Polysemy in function/non-lexical words.

[Legion]

In French, the word "que" handles a lot of different functions.


Interrogative pronoun (for inanimate, direct object):

Que vois-tu ?
What do you see?

It can extend to inanimate subjects, but only though some syntactic tricks:

Qu'arrive-t-il ?
Qu'est-ce qui arrive?
What is coming/what is going on?
but
*Qu'arrive?

It is also the relative pronoun for direct objects (both animate and inanimate):

L'homme que je connais.
The man I know.

Le livre que je veux.
The book I want.

It handles comparative (both equal and unequal):

Il est aussi grand que toi.
He is as tall as you.

Il est plus grand que toi.
He is taller than you.

It is also a subordinating conjunction:

Il m'a dit que tu étais malade.
He told me you were sick.

It's also a restrictive determiner:

Il n'y avait que trois personnes.
There were only three persons.

Il n'a fait que parler.
He just talked/He did nothing but talk.

Je ne veux que ton bonheur.
I just want you to be happy/All I want is your happiness.


That's not bad for two phonemes (and unstressed). Do you have example of function words in other languages that cover a wide range of functions?

[Soap]

Well English has the -of vs -'ve distinction, which often is merged as /əv/ in speech. Obviously not merged in writing, but growing up a lot of kids dont even realize theyre different words and write things like "I should of known". If you accept homophones of words that are otherwise unrelated there could be a lot of examples in a lot of languages. "than" and "then" are homophones in most sentences for me ... not even schwa, its /DEn/.

[Ser]

Well English has the -of vs -'ve distinction, which often is merged as /əv/ in speech. Obviously not merged in writing, but growing up a lot of kids dont even realize theyre different words and write things like "I should of known".

But that's something particular to the perfects of those modals, otherwise people don't say/write "I of" instead of "I've".

If you accept homophones of words that are otherwise unrelated there could be a lot of examples in a lot of languages. "than" and "then" are homophones in most sentences for me ... not even schwa, its /DEn/.

That's because both are diachronically the same word, from Old English þanne. Etymonline says, in fact, that they weren't distinguished in writing at all until about 1700.

[Legion]

Oh, and, there's more!!

Exclamatory adverb:

Que de poissons !
So many fish!

Interrogative causal adverb (a bit archaic):

Que ne l'a-t-il pas dit tout de suite ?
Why didn't he say so right away?

(more modern: Pourquoi ne l'a-t-il pas dit tout de suite ?)

[zompist]

And don't forget the ne...que construction!

I think etymologically all of these go back to two roots, though.

[linguoboy]

OMG Welsh yn /ən/ and Irish a /ə/.

[zompist]

English 'that' has a number of roles:

deictic adjective: that man

deictic pronoun: I'll have that too

Subordinator: I think that you're right

Intensive adjective: You're not that smart

Particle in expressions of desire: Oh, that I were famous!

'More of the same': 1066 and all that

[Ser]

And don't forget the ne...que construction!

He didn't forget it, he just calls it a "restrictive determiner". (:

I think etymologically all of these go back to two roots, though.

Yeah, the Trésor de la langue française in fact thinks about same, saying it comes from three Latin words: qui, quid (in their accusative forms) and quia:

Que conj. représente la conj. lat. quia [...]. Que rel. cas régime atone masc. sing. est issu du lat. class. quem (acc. masc. sing. du rel. qui) [...]. Que pron. interr. est issu du lat. quid « quelque chose, quoi », nomin. et acc. neutre sing. du pron. interr. quis en position atone (cf. quoi*).

http://www.cnrtl.fr/etymologie/que

[zompist]

And don't forget the ne...que construction!

He didn't forget it, he just calls it a "restrictive determiner". (:

Ah c'est vrai. I was reading and writing quickly in the middle of a game lobby.

[Jipí]

German der/die/das serve as both definite articles and relative pronouns, and in spoken language as ~demonstrative pronouns (der/die/das hier/da 'the here/there'); the articles themselves sometimes do double duty in declension (e.g. der is masc. nom. sg., fem. gen. sg., fem. dat. sg., gen. pl. depending on context). Also, als, which means both 'when' and 'than', and which has also has an aspect-ish function in some 'lects. And werden 'become' is used to form both passives (werden + past participle) and the future tense (werden + infinitive).

And ffffffffffffuuuu- communication works in spite of overlapping functions!!!!11!

[Legion]

OMG Welsh yn /ən/ and Irish a /ə/.

OMG... what do they mean?

[Salmoneus]

English 'that' has a number of roles:

deictic adjective: that man

deictic pronoun: I'll have that too

Subordinator: I think that you're right

Intensive adjective: You're not that smart

Particle in expressions of desire: Oh, that I were famous!

