Grammaticalizing from N to Q in Polish and Slovak pseudopartitives: A corpus-based study of case marking patterns

Introduction: We investigate a grammaticalization path from noun to quantifier through Polish and Slovak pseudopartitives, a binominal construction where one noun, N1, quantifies another noun, N2 (1a). In the Polish tradition, this process has been termed “numeralization” (Schabowska 1962; see also Rutkowski 2006), but it resembles patterns of change documented in English (Brems 2011) and Spanish (Verveckken 2015), where a reanalysis of the N1 from head to modifier has been proposed. There is evidence that some Polish N1s are shifting to allow a numeral-like subject-verb agreement pattern (e.g. Herda 2019), and we explore whether this change extends to case assignment. We find evidence of a numeral-like case pattern for some N1s, which appears to start in the locative, and argue that such N1s are undergoing a category change from N to Q.

Morphosyntactic patterns: In Polish and Slovak pseudopartitives, N2 typically occurs in the genitive, this resembling other types of binominals (1b); we term this the “N-pattern”. This behavior contrasts with numeral constructions, where the quantified noun is marked genitive in structural case contexts (nominative, accusative), but otherwise appears in the case of the case context (Babby 1987’s heterogenous/homogeneous distinction), the numeral being inflected (Slovak/Polish) or uninflected (Slovak) (2); we term this the “Q-pattern”. 

Method: We conducted a corpus study, investigating patterns of case assignment with select N1s. The list of N1s per language, as well as corpora can be found in Table 1/Appendix. A series of queries were constructed per case context per N1. These queries captured N-patterns, Q-patterns, and possible variations of these. Sample size was limited to 100 per query, and false positives were filtered out manually. Results in the locative, dative, and instrumental are discussed here.

N-pattern: All N1s except Slovak pár₂ and Polish parę ‘a few’ predominately showed the N-pattern, where N2 was genitive across different case contexts. These N1s seem to remain highly nominal.

Q-pattern: Several N1s additionally showed a Q-pattern, where N2 appeared in the case of the case context. For most N1s, this pattern was minimal, but some N1s showed it productively for the locative; see Table 1. The results suggest that these N1s have started grammaticalizing into quantifiers, with the locative leading the change. Slovak/Polish pár₂/parę predominantly showed a Q-pattern, suggesting they have already grammaticalized.

Analysis: Grammaticalization from N to Q interacts with case assignment. When N, an N1 absorbs the external case, but as Q, it fails to intervene, leading to the external case on N2. The availability of multiple patterns for a single N1 suggests a synchronic layering of grammaticalization steps (Brems 2011), which may eventually lead to divergence between the two patterns (as is likely for Polish parę/para) or a full shift to Q. That the Q-pattern infiltrates the case paradigm incrementally may further suggest a “climbing” of the case hierarchy (Caha 2009), where the locative, a smaller case than dative or instrumental, is affected first. Altogether, our findings provide evidence of ongoing change in the case patterns of Polish and Slovak N1s.

Table 1: Prevalence of N-pattern and Q-pattern per N1 per case in Polish and Slovak, with 0 indicating no attestations, <10 or =10 a small number of attestations, and ✓ what appears to be a productive pattern. The addition of an asterisk to the checkmark, i.e. ✓*, indicates that despite its prevalence, the pattern may not be productive (to be elaborated on in the talk).

Examples

1. Noun-like morphosyntax
   a.  Z      dziećmi           jest  kupa           roboty.    Polish
        S       deťmi              je     kopa           roboty.    Slovak
        With children.INST is     heap.NOM work.GEN
        ‘It’s a lot of work with children.’ (Polish: Herda 2019; Slovak: author translation)
   b.   Przegląd oferty       Polish
         Prehľad   ponuky    Slovak
         overview offer.GEN
         ‘an overview of the offer’
                                                                         
2. Numeral-like morphosyntax
   a.   Heterogenous case assignment                                                                                                                                     
         Pięć  książek       było                           czytanych.    Polish
         Päť   kníh              bolo                           čítaných.      Slovak
         Five  books.GEN were.N.SG(default) read.GEN 
         ‘Five books were read.’
   b.   Homogenous case assignment           
         z      pięcioma        dziećmi    Polish
         s      piatimi/päť     deťmi       Slovak
         with five.INST/.Ø   children.INST
         ‘with five children’

Corpora

• Polish Web 19 (plTenTen19), Sketch Engine, Jakubíček et al (2013), 4 bil. words
• Slovak Web Corpus (SNK Web 7.0), Jazykovedný ústav Ľ. Štúra SAV (2022), 5.3 bil. words