This was too long for a comment, and I think it starts to go towards an answer, so I've posted it as such.
Assumptions are, as I'm sure you're aware, often problematic. Modern Japanese, and to a lesser extent, Late Middle Japanese, are the odd men out when it comes to pronouns in Japonic languages. Old Japanese and the Ryukyuan languages have pronominal systems that do not really resemble their sister language.
Okinawan has what might be called a three dimensional pronominal system. First, there is a two-way distinction between non-polite and polite forms. Second, a four-way person distinction (first, second, third, and indefinite persons). And third, a two-way plurality contrast (singular and plural). This description is from Kinjou (1974)'s grammar of Naha dialect Okinawan.
All non-first person pronouns show the contrast between polite and non-polite forms. For instance, the non-polite form of the second person singular pronoun is 'yaa [ʔjaː], while the polite form is either 'unzu [ʔun.dʑu] or naa [naː]. Excluding the second person naa, all other polite pronouns are formed in an entirely regular way, by adding the contracted form of the attributive copula =n (uncontracted nu, homophonous in both forms with the genitive case clitic =nu) to the bare pronoun, along with a contracted form of the word tchu [ttɕu] 'person'. So the second person form given above, 'unzu, is from pre-Okinawan *uri nu ttchu, literally 'the person who is you'.
Plurality is similarly formed in a very regular fashion. The plural suffix -taa [taː] is added on to pronouns. Curiously, the pronouns lose any vowel length they may have had, and the intiial t- in the suffix geminates. So for instance, the first person singular pronoun is waa (in some morphophonological contexts, wan), while the plural is wattaa.
Person is interesting, because only the first and second person pronouns are not polysemous with some other function. The third person pronouns, for instance, are simply the demonstratives. Okinawan has a three-way contrast, between proximal, mesial, and distal, and speakers retain this even in their pronominal usages. So kuri is not only 'he/she', but 'he/she near me', and you can also have uri 'he/she near you', or even ari 'he/she near neither of us'. The indefinite person, taa doubles as the wh-word 'who'.
The Old Japanese pronominal system is even simpler than this, and is detailed in Vovin (2005)'s grammar of Western Old Japanese, as well as in Kupchik (2011)'s dissertation on the grammar of the various Eastern Old Japanese dialects.
So what does all this mean for your question? Well, I think Joe Pineda touched on this in his comment. It seems that the particular social circumstances of late medieval and early modern Japan radically altered the pronominal system into a very unusual form. However, it's not that unusual. The pronominal systems of many of the languages of Southeast Asia, like Vietnamese or Thai are usually held up as examples of extremely open (in the sense of an open class) pronominal systems. However, I wonder if it's even really the case that these words are pronouns anymore?
In any case, I think there are good arguments to be made that Japanese pronouns really are a special subclass of nominals (pronouns) that just happen to be less diachronically stable than, say, pronouns found in most other languages. See Ishiyama 2008 for more extensive arguments about this, but in summary, they don't exactly behave, morphosyntactically, like "regular" nouns (although they are much more noun-like than English pronouns); they are really not that open (you can't just go adding more to them at will); and they have very limited, but very pronoun-like, semantic content.