'More of the same': 1066 and all that

Not just deictic pronoun - also relative pronoun:

This is the horse and the hound and the horn
That belonged to the farmer sowing his corn
That kept the cock that crowed in the morn
That woke the priest all shaven and shorn
That married the man all tattered and torn
That kissed the maiden all forlorn
That milked the cow with the crumpled horn
That tossed the dog that worried the cat
That killed the rat that ate the malt
That lay in the house that Jack built.

There's also another use, I'm not sure exactly what you would call it. It belongs to 'that' in its role as a distal demonstrative, in that the plural form is 'those', but semantically and syntacticaly I think it's quite distinct - that is, its role as a sort of pronoun-of-free-selection modified by a relative clause.

What I mean by that is this: "That which makes us happy makes us free", "that which the palmerworm has left, has the locust eaten", "that which we call a rose, by any other name would smell as sweet", etc.

It can also have this 'free selection for relativisation' function as an adjective, as in "that government is best that governs least".

It may also be worth mentioning that 'that' is actually several different types of deictic: it can be used as a distal demonstrative ("look at that cloud!"), as the first-offered in a deictic contrast ("that one's nice, but this is better" - note how 'that' can be something you're holding, so it's not a normal distal), as an anaphoric textual deictic (see above where I say 'what I mean by that is this' - there's no distal/proximal distinction here and i'm not pointing at anything - both deictics here are standing for passages of text (utterances), and the difference is that 'that' refers to a preceding text and 'this' refers to an immediately following text). Also, like many 'distals' it's also used to indicate location at the interlocutor as well as a genuine distal - and in some cases it can also be used in this way to refer to past-tense events. The classic example of that is punching someone and then saying "take that!" - 'that' is not something the guy you've punched actually has, it's an event in the past (the punch), whereas the same event in the present or future is encoded as a proximal ('hey, laugh about THIS!').

Also, it turns clauses into subjects, as in "that you have no nose is obvious to me". It's tempting to think this is exactly the same as the normal subordinating function (ie the subordinating function creates the same nominal arguments but in object position), but i don't think it is. 'Feel' aside, you often can't turn these 'objects' into subjects through the passive as you should be able to: "I think that I might get a coat" cannot be turned into "That I might get a coat is thought by me" - well, technically I suppose that is legal, but it really doesn't seem the same. So this may or may not be a distinct function.

[Radius Solis]

Not just deictic pronoun - also relative pronoun:

I would understand "subordinator" on the list to cover this - relative clauses are, after all, a type of subclause. (And there's more unity between them than first meets the eye: in both relative clauses and other subclauses, "that" appears with clauses that are complements, generally not with adjuncts.)

edit: On the other hand I don't really see the "more of the same" one either. How is it different from the normal duty of a deictic pronoun? The "and all that" could be replaced without major changes in meaning by "and all of that", "and that whole business", "and that whole thing", or sometimes "and everything to do with it". These phrases all have 1. something to express universal inclusion (all, whole, everything) and 2. some pronoun or longer expression referring back to the 1066. So all of the above seem pretty compositional to me.

[Bob Johnson]

Well English has the -of vs -'ve distinction, which often is merged as /əv/ in speech. Obviously not merged in writing, but growing up a lot of kids dont even realize theyre different words and write things like "I should of known".

But that's something particular to the perfects of those modals, otherwise people don't say/write "I of" instead of "I've".

Just wait.

I think a better example though is /z/ which is possessive <the dog's running area> plural <the dogs run there> is <the dog's running> has <the dog's run already>.

but then that's not a word...

[Melteor]

^^You didn't mention i,t but the possessive is also ambiguous as to number (but distinguished in writing), hence "the dog's running area = the dogs' running area". I also have a it as a contraction of 'has', which is why we have constructions like "that dog's got big legs, dunnit = ?that dog HAS GOT big legs? == that dog HAS big legs" (incidentally I've seen typography recently that emphasizes <BIG. LEGS.>; I just think that's funny but apropos); I think it's really a phonological process that people feel like has gone too far, like the 'has' doesn't feel like a syllable anymore. OTOH I'm pretty sure that I can reinsert some high-central vowel that isn't schwa back in there if I need to make 'has' syllabic again, but then it still doesn't contrast with full 'is' (I can say something like, "He is got some funny things going on," and I'm kind of amazed I've never seen that in writing before. The vowel might be [e] though, which I have crop up as an allophone of /ə/ and maybe /I ɨ/. There's something going on with the final consonant of the noun, like I don't want to release it but sort of do anyway, which is different from the contracted copula 'is' and the possessive.) Most of these cases aren't quite as ambiguous as you think, but it's sometimes tricky i.e. "the dogs over there != the dog's over there"; I have a feeling I could distinguish them by intonation only, most of the time. It still counts even though it's a morpheme in my book.

