Talk:Grammatical number

From Wikipedia, the free encyclopedia

WikiProject Echo has identified Grammatical number as a foreign language featured article. You may be able to improve this article with information from the French language Wikipedia.

Contents 1 The inverse number scheme 1.1 reply 1.2 more on number toggling (inverse #) 2 Russian 3 Inflection 4 Merge with English plural? 5 Older/uncategorized discussion 6 Explanation wanted 7 Semantic and grammatical roles of number 8 semantic vs. syntactic uses of number in Slovene 9 Tlingit example 10 Two Mistakes 11 Number vs. Person 12 Merger proposal 13 Jemez noun class III description 14 Slovene removed 15 "An alleged example" 16 The commented out paragraph 17 Unnecessary comment? 18 Sanskrit congruence 19 Nullar Number

[edit] The inverse number scheme

I believe that the current last sentence of the first paragraph ("A unique number system called inverse number is found in the Kiowa-Tanoan language family of North America") should be removed from that paragraph. Since Ish ishwar reverted my removal of it, and I'm not interested in starting a revert war, I'm starting discussion over here. (I'm assuming it will be mostly me and Ish discussing it with each other, but obviously others are welcome to participate as well.) My reasons are as follows:

1. No other specific language or language family is mentioned in that paragraph, even English, even in the most minor way.
2. The Kiowa-Tanoan language family has only three grammatical numbers: singular, dual, and plural. Since this paragraph is primarily a description of the different grammatical numbers, and these three numbers are described earlier in the paragraph, nothing further need be said about this family in particular.
3. The current presentation makes it sound as though "inverse number" were another grammatical number, or perhaps a set of numbers that's completely different from singular/plural and variants. Neither is the case.
4. The Kiowa-Tanoan system is unusual in its way of marking number (that is, in its system of noun inflection), but since the first paragraph does not touch on methods of marking number, it is not the right place to point out this unusualness.

If you disagree with anything I've said, please reply (Ish or anyone else). Thanks. Ruakh 04:24, 15 Mar 2005 (UTC)

Agree. It's a very interesting phenomenon, but it doesn't add another systematic differentiation. It is sufficiently mentioned in its section. BTW, it is not that unique.

Some Russian words display in certain casus the same phenomenon of an inflected singular form dropping the ending in the plural (e.g. рыбу - рыб). Sebastian 05:05, 2005 Mar 15 (UTC) — Discussion continued below.

Hi. Yes, I am antiwar too. Perhaps we can edit-dance instead. Ish ishwar 04:26, 16 Mar 2005 (UTC)

Sorry, I don't understand. What do you mean by "edit-dance"? Ruakh 04:56, 16 Mar 2005 (UTC)

On point (1), I dont know if this is a major concern. Is it? I dont know. I think it nice to give endangered languages top billing — maybe a native speaker will be happy to have their language mentioned so prominently in cyberspace. If this is offensive to some readers, then please remove.

Regarding (2) & (3), I will probably need to think about this & reply more deeply later. My initial interpretation of the discussion I read in Mithun (I havent checked out the primary sources) is that yes, this is a distinct type of number that is different from singular, dual, plural, etc. numbers.

We usually speak of grammatical number as the intersection between semantics & morphology. So, Jemez nouns only have the two features, marked (=inverse) and unmarked, in its inflectional number category. The unmarked number varies semantically depending on the noun class.

You are suggesting that unmarked and marked number should be divided into semantic components 1) sg., 2) dual, 3) pl. It seems that one must do this in order to figure out the semantics of this inverse number. But this is not indicated morphologically. There is only one suffix that indicates the inverse of the usual unmarked number.

Trying to think of a parallel in English, I see that the word "pair" inherently refers to a dual referent. But, grammatically "pair" is singular. "Pairs" which is plural refers to 4 or any greater multiple of 2. Similarly, "scissors" has grammatical and semantic number mismatching in that it is grammatically plural but the referent is singular. "Pair" and "scissors" have the semantic number specified lexically which differs from grammatical number. I was thinking that Kiowa-Tanoan has these noun classes similarly specified in the lexicon with a particular semantic number like "pair" is specified as dual.

But I should think more about this.

Peace. - ishwar (SPEAK) 05:04, 2005 Mar 16 (UTC)

I don't think endangered languages should get top billing, no. I don't think an intentional bias toward narrowly-spoken languages is any improvement over the existing unintentional bias toward widely-spoken ones.

I understand how grammatical number could be viewed as you describe, as the intersection between semantics and morphology (and syntax); but the paragraph in question restricts its consideration to semantics, leaving morphological and syntactic details to later paragraphs.

