronounce them using different rules.

[There was a thread a while back in a.u.e about what's a phoneme? or something like that. This is what's a phoneme. Phonemes are part of languages, phones are part of Language.]

-- 4

>John Lawler writes:

 [regarding using Engl. it's violet to learn Germ. zwei]

>This method is certainly useful. The word "disregard" would work for the
>initial /sr-/ combination. Similarly, English speakers can be given a
>grasp of aspirated versus non-aspirated consonants by this method. For
>example, the difference between Mandarin Chinese zai4 and cai4 can be
>approximated by the difference between the the t-s-i sequence in "it's
>ironic" and the t-s-h-i sequence in "it's highly ironic".

>However, there's no guarantee that a student, even upon grasping the
>pronunciation of a new consonant or consonant cluster, will be successful
>at using it as an initial sound. Spanish-speakers, for example, have no
>trouble pronouncing /sp/ and /st/ medially, since their own language has
>these clusters, but many, when speaking English, fail to pronounce these
>as initials, prefixing them with at least a shwa. In fact, their own
>language at some point tacked the initial vowel e- onto all the words that
>in Latin began with sp- and st-, so it would seem the Spanish people
>either already had trouble assimilating initial sp- and st- when Latin was
>first introduced, or prepended the initial e- later for some reason and
>became "unable" to dispense with it.

That's true, and the method doesn't work so well for speakers of syllable-timed languages like Spanish, which just don't have the resources (i.e, the requisite "rules", or linguistic subroutines) that are necessary to get rid of the extra syllables. Spanish speech apportions approximately the same amount of time to every syllable, so there's no need to reduce them, and consequently there's no need for vowl r@duction, final consonan deletion, unstressed sylble deletion, assimilation, cluster simpification, and all those other rules that English speakers practice minutely.

It works better for English speakers, though, for that reason. English is stress-timed, apportioning approximately the same amount of time to stress groups of syllables between major stresses, no matter how many syllables they have. This is the measure that English uses for poetry scansion

   'Twas the eighteenth of April in Seventy-Five
   Hardly a man is now alive
   Who remembers that famous day and year
and it helps delimit constructions, and so on. But it sure teaches English speakers how to get rid of unnecessary unstressed syllables, which they do with great abandon.

You're right, of course, that there's no guarantee, but when is there ever in teaching?

-- followup - 5

>>> When we say, "It's raining," to what does "it" refer?

>> Let me guess.  Some teacher in grade school taught you
>> that all pronouns must have an antecedent...

>In my case, Mrs. Morris.  Emily Morris, R.I.P.  A great teacher.  We
>had to diagram stanzas from "Jabberwocky" to make sure we understood,
>at least at the 7th or 8th grade level, what we were doing.

In my case, a whole series of nuns. By the time I reached (public) high school (Midwestern US, late 1950s) I found I knew traditional grammar better than most of my teachers, simply by having paid attention in grammar school. The fact that Latin was a familiar part of every Catholic school day helped; no doubt my teachers had studied Latin, anyway. And later I learned that there was a whole lot more to English, and to English grammar, than they'd taught us there, too.

But that's hardly a surprise; there's a whole lot more to mathematics and science and art and literature and history -- to name a few things -- than anybody is taught in grammar school. Nobody would seriously propose that we freeze our understandings of these subjects, or of grammar, at the 7th-grade level (though it appears there are some people who do so).

>> All the common contractions with subject it are really grammar
>> markings:
>>          it's  [meaning 'it is']     Progressive or Passive
>>                [meaning 'it has']    Perfect
>>          it'd  [meaning 'it would']  Conditional
>>                [meaning 'it had']    Pluperfect
>>          it'll [meaning 'it will']   Future

>Well, to a descriptivist (a self-confessed one, I've read your
>confessions myself), ...

Whoa - wait a minute; I haven't confessed to anything. You have apparently accused, tried, and convicted me of being a "Descriptivist". I'm not copping to that. While it is of course your privilege as a speaker of English to mean whatever you intend to mean by whatever you say (Die Gedanken Sind Frei, after all), you shouldn't be surprised if others do not always acquiesce tamely.