Salmoneus: This might be an old wive's tale but I've heard 'what' is a contraction of 'that which'.

OP: Sometimes I have double-that "that that", and I'm not entirely sure why...Well, one is sort of making an idea (i.e. "That we should be faced with this is unprecedented") and the other is pointing (i.e. "That's a bad idea"); so I could say something like, "Not that that was my intention..." but honestly I have yet to meet a teacher who puts up with it in writing, so I'm out of the habit. (Thanks jmcd for reminding me of this one!)

[Radius Solis]

I think a better example though is /z/ which is possessive <the dog's running area> plural <the dogs run there> is <the dog's running> has <the dog's run already>.

but then that's not a word...

No, but you missed two. Our -s has six functions; the aforementioned:
1. the plural suffix
2. the possessive clitic
3. contracted "is"
4. contracted "has"
...plus:
5. the 3sg/present verb suffix "The dog runs."
6. the little-used modifier derivation "Evenings I walk the dog." (also lexicalized in e.g. "towards", "overseas")

But there are three etymological sources, not just one: the plural, the OE genitive gave rise to both 2 and 6; and 3 and 4 are from the same verb ending in 5.

[Salmoneus]

Not just deictic pronoun - also relative pronoun:

I would understand "subordinator" on the list to cover this - relative clauses are, after all, a type of subclause. (And there's more unity between them than first meets the eye: in both relative clauses and other subclauses, "that" appears with clauses that are complements, generally not with adjuncts.)

This seems like hand-waving to me. Sure, you can define some super-category 'subclause' that includes both these phenomena. But then you may as well say that 'that' only ever has one use, as a 'that'-ifier, and then just define all the other uses as examples of thatification.

After all, it's clear that these two uses of 'that' are entirely distinct from one another syntactically in reality, regardless of how chomsky files them in his mental filing-cabinet. The relative pronoun, for instance, is followed by a sequence of words forming a verbal phrase that is missing an argument, leading us to want to analyse the pronoun as in fact being that argument, as it shares its semantics (and in other languages similar constructions are explicitly marked as verb arguments). "I ate the dog that you kicked", for instance, has this verb phrase "you kicked", which lacks an object, and lacks its full meaning, until you realise that 'that' is what is kicked. Even more strikingly, "I ate the dog that bit you" - "bit you" is not even a valid clause in its own right. But "that bit you" IS a valid clause, because 'that' is what bit you.

In the other case, however, the subordinator actually does subordinate - that is, it indicates that the following, by itself completely valid, clause is in fact subordinate to the preceding one. In "you must realise that 'that' is what is kicked", there are two potentially independent clauses, "you must realise" and "'that' is what is kicked", and "that" acts to subordinate the latter to the former.

Since "I ate the dog that bit you" does not actually subordinate a clause, I wouldn't call it a subordinator.

[Melteor]

^^Well, this certainly does happen in certain dialects of English, which would use 'what' instead of 'that' here, i.e. "I ate the dog what bit you."

[ná'oolkiłí]

I suppose it depends on your definition of subordination, Sal. Usually it just means embedding of a clause. In this sense, both [ John said [ that he saw Mary ] ] and [ I ate the dog [ that bit you ] ] are instances of subordination, even if in the latter the subclause is 'incomplete' (but note that resumptive pronouns are very common in colloquial spoken English, eg I ate the dog that it bit you). I wouldn't say this is handwavy, it's just a definition of recursive clause embedding.

But of course you're right in saying relatives and regular embedded clauses are different syntactically. Usually in the case of the latter you just say that a verb takes a CP ( = S; a clause) complement. There are several analyses of relative clauses, but generally you either say that the head noun takes a CP as its complement which contains a silent element that is linked to the head ( [_DP the dog_i [_CP that Ø_i bit you ] ] ) or that the head actually originates within the embedded clause and raises up ( [_DP [_CP that the dog bit you ] ] → [_DP the dog [_CP that bit you ] ] ).

The syntax of relative clauses is very complicated, and a matter of quite a bit of debate. It is possible to analyze the that in relatives as a demonstrative pronoun that stands in for the head — something like this is normally said for wh-relatives (the man I saw who → the man who I saw). But it is also said that the that is a clausal subordinator, like it is in contexts of regular clause embedding.