I stand by my assertion that semantically speaking, Jemez has three numbers, with an unusual system for marking them. I might be convinced otherwise if you could demonstrate either (1) that statistically, unmarked nouns of each word class occur much more often than the corresponding marked nouns in everyday speech or (2) that syntactically, neighboring words (determiners, adjectives, verbs, or the like) do not change between noun classes, but do change when a noun enters inverse number. Even if that is the case, I'd be opposed to the current sentence, though, and would prefer something like, "Some languages have completely different systems of grammatical number; see [[#Inverse number|the section on ''inverse number'']] for an example."

Your English example ("pair") doesn't quite work, since according to the table, dual number is always unexpected in Jemez, regardless of noun class. I see where you're going with your example, but I think I'm leaning toward an Occam's Razor approach and saying that the explanation currently given in the Inverse number section is the simplest and best (for an English wiki, at least). Ruakh 16:38, 16 Mar 2005 (UTC)

RE edit-dancing

You spoke of editing & arguments in terms of war. This is a very common metaphor in English. I am told that in Tibetan language/culture arguments are spoken of in terms of dance. I like this visual image, so I suggest that we use the nicer metaphor. - ishwar (SPEAK) 05:17, 2005 Mar 16 (UTC)

I didn't speak of arguments in terms of war; I spoke of them as an alternative to a revert war. That is an interesting cultural point, though. I wonder if the Tibetan use of the dance metaphor occurred naturally, or if someone at some point made a conscious decision to adopt it instead of an existing metaphor. I also wonder if the connotation of dancing is less positive in Tibet than here. ;-) Ruakh 16:38, 16 Mar 2005 (UTC)

[edit] reply

Hi.

Yes, I think that your description is probably a better one. So, nice work! I thank you helping me out. I dont know if a statistical data would convince of anything though... I am thinking about trying to get some more information on inverse number. If I can get an electronic copy of some papers & you are interested, leave me a note & I will send some your way.

Re lang billing: I think that I will always disagree about this. I am for the injured & the marginalized. Any help is good. I think that bias is inevitable so it is nice err on the side of generosity. The past injustices are still in us and I feel them. But, this is a personal feeling. I know that others do not share my philosophy. Since you do not like to have this language family mentioned in the first paragraph, I encourage you to remove it.

Re war: I just noted that you envoked this metaphor. You then suggested that you did not want to enact it. I agreed. I think metaphor is tremendous. There is a nice book called Metaphors we live by linguist George Lakoff & philosopher Mark Johnson. The ARGUMENT IS WAR metaphor is one of first metaphors that they mention in this book. Here are some pieces of it online if you are interested: http://theliterarylink.com/metaphors.html. I dont know if Lakoff has ever written anything on Buddish argument—I just heard this from one of his friends. Peace. — ishwar (SPEAK) 02:01, 2005 Mar 22 (UTC)

[edit] more on number toggling (inverse #)

Hi. I said I would look up some more about this. Here is description from Weigel (1993:468-469, 477):

"1.3 Kiowa noun number marking. Nouns in Kiowa (Tanoan; southern U.S. plains) may be singular, dual or plural but may also be marked as ambiguous between two values. (In some cases this ambiguity will be eliminated by coreferential pronominal affixes on the verb.) Thus noun stems are divided into three classes that are inherently either sg,du or du,pl or du.⁴ (A fourth class is not marked for number at all and is not relevant here.) The "inverse" suffix -GO, (with extensive phonologically conditioned allomorphy) toggles the underlying number marking (Watkins 1984:78-90):

	(4a)	[cę̂·]sg,du	+GO	[cę̂·gɔ̀]pl	'horse'
		[t’ą́y]sg,du	+GO	[t’ą́ymɔ̀]pl	'egg'
	(4b)	[tʰǫ́·sè]du,pl	+GO	[tʰǫ́·sègɔ̀]sg	'bone'
		[tʰá·]du,pl	+GO	[tʰê]sg	'wife'
	(4c)	[k’ɔ̂n]du	+GO	[k’ɔ̨̂·dɔ̀]sg,pl	'tomato'
		[é·]du	+GO	[é·bɔ́]sg,pl	'bread, grain'

"Notice that sg and pl toggle, but stems are all marked +du, and affixed forms -du. The resulting rule is:

(5)   [[noun stem]_{+αsg,+du,+βpl} GO]_{-αsg,-du,-βpl}

"Are there credible alternatives to analyzing these data as instances of morphosyntactic toggles? There seem to be two candidates: (i) renanalysis as an exchange rule (reversing the logic we just applied to Dinka); or (ii) reanalysis as several 'normal' rules. As we will see, the first option is impossible, while the second is highly undesirable.