"Descriptivist" appears to be a label like "Liberal" or "Christian" or "Counter-revolutionary" that gets tacked on to some group of people of whom the speaker disapproves. It's a private label that means whatever you intend it to mean, like Humpty Dumpty's "Glory". It is a political term, or a moral term, but not a linguistic term. Unless you believe that grammar is a political issue or a moral issue, in which case we part company, because I don't.

If you want a label that I'll accept, try linguist. Linguistics is a science, and of course scientists must describe the objects of their study; again, nothing new. There are lots of other things we do, too; though describing phenomena as completely and accurately as possible is very important. It turns out in linguistics, as it does in all sciences, that you have to observe very carefully, and report what you find very, very carefully, if you want to learn anything you didn't already know. Or thought you knew. So in that sense I'm a (lowercase) "descriptivist", like any other scientist, though the word is not one I have any use for.

But you appear to mean something different by "Descriptivist", some kind of side-taking in somebody else's perennial dispute, and I'm afraid that, frankly, I don't give a damn about any of that. I post and publish my grammatical pieces in the hope of providing satisfying answers to questions. No one is under any obligation to read or understand them, nor to agree with them, believe them, or find them satisfying. Nor is anyone under any obligation to speak or write as I, or you, or our grammar-school teachers might have them speak or write.

> ... they're just "grammar markings."

Nor did I say they were "just grammar markings". I said they were "really grammar markings". Which is to say that thinking of them that way pays off in terms of understanding other things. For instance, the history of English shows that there has been a wholesale loss of inflections in the last millenium. For those whose understanding of grammar begins and ends in inflection, we may appear to have lost our grammar, and this might cause a little righteous panic among the naive.

But these pieces of grammar haven't been lost at all -- they're just not inflections anymore. Now they're syntactic constructions, a different part of grammar, but still part of grammar. Though a part of grammar not sanctified by Latin, because Classical Latin hadn't lost its inflections and therefore didn't have anything like the syntactic complexity that Modern English does.

>But back in the dim old early history of English, nobody sat down and
>said, with or without benefit of grammar or vowel shifting, "Hey, guys
>and gals, let's put together some things to use as grammar markings.
>And let's use a grammar marking to talk about weather.  Cool, eh?"

Of course not. Not only isn't there the slightest evidence that such a thing happened, but things like that rarely ever happen (or if they do, rarely, they never affect anything), because language change is both unconscious and social, not conscious and individual. Not that I ever said anything different; but perhaps you think of this as (a caricature of) the story I am telling. Sorry to disillusion you, but no.

>Rather, someone got into the habit of saying "The weather is rainy," or
>"the weather is raining," or possibly "the sky is raining today" (in
>some pre-modern form or another). And, over the years, that became
>shortened and shortened, to the form we see today: "It's raining."

Nice story. I see you understandably prefer your own story to the one you imagined I was telling. Your privilege. My take on this story is precisely the same as the one above: things like that may happen, to individuals, rarely, but they're irrelevant to the language. And of course there's no evidence for this story either. But tastes differ, and if you prefer this mythology, go for it.
De gustibus non disputandum est.

>Then, and only then, could a descriptivist look at the construct and
>say, "That's a grammar marking, it isn't a referential pronoun."  It
>isn't a referential pronoun mainly because the reference was dropped
>before that particular descriptivist was born.

I wouldn't venture to guess when any "Descriptivist" might be able to do anything, nor what they might venture to do, nor what motives they might have for doing so. This is your story, not mine, so I can't argue with it. Maybe something interesting happens after the reference drops.

>> Solution: a Dummy Subject, obeying the letter but not the spirit of
>> The Rule.  One of several: this particular construction is called
>> Weather 'it', to distinguish it from Distance 'it' in

>>     It's a long way to Schenectady

>> or Extraposition 'it' in

>>     It's important to me that she be present.

>Obviously, the sentence springs from "That she be present is important
>to me," or "She being present is important to me," or something similar.

Right. That's what Extraposition means. It's a technical term, like "Subjunctive" or "Nominative" or "Indirect Object". It's just that Extraposition wasn't invented by the Latin grammarians like those others (though it was formed from Latin roots, like most grammatical terminology used in English). Latin didn't require a separate subject word the way English does, and didn't have strict word order the way English does, and consequently didn't require rules like Extraposition that readjust the word order and leave dummy subject words behind, the way English does.