[Salmoneus]

I suppose it depends on your definition of subordination, Sal. Usually it just means embedding of a clause. In this sense, both [ John said [ that he saw Mary ] ] and [ I ate the dog [ that bit you ] ] are instances of subordination, even if in the latter the subclause is 'incomplete'

I understood subordination as an independent clause being subordinated - I don't understand the concept of something being subordinated if it is not independent. It has to exist to be subordinated, as it were - subordinating implies it moves from not being subordinate to being subordinate. It's like you're telling me that "killing" doesn't necessarily imply the victim was alive to begin with.

(but note that resumptive pronouns are very common in colloquial spoken English, eg I ate the dog that it bit you).

I have never, ever, heard this, and i would assume that anyone who said it was non-native. I suspect English may colloquially use resumptive pronouns now and then in very complicated the speaker-has-got-lost circumstances, but in a simple relative clause? I would stare at the speaker, too shocked and confused to summon a reply. It is absolutely illegal in my dialect.

[Yng]

OMG... what do they mean?

Can't speak for Irish, but Welsh yn is an adverbial marker for adjectives and sometimes nouns:

mae hi'n rhedeg yn gyflym! - she runs fast!
dos adref yn fachgen da - go home like (as) a good boy

It also, in what I would analyse as basically the same function, adds nominal/adjectival complements to copulaic verbs:

Mae'r bechgyn yn hapus ar hyn o bryd - the boys are happy at the moment
Ti sy'n gachgi! - it's you who's a coward!

A homophonous particle, but one which doesn't trigger mutation, adds verbnominal complements to copulaic verbs:

Mae hi'n rhedeg - she's running

A third yn, homophonous in most cases but with a set of distinct allomorphs and triggering the nasal mutation, is a locative preposition meaning 'in':

Yn y tŷ - in the house
Yng Nghymru - in Wales

A fourth yn, which is nonstandard and usually triggers the nasal or soft mutation depending on dialect, is a variant of the first person singular possessive:

yn nhad - my father

A fifth, more dubious yn is a reflex of oni in some dialects and is used in tag questions:

mae e'n dod, yndyw e? - he's coming, isn't he?

Arabic ما mā has a number of functions in MSA. It can be used as a more informal negative marker for the past (and was generalised by all dialects as a negative marker):

mā qultu šayʾā! - I didn't say a thing!

It's used for 'what' in verbless sentences:

mā huwwa l-Qurʾān? - what is the Qur'an?

It is used to turn some prepositions into conjunctions:

ʿinda-mā kuntu fī ʾUrūbā - when I was in Europe (< ʿinda 'by')

baʿda-mā quddima li-r-raʾīsi ta-ʿāzī-h - after his condolences had been presented to the president (also a cool example of passivisation which nonetheless reserves VAO word order because a light adverbial has been moved before a heavier subject)

And to produce various conditional conjunctions which translate with '-ever':

ʾiḏā mā futiḥa bābu l-ḥiwār - if the door of discussion is ever opened

mah-mā qālati l-Wilayātu l-Muttaḥidah - whatever the US say

And it's also used as a generic relative pronoun:

lā ʾafhamu mā taqūlīna-h - I don't understand what you're saying

[Ser]

Arabic ما mā has a number of functions in MSA. It can be used as a more informal negative marker for the past (and was generalised by all dialects as a negative marker):

mā qultu šayʾā! - I didn't say a thing!

It's used for 'what' in verbless sentences:

mā huwwa l-Qurʾān? - what is the Qur'an?

It is used to turn some prepositions into conjunctions:

ʿinda-mā kuntu fī ʾUrūbā - when I was in Europe (< ʿinda 'by')

baʿda-mā quddima li-r-raʾīsi ta-ʿāzī-h - after his condolences had been presented to the president (also a cool example of passivisation which nonetheless reserves VAO word order because a light adverbial has been moved before a heavier subject)

And to produce various conditional conjunctions which translate with '-ever':

ʾiḏā mā futiḥa bābu l-ḥiwār - if the door of discussion is ever opened

mah-mā qālati l-Wilayātu l-Muttaḥidah - whatever the US say

And it's also used as a generic relative pronoun:

lā ʾafhamu mā taqūlīna-h - I don't understand what you're saying

It also derives some interrogrative pronouns from prepositions (e.g. لم lima 'why' < li-mā for-what, مما mimmā 'what...from' < min-mā from-what).

It can also attach to the head of an indefinite NP as a modifier, to make it generic. E.g. شخص šaxṣun 'a person' > شخص ما šaxṣun mā 'somebody, some person'; طالب ṭālibun 'a student' > طالب ما ṭālibun mā 'some student'.