"An exchange feature analysis of Kiowa would toggle the phonology, in the hope that the morphology would then act normally. In other words, such a rule would insert the material -GO (or some other allomorph) when it is absent and delete it when present. .... However, the Kiowa data differ from the Dinka in that this procedure will not result in a rule like (2), with a constant morphoysyntactic value in the input.

(2)   [...[+αlong]...]_-pl → [...[-αlong]...]_pl

"The reason for this is that Kiowa has six possible configurations for number marking, rather than two. (Of the 2³=8 possible sets of values for three binary features, only the combinations {sg,du,pl} and {-sg,-du,-pl} are not instantiated either on the stem or the affixed form of some class.) Thus there is no single configuration of number feature values that can serve as a constant input, as does [-pl] in (2), and at the same time yield the required configurations in the output. At best a set of rules would be necessary. A further problem is created by the values of [du]. This feature does not toggle in (5); its value is [-du] if and only if the affix -GO is present. Thus if -GO is made to toggle, so will [du]. The result would be a process that is at once both an exchange rule and a morphosyntactic toggle!

"If a single rule with constant morphosyntactic input and output is not possible, certainly a set of rules corresponding to the three noun classes is:

	(6a)		[[stem]class I -GO]pl
	(6b)		[[stem]class II -GO]sg
	(6c)		[[stem]class III -GO]sg,pl

"If it were not the case that the allomorphy of -GO is unrelated to noun class membership (Watkins 1984:80), then this might be a sensible solution. As it stands, however, (6a-c) is nothing more than a set of apparent coincidences. This multiple rule representation leaves unexplained (i) why three different rules, with different morphosyntactic consequences, have the same phonological shape and exhibit the same allomorphy, and (ii) the fact that the input and output of each rule have opposite morphosyntactic feature values. By rejecting the toggling device and thus failing to express the generalization that it embodies, (6a-c) mandate an unacceptable level of redundancy and opacity. It seems we have to accept that toggles are real."

"[footnote] 4. Multiple feature marking is another rare but attested phenomenon. Some examples are collected in Lieber (1992:93-101). Those accustomed to representing singular/dual/plural number systems with a pair of binary values may find representations of the sort 'X_sg,pl' paradoxical: i.e. if singular = [+sg,-pl], dual = [-sg,-pl], and plural = [-sg,+pl], then most of the Kiowa forms are contradictorily marked. But of course, 'X_sg,pl' does not mean X is both singular and plural, it means that the form X is ambiguous between singular and plural; it can be used to fill a syntactic slot requiring either a singular or plural noun. Because there are six such attested distinct morphosyntactic configurations, no pair of binary features suffices to distinguish them."

The rest of the paper is about the implications of toggles on linguistic theory. Anyway, just in case anyone is interested. peace – ishwar  (speak) 16:16, 13 September 2005 (UTC)

If you can write an article on the subject that's neither plagiarized nor original research, I think you should, and then add a link to it from this article; it's quite an interesting topic. Ruakh 16:49, 13 September 2005 (UTC)

Sure. Sometime in the unknown future. It would nice to compare what happens in the different languages as we can see that Kiowa is different from Jemez. I would also need to get a copy of the Jemez dissertation. But, I guess first I would like to get a page on simply the Jemez language and people. – ishwar  (speak) 17:29, 13 September 2005 (UTC)

[edit] Russian

(In reply to #RussianExample)

Hi Sebastian.

I dont know Russian and I cant figure out your explanation. Can you elaborate further? Thank you. Peace. - ishwar (SPEAK) 05:20, 2005 Mar 16 (UTC)

Well, of course it's not as cristal clear as the Kiowa-Tanoan example. Regarding it as inverted grammar wouldn't be a helpful concept because it only affects two oblique cases, while all the other cases have different endings. — See Russian_grammar#Nouns, Second Declension. Sebastian 06:15, 2005 Mar 16 (UTC)

[edit] Inflection

What does "commonly" mean in the sentence "Languages that distinguish grammatical number commonly do so by inflection."? Are there any exceptions? Sebastian 02:54, 2005 Mar 16 (UTC)

I guess it depends what it means. If it means "inflection of the noun," then spoken French is an exception: it usually distinguishes grammatical number, generally through the use of determiners. If not, then I'm not aware of any exception, but there's bound to be some language that uses particles that some grammarian considers to be a separate word. Personally, I wouldn't feel comfortable changing it to drop the word "commonly" unless I had a source to cite, but if you make that change, I won't revert. A better change might be to add "of the noun" to the end of that sentence, and then add "as well" somewhere in the next one. Ruakh 04:50, 16 Mar 2005 (UTC)

Some languages indicate number through derivation not inflection. So, yes, there are exceptions. I will have to look up some languages for you.

