Abstract
This paper contributes to our cross-linguistic understanding of pluractional adverbials through an in-depth, corpus-assisted study of the N(um)-nkéd construction in Late Old and Early Middle Hungarian. We argue that N(um)-nkéd pluractionals are (i) mereological-only, (ii) they can be associated with the agent, theme, time or location of the eventuality, (iii) they can modify states as well as events and (iv) they cannot instantiate pluractional comparisons across substates. These findings call for a more fine-grained cross-linguistic approach to pluractional adverbials, especially in terms of the mereological-scalar dichotomy: in addition to (i) context and (ii) the type of the N(um)-denotation, (iii) the morphosyntactic makeup of the pluractional also has to be taken into account. Adopting a diachronic approach will also enable us to shed light on a somewhat neglected aspect of pluractional adverbials: their functional load, especially in terms of the division of labour vis-à-vis universal quantifiers (‘day-by-day’ vs. ‘every day’) and distributive operators (‘all the boys one-by-one’ vs. ‘each boy’). By observing changes playing out in the Late Old Hungarian to Early Middle Hungarian as evidenced in corpora, we will show that the development and spread of bona fide universal quantifiers and of the partitive-distributive suffix -ik indeed happened in tandem with a sharp reduction of the frequency of the relevant types of pluractional adverbials.
1 Introduction
This paper aims to contribute to our cross-linguistic understanding of pluractional adverbials through an in-depth, corpus-assisted study of the N(um)-nkéd1 construction in Late Old and Early Middle Hungarian. The study of pluractional adverbials has expanded considerably in recent decades (see Beck 2021; Wu 2023 for recent overviews). While different authors have covered a slightly differing range of constructions,2 the empirical core of the phenomenon can be characterized as follows: pluractional adverbials (the term we are going to adopt as it seems to be maximally neutral to us) such as as ajtó-nkéd ‘door-plact’ below encode a relationship between an eventuality (event or state) and an N(um)-denotation.
| És | reggel | az | fráterek | ajtó-nkéd | kenyeret | koldulának. |
| and | morning | the | friars | door-plact | bread | begged |
| ‘And in the morning, the friars went begging for bread, from door to door.’ | ||||||
| (Domonkos Codex, 1517) | ||||||
Specifically, the eventuality is partitioned3 into subeventualities; and some parameter of the eventuality (agent, theme, location or time) is partitioned in terms of the N(um)-denotation; and there is a one-to-one (bijective) relation between the two partitions:
| subeventuality1 | <---> | N(um)-denotation1 |
| subeventuality2 | <---> | N(um)-denotation2 |
| subeventuality3 | <---> | N(um)-denotation3 |
| … | <---> | … |
| subeventualityn | <---> | N(um)-denotationn |
Paraphrasing Beck & von Stechow (2007): (1) is true of a begging event e iff the relevant division of the spatial extension of e is into doors, and each door was the location of a relevant subevent of e, and each relevant subevent of e took place at one of the doors.
While at this level of detail, pluractional adverbials are pretty much alike, there is in fact a fair amount of variation which, to our mind, seems to call for a systematic formal typological approach, to which we wish to contribute in this paper.
One parametric difference across languages seems to be whether the eventualities that are being partitioned include events and states (as was the case in Old Hungarian) or only events (as seems to be the case in Modern Hungarian).
Another point of variation is whether the subeventualites are obligatorily temporally ordered or not. While it has been assumed in much (though not all) of the literature4 (either tacitly or explicitly) that temporal ordering (or sequencing) is a hard-wired part of the semantics of pluractional adverbials, we will provide evidence that this was emphatically not the case with Old Hungarian N(um)-nkéd. Note that cross-linguistically, there seems to be a correlation between the obligatoriness of sequencing and the morphological makeup of the adverbial. ‘(Preposition) N Preposition N’-type pluractionals (e.g. in English and German) and ‘N-Postposition N-Postposition’-type pluractionals (e.g. in Hungarian) are obligatorily sequenced whereas pluractionals of ‘N-suffix’ type (N-wise in English, N-weise in German, N(um)-nkéd in Hungarian) are not. (Whether this rough observation holds on a larger sample of languages remains to be seen.)