The languages have changed a bit since Pompeii, and we've learned a few other things since then, too. Nowadays we rarely talk about one sentence "springing from" another, though I imagine the Romans liked the image, like Athena springing from the forehead of Zeus. Mythology is great fun; but so is science.

(And the subject of a gerund is traditionally in the Possessive Case, or the Objective, nor the Nominative; that should probably be Her being present is important to me. Of course, getting the details right is less important if all you're doing is telling a story.)

>So the question to me is not what a descriptivist decides the grammar
>happens to be in 1996 (or perhaps the concept of Weather 'it' and
>Distance 'it' came along in 1995, or 1895), but how the original form
>mutated into the current form.

Well, if that's the question to you, you might like some references you could use in beginning to answer it, if you ever want to. I'd recommend David Crystal's Cambridge Encyclopedia of Language and Cambridge Encyclopedia of the English Language, first. After you go through them, the places to look are in the OED, and in Jespersen's English Grammar, and in Baugh's (or anybody else's) History of the English Language, and in Manfred Görlach's Introduction to Early Modern English.

But you appear to be satisfied enough with your own story, so perhaps there's no need. (It's too bad, though, after going to the trouble of making up this "Descriptivist", that you then profess to be uninterested in what they're doing. Oh, well.)

>I enjoy /enormously/ reading your explanations of language from
>your point of view, but you seem to view it as a static thing to
>be examined and measured on a day certain.

Thank you. Though I'm unaware of ever having had that idea, much less ever saying so. I view language as an enormously complex phenomenon that is close to the human soul, and I view it with enormous respect. Awe, in fact. Some things about it can be measured and examined, like actual behaviors; and some parts can be statistically treated; and others can be predicted by rule. And everything has a history. These things can be learned. I believe that those things about language that are knowable should be known by those interested in forming opinions about it; contrariwise, opinions about language formed by those who ignore linguistic reality are rarely valuable, though they may, of course, be lots of fun.

You may have noticed the Sapir quotation in my .sig.
Here's the whole paragraph it comes from:

"Everything that we have so far seen to be true of language points to the fact that it is the most significant and colossal work that the human spirit has evolved -- nothing short of a finished form of expression for all communicable experience. This form may be endlessly varied by the individual without thereby losing its distinctive contours; and it is constantly reshaping itself as is all art. Language is the most massive and inclusive art we know, a mountainous and anonymous work of unconscious generations."

I respect language far too much to settle for comforting myths about it. Stories are fun, but they're no substitute (imho) for accuracy and clarity.

>I don't think assigning a name or recognizing a null value in the
>sentence matters much, because the sentence we use today is not the
>sentence that prompted original use of the pronoun or its antecedent.

Sentence? Singular? You think one sentence was the First Weather It? A separate creation myth for every syntactic rule? Wow. You don't think small.

>Today (at least where I am), it suffices to look up at the sky and say
>"Raining," or perhaps "Ugh." But, the former isn't a full thought,
>it's a convenient short form,

What criteria are you using for full thought? In Malay, when one says Hujan (one word, one morpheme, indifferently noun or verb, meaning 'rain'), one often means what one might mean in English by (It's) raining, with or without the dummy subject and form of be that's sanctioned in English. There's no other way to say it in Malay, though, and I'm sure you wouldn't want to say that Malay speakers are unable to complete a thought on this subject, just because they don't use a dummy subject.

My guess is that you don't have a good definition of full thought that's separate from the rest of English grammar. Explaining the construction by reference to such a criterion is name magic, just assigning a name with a null value. Maybe you're right; maybe it doesn't matter much.

>and that is probably what "It is raining" is, too.

If you say so. As I said, you are at liberty to believe anything you please, with or without evidence. This way of thinking, like all ways of thinking, is optional, as is thinking itself, or even wanting to.
I think I'll pass, myself.

  - John Lawler       Linguistics Department and Residential College     University of Michigan

    "Language is the most  massive  and  inclusive  art  we know,  a           - Edward Sapir
      mountainous and anonymous work of unconscious generations."       Language (1921)

More English Grammar   More About Language   The Eclectic Company   The Chomskybot