French still indicates number through inflection even though it is not indicated on nouns. Many languages indicate grammatical information like number on the verb. In head-marking languages, number is indicated on the heads of phrases, which means that in these languages prepositions can be inflected for number. I believe that all European languages (except Basque) are dependent-marking, not head-marking.

If speaking of isolating languages that use separate words (as opposed to affixes), then we dont need to talk about inflection or derivation since there is no morphology involved.

Peace. - ishwar (SPEAK) 05:34, 2005 Mar 16 (UTC)

Re: "French still indicates number through inflection even though it is not indicated on nouns": yes, that was my point.

Re: "Some languages indicate number through derivation not inflection": that seems like a contradiction in terms. The difference between derivation and inflection is primarily a semantic one: inflection changes number, gender, tense, aspect, etc., while derivation changes part of speech, word function, etc. Or do I misunderstand something?

Re: isolating languages: I'm not so sure. Firstly, I'm not aware of any isolating languages that mark number in a separate word. (That's not to say that it doesn't exist, simply that I'm unaware of it.) Even Chinese, in the rare instances where it distinguishes grammatical number, does so by what is inarguably inflection. Secondly, if there is a language where number is marked in what could be considered a separate word, that doesn't mean it couldn't also/instead be considered part of the same word. (Consider Japanese particles, which are considered by some to be separate words, and by some to be inflection; see Inflection#Inflection_in_various_languages.) Ruakh 16:47, 17 Mar 2005 (UTC)

No, there is no contradiction. Different languages do the same thing sometimes through derivational processes, others through inflectional processes. The difference between derivation & inflection is not necessarily only based on semantics. In fact, most criteria used to distinguish between the two is not based on semantics since semantics is rather unreliable. There are a number of different criteria (which are not mentioned in this wikipedia article). I created the bibliography for that article. If you consult practically any book on morphology there will a chapter (or more) dedicated to the issue of inflection & derivation. In a nutshell, inflection is relevant to syntax while derivation is not. There probably is a continuum between the two. It is an interesting topic which I should learn more about.

About isolating languages, since these languages (in theory) dont have any morphological things like affixes and the like, they are generally described as having no morphology. The things that synthetic languages indicated with morphological processes are indicated by syntax. Of course, are there any purely isolating languages out there? I dont think so. The two textbook examples are Vietnamese & Chinese languages. But, both of these have morphology even though it is rather minimal. If something was a separate word, I think that I would call it a separate word & not part of another word. However, it is very hard to define formally what a word is especially cross-linguistically. Japanese particles are in between words and affixes. They are something like postpositional enclitics. Clitics are essentially equivalent to affixes at the phrase level rather than the word level. But, one could perhaps argue for a continnum between a free word and a bound morpheme (perhaps envoking a notion of a prototypical word). Again much if these issues can be read about in morphology texts, much more eloquently and skillfully than I can. Cheers! — ishwar (SPEAK) 02:25, 2005 Mar 22 (UTC)

[edit] Merge with English plural?

It seems that much of the information on formation of English plurals duplicates information at English plural; I think that section should be removed (or greatly shortened), with a link to English plural, and whatever information is here that's missing from English plural should be added to English plural. Formation of plurals in other languages can be moved to other pages as well. Does anyone have any thoughts on the matter? - Ruakh 00:19, 22 Dec 2003 (UTC)

This still sounds like a good idea. What became of it? Shouldn't we at least tag the pages with {{merge}}? Sebastian 02:54, 2005 Mar 16 (UTC)

I don't know about merging them, but a page on grammatical number certainly shouldn't describe how to pronounce the English plural suffix. It would be better to focus on how plurals (and other types of number) are formed and used in general, rather than how a specific language does so. --Damezi 07:31, 31 Mar 2005 (UTC)

Don't merge them

[edit] Older/uncategorized discussion

In making the English plural rules a little more accurate I've removed the fourth alternant " -/dz/ next to n. Some dialects may not have this one.", because if this occurs it's a phonetic process, not a phonological one, and dosn't bear on the plural morphology. - Gritchka 13:50, 10 Jan 2003 (UTC)

[edit] Explanation wanted

Early in the article, the claim appears that:

Grammatical number is distinct from the use of numerals to specify the exact quantify of a noun; number is usually vague.