[Yng]

It can also attach to the head of an indefinite NP as a modifier, to make it generic. شخص šaxṣun 'a person' > شخص ما šaxṣun mā 'somebody'

And also, apparently, to add a sense of approximation:

yastaġriqu mā bayna šahrayni wa-ṯalāṯah - it will last between two and three months

[zompist]

This seems like hand-waving to me. Sure, you can define some super-category 'subclause' that includes both these phenomena.

Or, you know, "clause". This is exactly the same classification as in traditional grammar.

In the other case, however, the subordinator actually does subordinate - that is, it indicates that the following, by itself completely valid, clause is in fact subordinate to the preceding one.

Traditional and generative grammar agree that relative and subordinate clauses are not the same thing, but are subclasses of something.
But traditional grammar seems to have a weird idea of what the difference is. Why is the notion of "independent clause", or something "completely valid by itself" as you say, important? What does it let you do?

The idea of a clause is universal, except in Pirahã— all languages have structure, and have ways of embedding a sentence in another one. Not all languages have this notion that the embedded clause is sometimes identical to how it'd appear as a main sentence. German, for instance, has a different verb ordering rule for subclauses. Quechua often uses a very different structure (something like a participle) for subclauses. The NW Caucasian languages have special verb forms used for both relative and subordinated clauses; so does Nishnaabemwin. And these are just examples I have at hand.

You can even find this in English.

* In I wish I were rich, the subordinated clause cannot occur in the same form as an independent sentence.

* In He said that he would take the garbage out the subordinated clause would have a different meaning as an independent sentence.

* And for that matter, for He told me I was crazy, although it's perfectly true that "I was crazy" is a valid sentence, it's not what he said— what he said was "You're crazy".

The important difference thus isn't whether the embedding preserves the surface form of the subclause, but whether the subclause serves as an entire argument, or as a modifier to an argument.

It's not some kind of crazy amazing coincidence that we use "that" in English, and "que" in French, for both types of embedding. Languages don't have to use the same mechanisms for both (again, off the top of my head, Russian and Mandarin don't), but it's no great surprise that some do.

[ná'oolkiłí]

(but note that resumptive pronouns are very common in colloquial spoken English, eg I ate the dog that it bit you).

I have never, ever, heard this, and i would assume that anyone who said it was non-native. I suspect English may colloquially use resumptive pronouns now and then in very complicated the speaker-has-got-lost circumstances, but in a simple relative clause? I would stare at the speaker, too shocked and confused to summon a reply. It is absolutely illegal in my dialect.

Well I for one hear it all the time, especially among younger speakers. I think it's one of those things you don't realize you actually ever say, and can't even judge if you don't produce it spontaneously. But you're probably right that the complexity of the structure is important. I ate the dog that it bit you isn't the best example; if you have a heavier noun phrase or if you're relativizing something less accessible than a direct object, then resuption is more likely. Where is also a common relativizer in these cases. How do you feel about That thing where you have to put it on the corner of an envelope so it gets sent?

———

The Georgian word რომ rom is pretty polysemous. It signals a bunch of types of subordination: complement clauses, relative clause, counterfactual conditionals, 'when' clauses, 'so that' clauses, and probably others I can't think of right now.

მითხრეს, რომ იწვიმებდა
mitxres, rom ic̣vimebda
they.told.me ROM it.would.rain
'They told me it would rain'

დედას რომ კვიცი წინ წაუხტება, მგელი შეჭამსო
deda-s rom ḳvici c̣in c̣auxṭeba, mgeli šeč̣ams=o
mother-DAT ROM foal in.front it.leaps.out.in.front.of.it wolf it.eats.it.up=QUOT
'The wolf eats the foal that leaps out in front of its mother'

რომ დაგენახათ, ვეღარც კი იცნობდით
rom dagenaxat, veğarc ḳi icnobdit
ROM you.had.seen.him can't.anymore EMPH you.would.know.him
'If you had seen him you wouldn't have recognized him'

შებინდებული იყო, კირილემ პლატონი რომ გამოაღვიძა
šebindebuli iqo, ḳirile-m ṗlaṭoni rom gamoağvija
darkened it.was Ḳirile-ERG Ṗlaṭon ROM he.woke.him.up
'It was dusk when Ḳirile woke up Ṗlaṭon'

ყველს დალბობა უნდა, რომ სიმლაშე გამოეცალოს
qvel-s dalboba unda, rom simlaše gamoecalos
cheese-DAT soak.MASDAR it.wants/needs ROM saltiness it.may.be.removed.from.it
'Cheese needs to be soaked to remove excess salt'