What is the distribution of pronominal "one"?

(Based on the comments this question has received, more is needed to avoid confusion. The original question remains as stated below the line below. What is added here now is a more complete rendition that can avoid confusion.)

There are two versions of the word one in English, the cardinal number one and the pronominal count noun one. My question concerns the distribution of the latter. Why is the appearance of the pronominal count noun one immediately after certain determiners (i.e. weak determiners: a, my, your, his, her, our, their, etc.) marginal to bad for many speakers. Some examples are next:

(1) I like your pets. *Do you like my ones? vs. Do you like mine?

(2) Your kittens are cute, *but their ones are cuter. vs. but theirs are cuter.

(3) Our pastries are good, *but your ones are better vs. but yours are better.

The plural ones is used in these examples to avoid confusion with cardinal one, for plural ones is necessarily the pronominal count noun; the cardinal number one is always singular (of course).

The fascinating aspect of this phenomenon is that the examples just produced become fully acceptable if an adjective appears between the determiner and ones:

(1') I like your small pets. Do you like my big ones?

(2') Your gray kittens are cute, but their black ones are cuter.

(3') Our European pastries are good, but your Asian ones are better.

So my question concerns this mysterious state of affairs. What explains the inability of the pronominal count noun one(s) to appear immediately after a weak determiner?

The following data set illustrates a mysterious aspect of the distribution of the indefinite pronominal count noun one in English:

(1) *a one, *my one, *your one, *his one, *her one, *its one, *our one, *their one

(2) a big one, my big one, your big one, his big one, her big one, its big one, our big one, their big one

(3) the one, this one, that one, these ones, those ones

(4) the big one, this big one, that big one, these big ones, those big ones

Why are the examples in (1) bad, whereas the similar examples in (2-4) are fine? What literature addresses this phenomenon directly?

• Comments are not for extended discussion; this conversation has been moved to chat.
– prash
Apr 7, 2020 at 8:14

I will now attempt to answer my own question. I believe the distribution of strong vs. weak forms in Germanic is the core explanation for the curious acceptability contrast produced in the question. Modern German provides a clue in this regard. Adjectives following certain determiners in German, which can be called strong determiners, take endings that are distinct from adjectives that follow certain other determiners, which can be called weak determiners. It can be assumed that this distinction between strong and weak determiners (and strong and weak endings) exists across Germanic languages more generally. What we can observe, then, is that the pronominal count noun one(s) in English is licensed by a preceding strong determiner or by what would be a preceding strong form in Germanic. If no such strong determiner or strong form appears, the pronominal count noun one cannot appear -- but cardinal one CAN, of course.

The next data set from German illustrates the core observation:

(1) das klein-e Haus 'the small house'

(2) ein klein-es Haus 'a small house'

The definite article (der, die, das, etc.) is a strong determiner. After this strong determiner, the ending that appears on the adjective (just -e) is weak. In contrast, the indefinite article ein- in (2) is a weak determiner. After this weak determiner, the ending that appears on the adjective is strong.

The pronominal count noun one(s) in English can appear immediately after the English versions of those determiners in German that are strong. It cannot appear immediately after the English versions of those determiners in German that are weak. The observed correspondence across the languages is almost complete. The next table lists the weak determiners across the languages:

The pronominal count noun one(s) cannot appear immediately after almost any one of the weak determiners listed in this table. If it immediately follows a strong determiner, however, it CAN appear. The strong determiners across English and German are given in the next table:

Switching now to the fact that pronominal one(s) can appear immediately after an adjective following a weak determiner (e.g. *my ones vs. my big ones), the intervening adjective would have a strong ending in German (e.g. ein klein-es Haus). Thus, the relevant generalization is that the pronominal count noun one(s) in English must be licensed by what would be a preceding strong form in Germanic, be this strong form a strong determiner or a strong inflectional suffix on an adjective.

The plausibility of this explanation is of course reliant on demonstrating that the relevant pattern existed in earlier forms of English. Certainly we know that Old English had the morphological richness associated with modern German. My coauthor is currently investigating the extent to which the observed pattern did or did not exist in Old and Middle English.

• Would that approach mean having to regard one as cardinal in sentences like I'll get one from the shop, i.e. as having the same function as two in I'll get two from the shop? Apr 8, 2020 at 10:40
• Yes. But that is not new. Huddleston and Pullum (2002) already make that assumption. Apr 8, 2020 at 10:53
• Strong and weak declensions did indeed exist in Old English, but even without checking very deeply, I’m quite certain they did not survive far into Middle English. It lasted longest in monosyllabics, but even there, by late Middle English, it was gone – the weak ending -e had become a simple orthographic variant. The real question this answer poses is thus when the construction my one arose. I’d be very surprised if it dates all the way back to ME, much less OE – I would guess it’s a construction that’s arisen within the past 2–300 years, which would weaken this argument. Apr 10, 2020 at 21:40
• @Janus Bahs Jacquet I do not follow your reasoning. The construction my one, where one is prominal one (not cardinal one), is marginal to bad for many speakers. If some accept it, though, then that would be consistent with the assumption that it has entered the language only recently and is not yet widely accepted. Apr 11, 2020 at 14:38
• @TimOsborne Yes, but if it entered the language recently, and the strong/weak distinction disappeared long ago, then how are the two related? You cannot really argue that it arose to fill a gap that had been gone for centuries with no ill effects. Apr 11, 2020 at 14:41

If you're interested in describing speakers' knowledge of these contrasts, then you should take a look at Richard Kayne's (2017) paper 'English One and Ones as Complex Determiners.'

He notes that adjectives need to have some sort of informational focus in order to license 'one.' For instance he gives:

People who read interesting books generally profit considerably from the reading of those interesting books/*ones

Also, what determines whether a determiner is 'strong' or 'weak' in English for you? A number of determiners normally considered 'strong' like 'most' don't allow bare 'one'/'ones':

I was surprised to like the movie; I don't like most movies/*ones.

• Thanks for the tip. I will certainly take a look at Kayne's paper. Your comment about informational focus points to something I was aware of, although I have characterized the observation in terms of contrastive focus. Concerning most, it appears to be weak. I note, however, that its German equivalent, i.e. die meisten, is complex and is therefore better viewed as strong. There therefore appears to be a mismatch across the languages, although that is perhaps not a problem, since there is no clear correspondence due to the definite article die appearing with meisten but not with most. Apr 9, 2020 at 22:32
• If you are going to count most, any, some etc. as determiners then I suspect some version of your rule is valid regardless of dialect, and what changes is just the set of strong determiners. I hadn't been considering them as determiners at all though. There are a few ones left on the shelf is marginal to bad for me but otherwise anything with a quantifier is just impossible. Apr 10, 2020 at 10:13