This sounds odd to me: what is intended by this? --- Charles Stewart 17:39, 17 Jun 2005 (UTC)

I assume it refers to the fact that grammatical number usually doesn't tell you exactly how many items are under consideration. For example, in English, we use the plural for 0 gallons, 0.5 gallons, 2 gallons, or 3,631,298 gallons of milk. Now, English's singular does give an exact quantity (which, incidentally, was probably the intended word, as quantify is the wrong part of speech), and some languages have dual or even trial numbers, but overall, grammatical number tends to be very vague. Feel free to rewrite the section to make it more clear. Ruakh 03:03, 18 Jun 2005 (UTC)

hi. yes, i, too, think it can be better written. please write if you have time. peace – ishwar  (speak) 04:51, 2005 Jun 18 (UTC)

[edit] Semantic and grammatical roles of number

I moved this section from the article, because it doesn't relate to the topic of the article; however, it does seem encyclopedic, and I hope it will find a home in some article. Ruakh 03:09, 18 Jun 2005 (UTC)

The most characteristic use of numbers is as cardinal quantifiers, but it is possible to use numbers as ordinal quantifiers and as nominals (Wiese 1993). For example, in English:

1. In There are twenty horses in the ring, twenty is used to indicate the cardinality of the collection "the number of horses in the ring";
2. In Paul finished twentieth in the race, twenty is used to indicate the ordinality of Paul's position in the sequence "the finishing positions in the race";
3. In London's number twenty bus goes to Waterloo, twenty is used as an adjective to select which of "London's buses" we are talking about.

The above examples see the number 20 being used in three different roles, and as both quantifier and adjective. Additionally, number may be used as a noun, as indicated in "The number of horses is twenty", and "Paul's finishing position in the race was twentieth".

Not all of these semantic and grammatical roles occur in all languages. For example, languages originating in cultures that do not need to rapidly generate many tags for similar objects will generally not need nominal uses of number.

hi. yes, this a bit tangential.

i do wish that something could be written about grammatical number vs semantic number, though. cheers – ishwar  (speak) 04:53, 2005 Jun 18 (UTC)

[edit] semantic vs. syntactic uses of number in Slovene

The Grammatical number#Slovene section looks quite obscure since it doesn't make a clear distinction between the uses of the forms for indicating the semantic number and the uses due to syntactic control by the numerals. Particularly, in

When a number reaches one hundred and one(two) (or several hundred or thousand), singular and dual are used again. (ena knjiga (one book) (singular),dve knjigi (two books) (dual), pet knjig (five books) (plural), sto ena knjiga (101 books) sto dve knjigi(102 books))

it is obviously the latter case (of "syntactic" number), but the description doesn't state it explicitly. Probably, the table of number forms in Slovene would look differently for semantic number of stand-alone nouns and for "syntactic" number of nouns controlled by numerals. (As it would in Russian, see section Счётная форма [Counting form, or Adnumerative] in A.A. Zaliznyak, Русское именное словоизменение [Russian nominal inflection], pp. 46–48.)--Imz 19:45, 28 November 2005 (UTC)

This use of the terms semantic and syntactic confuses me, but I think I see what you're saying. In Slovene, if I say those books, then I'll use the plural form of book, even if I happen to be talking about exactly two books; but if I say those two books, then I use a specific form of book that agrees with the number two? Ruakh 23:28, 29 November 2005 (UTC)

That's true. Those books would be (tiste) knjige, while those two books would be (tisti/tistile) knjigi. --Eleassar ^{my talk} 14:28, 5 May 2006 (UTC)

I wouldn't agree, and above two statements are false or conflicting. I might be talking about exactly two books while using the form those books (ti knjigi) (no use of 'two'). While those books meaning more then two books would be te knjige. Actually the Ruakh's statement is typical example if dissapearance of dual in informal slovene talk. Young people would use the plural form although meaning those (two) books. I know this section has been moved anyway, but just wanted to point out to anyway research this here. --User:Lukanova--222.152.244.105 13:14, 13 January 2007 (UTC)

I just added a rewrite tag. The current text suggests incorrectly that Slovenian nouns can inflect for 4 numbers, but in fact there are "only" 3 numbers in Slovenian (singular, dual, plural). The difference in the syntactic structure of noun phrases involving the numerals (2)-3-4 vs. 5-and-above is found in most other Slavic languages, and it is not a question of grammatical number (as Imz explained above). I would delete this section altogether (and give Slovenian a mention under "Dual number"), or expand it to discuss Slavic in general. CapnPrep 01:12, 5 November 2006 (UTC)

[edit] Tlingit example

Some languages differentiate between a basic form (collective) which is indifferent in respect to number, and a more complicated derived form for single entities (singulative). For example, in Tlingit (a Native-American language spoken in Alaska), the plural of tɬingit ("man") is tɬingitq, which really refers to a unitary group of men, not to a random unconnected group of men. Similarly, the plural of q'aat ("island"), which is q'aat'q'i, really means something like "archipelago".