Languages also differ in terms of what the subeventualities are partitioned to, i.e. what the N(um)-denotation relates to: while in the languages examined so far (and in Old Hungarian too), pluractional adverbials as a family of constructions can target the agent, patient, time and location of the event, it is conceivable that in some languages, pluractionals are more limited. Note also that different pluractionals can have a different range within the same language. While N(um)-nkéd in Old Hungarian (and its Modern Hungarian cognate N(um)-nként) covers agent, patient, time and location, the N-ről N-re (N-ela N-sub) ‘from N to N’ construction appears to be limited to time and location.
The exact nature of how the N(um)-denotation is being partitioned and mapped to the subeventualites is also subject to cross-linguistic variation. To date, most of the work has been done on a sample of two languages: English (mostly) and German (to a lesser extent), with the notable exception of Vlášková & Dočekal's (2020) analysis of pluractionals in Czech. The following main subtypes have been identified:5
mereological mapping: here, the N-denotation is partitioned either in terms of a subset-set relation (in which case N is typically a numeral or a sortal noun) or a part-whole relation, and each subeventuality is mapped to exactly one subset or part of the N-denotation:6
| The friars entered the chapter one-by-one. |
| The faithful celebrated the Passover house-by-house. |
| The size of the congregation grew day-by-day. |
| Mary ate the bread slice-by-slice. |
Mereological mapping may apply to participants (agent and theme, or more generally, some theta role of the event) or the theme or location of the eventuality. Mereological mapping can apply in the case of events and also in the case of states. The state may be constant such as in (7) below:
| kor-onkéd | dagályosok | voltatok |
| time.interval-plact | boastful | be.past.2sg |
| ‘you have always been boastful’ (lit. ‘you were boastful from minimal time interval to minimal time interval’, i.e. ‘each time interval is such that you were boastful in it’) | ||
| (Jordánszky C., 1516–1519, Deut 31:27) | ||
Or the state may be changing in degree (in so-called pluractional comparisons):
| She felt better day-by-day. |
| ‘Each day was such that she felt better than on the day before.’ |
scalar mapping (events): here, an event and its subevents refer to a change in one of the participants and the N-denotation refers to unit of the scale along which the change can be measured (that is, N is typically a unit noun in this case):7
| The crack widened inch-by-inch. |
Since pluractionals in English (specifically, the N(um)-by-N(um) construction) have been shown to exhibit all these readings, there has been a push in the literature to come up with unified models that cover all these readings. Beck & von Stechow's (2007) and Brasoveanu & Henderson's (2009) exclusively mereological approach has been shown by Henderson (2013) to be unable to cover scalar mapping. Henderson's (2013) stipulated that while Num-by-Num pluractionals are mereological, Noun-by-Noun pluractionals are exclusively scalar.8 This approach has been criticized by Wu (2023) on theoretical and empirical grounds. Wu (2023) in turn advocated for a unified (but underspecified) semantics for N(um)-by-N(um): in his model, all N(um)-by-N(um) pluractionals have the same lexical entry, however, this entry contains an underspecified free interval function which is contextually resolved either to a so-called count function (resulting in a mereological reading related to an eventuality participant), or to a path function (resulting in a mereological reading related to the time or location of the eventuality) or to a trace function (resulting in a scalar reading).
While such a unified account seems warranted for N(um)-by-N(um) in English, it remains to be seen whether pluractionals exhibit this same plasticitiy in terms of mereological and scalar readings cross-linguistically. Indeed, Vlášková & Dočekal (2020) have argued that as far as Czech is concerned, Noun-Preposition-Noun pluractionals are scalar, wherease Num-Preposition-Num pluractionals are mereological. (This of course could be technically handled by tweaking Wu's (2023) model so that the resolution of the underspecified interval function is not contextual but rather, is rule-based: in the case of nouns, it is resolved to a trace function, in the case of numerals, to a mereological count function.) As we will see, in Old Hungarian, N(um)-nkéd pluractionals (such as nap-onkéd ‘day-plact’) are strictly limited to a mereological interpretation, whereas N-Postposition-N-Postposition pluractionals (such as nap-ról nap-ra ‘from day to day’) are strictly limited to a scalar interpretation. Within the mereological interpretation, so-called pluractional comparisons across substates are unattested with N(um)-nkéd.