There are too many problems with these examples so I've deleted them. For starters the examples are incorrect. The word lingít /ɬɪŋɡɪ́t/ means "person, people, Tlingit" and its (rarely used) plural is lingítx' /ɬɪŋɡɪ´tx’/ which means "people, group of people". An island is x'áat' /x’áːt’/ and its plural would be *x'áat'x' /x’áːt’x’/ "islands, group of islands" but I've never seen or heard this used before. Tlingit plurals seem to be nonproductive, i.e. they are restricted to only a small poorly defined class of nouns. Otherwise the language has no grammatical number marked for nouns, only for pronouns. Furthermore the basic form of a noun is unnumbered-singular, not collective with a derived singular. I've deleted the part of this paragraph which refers to Tlingit, but I left the initial sentence. Hopefully someone will find a real example of this.

By the way, judging from the examples I'm betting that this was taken from either Swanton's (1911) Tlingit grammar sketch or Boas's (1918) Tlingit grammar, both of which are unreliable. Swanton spent very little time studying Tlingit and made many mistakes with the phonology, and Boas relied on only one informant for his work. Both are admirable for their efforts at the time, but should not be used as reliable sources of information. Please see a modern source for Tlingit language information in the future. — Jéioosh 20:15, 24 January 2006 (UTC)

[edit] Two Mistakes

Two serious mistakes:

"In many languages, such as English, number is obligatorily expressed in every grammatical context". True of some languages, but definitely not English. For example, English has no agreement in number (or anything else) with adjectives--unlike, say, most Romance languages. English also does not have agreement in number with past-tense verbs, or 1st or 2nd person present tense.

"Grammatical number may be thought of as the indication of semantic number through grammar." Not always, and this is the crux of a long-standing problems in linguistics. Let me go gather some refs and I'll explain it, but briefly: Grammatical number is sometimes morphological number, and sometimes semantic number. These are often the same, but not always, and it's not entirely clear when semantics wins and when morphology wins, much less why.

[edit] Number vs. Person

Sorry if this is long and rambling yet still vague. If anyone reads this and would like me to dig up more refs, get my thoughts in order, and add to the article--or just wants me to explain what I'm talking about--just ask.

There's a recent paper by Michael Cysouw with some great background on plural morphology and a survey in a wide variety of languages. I couldn't find it, but while searching, I found a good paper by Stephen Wecsler ([1]). Wecsler gives a good background of number mismatch and a variety of related issues, including everything I was trying to get across, plus related issues in T/V (formal 2nd) plural, 3rd plural, conjunctions, pluralia tantra, etc.

Wechsler also has a fascinating solution, but I don't know enough to judge it beyond that. In case you're curious, but don't want to read Wecsler's paper, the idea is simple: Instead of 2 numbers and 3 persons, we have 2 numbers and 5 persons, but only one can be marked. He develops some implications (in LPG) and explores a few languages, but it's clearly not a completely-developed theory. Anyway, you get:

• unmarked (traditional 3rd sing)
• plural (traditional 3rd pl)
• 1st person, or 1s (traditional 1st sing)
• 1st aggregate, or 1a (traditional 1st pl, mostly)
• 2nd person, or 2a (traditional 2nd pl, or formal/deferential/etc. 2nd fake pl ["vous"/"usted"/etc.])
• 2nd intimate, or 2s (traditional 2nd sing, informal/intimate/etc. ["tu"/etc.])

Anyway, back to the main story. I'll stick to 1st person plural in Indo-European languages with a 2-number/3-person system, and pretend gender and other features don't exist. But most of this applies to 2nd plural (and some of it to 3rd, and probably to separate local/distal and so forth), and to non-IE languages, and (I don't know, but I'd bet) to 3-number systems and beyond.

A probably radical way to state the issues:

1. Unlike 3rd person, 1st plural is not semantically, morphologically, or syntactically the plural of 1st person.

2. In many languages, there are a variety of cases where plural morphology can be used for singular semantics (and vice-versa). If other parts of the phrase must agree with the plural/singular element, some parts will take singular, some plural. There is some variation among languages, and among lexical roots (for the mixed element, or for the main verb). This variation has not been entirely characterized, much less explained. (It's what Wechsler is trying to solve in the link.)

Let me explain #1:

A true 1st plural would refer to a group of speakers. The English 1st plural can refer to this, but it can just as easily refer to inclusive we, exclusive we, complete we, and special cases like authorial we, royal we, etc. Unlike English, many languages make distinctions among these uses. You might have two extra lexemes for exclusive and complete. You might not have any of the artificial plurals, and instead use artificial third person when talking to/being the King. Almost any partition you can imagine seems to be instantiated. Except that, at least according to (my recollection of) Cysouw's survey, no language has a special word, affix, or particle for the true plural "we". By contrast, it's pretty obvious what 3rd person plural means: more than one person who isn't a participant in the discussion (or more than 2 in dual languages, etc.; from now on I'm going to ignore other numbers besides singular and plural).

Languages with a partially regular system of pronouns, affixes, etc., are far more likely to have a relationship between 3rd singular and plural morphemes than between 1st singular and plural. (Unfortunately, I didn't find a ref for this, and you don't want to take my word for it.)

Finally, most languages have cases where a word with plural morphology is used with singular semantics. (I mentioned the common cases above. I don't know how common the opposite mismatch is, or cases such as dual/plural mismatch.)

This number mismatch is bad enough, but often a mismatched agreement and an expected ("normal") agreement appear in the same phrase, as in my French example (I found it in Wechsler's paper; either it's a canonical example or I got very lucky):

• Nous avons toujours été loyal envers la grammaire générative ("We have always been loyal to generative grammar") has "nous" ("single authorial we"), "été" ("are") and "loyal" ("loyal" singular). In this sentence, "nous" is clearly morphologically plural but semantically singular.

The question is, how does the verb agree with the morphology while the adjective agrees with the semantics?

If this were always true, the problem would be easy, but in many similar cases, a mixed word takes plural agreement everywhere, and probably other combinations are also available. For optimalists, it's just a question of setting the strengths of the two conditions and seeing which one wins, but unless you love optimality theory this isn't likely to be sufficient explanation. Other simple solutions (split the number feature in two, add more information to the lexicon, etc.) don't work either.

For typical rules-based approaches (including GG/TG/G&B/whatever Chomsky calls it this decade) there has to be a principled reason to pick one or the other in each instance, in each language. Presumably there's at least one parameter involved to distinguish between different classes of languages, but I don't know if we have enough data to even guess well yet.

(There are probably good English examples, but off the top of my head the best I can do is, "Ya'll set yourself down over there." But you could argue that "ya'll" is both sing and pl, just like "you", at least in dialects where it can be used in this sentence, and therefore there's no mismatch here. In a language with a small set of pronouns, no adjective inflection, and minimal verbal inflection, it's probably not as common.) 69.107.50.208 13:22, 28 February 2006 (UTC)

[edit] Merger proposal

I am proposing a merger of this entry with plural, nullar, trial, paucal, collective noun and singulative. Here are my reasons:

• Most of those other entries have very little in them. They could easily become sections, or even paragraphs in this one;
• One exception is the plural entry, which does contain quite a bit, but a lot of what it has could just as well be in a general entry about 'grammatical number';
• I think that bringing what is in the plural entry here might improve the quality of this entry;

A related page which is probably best left separate is the one on the dual number, which seems too large to merge with this one. FilipeS 19:58, 20 April 2006 (UTC)

Merging plural, nullar, and paucal into this article makes sense, but you should be sure to propose this on the relevant talk-pages so people watching those articles have a chance to comment. Merging trial would be a bad idea, but I'd support merging trial (grammatical number). ;-) Ruakh 16:03, 21 April 2006 (UTC)

Oh, and I forgot to say that I'm less convinced about merging collective noun and singulative into this article. Ruakh 16:05, 21 April 2006 (UTC)

I support some merger. -lethe ^{talk +} 16:08, 21 April 2006 (UTC)

Oops, thanks for the heads up, Ruakh! I've corrected the link on the main page. I will copy the merger proposal above to the talk pages of the other articles.

As for 'collective noun', my opinion is that, while the collective is not a grammatical number, the only time we need to talk about collective nouns is when we have to explain whether they take singular or plural verbs and determiners. For example, when we discuss whether to say 'The committee is meeting' or 'The committee are meeting'. And this has to do with number agreement.

Why would you single out the singulative? ;-) FilipeS 14:39, 22 April 2006 (UTC)

But the Collective noun article gives a good deal of encyclopedic information that's completely irrelevant to the notion of grammatical number. Ruakh 15:44, 22 April 2006 (UTC)

I see what you mean. That does seem too specific of collective nouns to be in an article about grammatical number... FilipeS 15:55, 22 April 2006 (UTC)

I'm withdrawing 'collective noun' from the list. FilipeS 20:13, 29 April 2006 (UTC)

[edit] Jemez noun class III description

Jemez noun class III is given as "other animate nouns." Judging from the fact that II is "some inanimate nouns", shouldn't III be "other inanimate nouns"? (I don't know for sure, but even if III does contain "other animate nouns", class I should say "some animate nouns". 198.150.76.150 18:59, 27 October 2006 (UTC)

Oh my, you're right. It looks like I accidentally introduced that error more than a year and a half ago when I restructured that section (diff). Thanks for pointing it out! Ruakh 21:57, 27 October 2006 (UTC)

[edit] Slovene removed

I have decided to be bold, and remove the Slovene example from the page. Reasons:

• According to this conversation, the example contains some inaccuracies.
• The example is redundant, since Slovene has a singular/dual/plural number system, like Hebrew, for which there is already an example.
• Number in the Slavic languages (and in particular the dual) is extensively discussed in the Dual grammatical number article.

I will, however, copy the Slovene examples to the talk page of the Dual grammatical number article, in case someone wishes to improve it, and add it to the other article. FilipeS 14:18, 5 November 2006 (UTC)

[edit] "An alleged example"

An alleged example of this situation is Desano, a Tucanoan language of Colombia. Cf. gasi "canoes" vs. gasiru "canoe"; yukü "trees" vs. yukügü "tree".

I have added an "unsourced" tag to these oddly worded sentences. FilipeS 14:33, 5 November 2006 (UTC)

[edit] The commented out paragraph

Verbs may agree with the number of nouns to which they refer, even when there is no other form of number agreement in a language. For more information on this special type of number agreement, see Grammatical person, Verb, and English verbs.

This is supposed to mean that even in languages where nouns are not inflected at all, verbs may be inflected for number. To tell the truth, I can't think of any example right now, though. FilipeS 19:10, 31 December 2006 (UTC)

[edit] Unnecessary comment?

The following was deleted for being an 'unnecessary comment':

(The s in grands is pronounced in this case due to liaison.)

The problem is that earlier in the article the point is made that the plural ending -(e)s is normally silent in French. FilipeS 19:12, 31 December 2006 (UTC)

I deleted that, because it's irrelevant to the point being made: the ending change occurs whether or not liaison is triggered, and there was nothing special about the example that makes its particular pronunciation relevant. Our example could as easily have been about un grand crayon and deux grands crayons.

Incidentally, what's this about a plural ending -(e)s? I can't think of a single French noun or adjective that forms its plural by adding -es; am I overlooking something obvious?

—Ruakh_TALK 22:37, 31 December 2006 (UTC)

It's not irrelevant. If there were no difference in pronunciation, then the ending change in writing wouldn't prove anything about whether there are number inflections in the French language. In un grand crayon / deux grands crayons, for instance, the singular and the plural forms of grand are the same. The fact that you write them differently is a totally arbitrary convention. You could just as easily write un gran crayon and deux gran crayons without any loss of linguistic information -- were it not for the liaison, the only case where there is a real difference between the singular and the plural of the adjective.

As for the typical plural suffix, I think you may be right. It's -s, rather than -(e)s. FilipeS 12:32, 2 January 2007 (UTC)

If the pronunciation is what's important, and the orthography is secondary at best, then we should give IPA for the example (either instead of the normal French spelling, or in addition to it). —Ruakh_TALK 03:29, 3 January 2007 (UTC)

[edit] Sanskrit congruence

"Sometimes, grammatical number will not represent the actual quantity. For example, in Ancient Greek and in Sanskrit, neuter plurals took a singular verb." This is very true for Ancient Greek, but is never found in Classical Sanskrit. I checked my grammar book and asked my professor of Sanskrit, but he knew nothing of such a rule. (And there is no scarcity of rules in Sanskrit!) Therefore I am going to delete the mention of Sanskrit in the cited sentence. —The preceding unsigned comment was added by 84.58.229.134 (talk) 19:59, 15 January 2007 (UTC).

Your investigation and contribution are appreciated. :-) —Ruakh_TALK 04:44, 16 January 2007 (UTC)

[edit] Nullar Number

I removed the Nullar Number section as Latvian does not have a nullar number, nor can I find any details in linguistic literature of such a thing existing. The only references I can find to its existence are mirror pages to Wikipedia, or conlangs. See Talk:Plural#Nullar_numbers for more information. 86.143.131.161 15:13, 4 February 2007 (UTC)