Thus, it appears that the availability of mereological vs. scalar interpretations can depend on (i) context (as advocated by Wu 2023 for English N(um)-by-N(um)), (ii) the type of the “N-denotation” (noun vs numeral, as shown by Vlášková & Dočekal 2020 for Czech) and on (iii) the structure of the pluractional (as we submit in the present study concerning Old Hungarian). Any cross-linguistically valid unified model of pluractionals has to take all three factors into account. Our purpose in this paper is not to develop such a full-fledged model: we content ourselves with (i) delineating which existing models fit the data from Old Hungarian and (ii) to precisely characterising how data from Old Hungarian augment the cross-linguistic explanandum with regard to pluractional adverbials.
Another oft-noted point of variation is whether a pluractional is strictly an adverbial adjunct (e.g. from day to day) or can also occupy an argument position (e.g. day after day):
| We cried from day to day / day after day. |
| *From day to day / Day after day has passed. |
As we will see, N(um)-nkéd is strictly an adjunct, as is N-ről N-re ‘from N after N’. (Only N N után ‘N after N’ can be used as an argument.)
Finally, adopting a diachronic approach will enable us to shed light on a somewhat neglected aspect of pluractional adverbials: their functional load, especially in terms of the division of labour vis-à-vis universal quantifiers and distributive operators. Pluractional adverbials often have a semantic import similar to universal quantifiers (e.g. nap-onkéd ‘day-by-day’ vs. minden nap ‘every day’) or to distributive operators (a fiúk mind egyenként ‘the boys all one-by-one’ vs. minden-ik fiú ‘every-distr boy’). By carefully observing changes playing out in the Late Old Hungarian to Early Middle Hungarian as evidenced in corpora, we will show that the development and spread of bona fide universal quantifiers (Bende-Farkas 2015, 2019) and of the partitive-distributive suffix -ik (É. Kiss & Tánczos 2018) indeed happened in tandem with a reduction of the functional load of certain types of pluractional adverbials.
The paper is organized as follows. In Section 2 we discuss the main characteristics of N(um)-nkéd pluractionals in Old Hungarian (types of N(um)-denotations, types of eventuality-N(um)-denotation relations). In Section 3, we discuss grammaticalization pathways taken and not taken. In Section 4, we consider the possible effect of the originals in case of translated texts. In Section 5, we canvas alternative expressions of pluractionality in Old Hungarian. In Section 6, we take a brief look at the landscape of Modern Hungarian. The Appendix contains aggregated data drawn from our corpus. Non-aggregated data are provided in the electronic supplement.
2 N(um)-nkéd constructions in Old Hungarian: the main characteristics
Data presented in this section are based on our detailed analysis of the so-called Old Hungarian Corpus (which contains Late Old Hungarian texts of various type plus five Bible translations representing the Early Middle Hungarian period; Simon & Sass 2012; Simon 2014). Using a broad search strategy in order to minimize the chance of false negatives and to cover various alternative spellings, we covered the totality of the corpus (including the morphologically unparsed part as well) and we identified altogether 723 cases of N(um)-nkéd pluractionals, each of which was analysed separately. In addition to exploring the availability and relative frequency of the various subtypes of pluractionals over time in the corpus as a whole, we also carried out a concordance-based analysis using the various Bible translations: this made it possible (i) to carry out a more detailed analysis of the division of labour between various types of pluractionals and between pluractionals and other elements and (ii) to map the diachronic changes in the frequency of various types of pluractionals.
Our main empirical findings are the following:
N(um)-nkéd pluractionals applied to the agent, theme, location or time of the event
N(um)-nkéd pluractionals applied to events and states
N(um)-nkéd pluractionals were limited to a mereological reading
N(um)-nkéd pluractionals could not instantiate pluractional comparisons across substates
Generally speaking, -nkéd9 in Old and Early Middle Hungarian encodes a relation between an eventuality (event or state) and an N-denotation, specifically:
the eventuality is partitioned10 into subeventualities; and some parameter of the eventuality (agent, theme, location or time) is partitioned in terms of the N-denotation
there is a one-to-one (bijective) relation between the two partitions:
| subeventuality1 | <---> | N-denotation1 |
| subeventuality2 | <---> | N-denotation2 |
| subeventuality3 | <---> | N-denotation3 |
| … | <---> | … |
| subeventualityn | <---> | N-denotationn |
Consider:
| És | reggel | az | fráterek | ajtó-nkéd | kenyeret | koldulának. |
| and | morning | the | friars | door-plact | bread | begged |
| ‘And in the morning, the friars went begging for bread, from door to door.’ | ||||||
| (Domonkos C., 1517) | ||||||
In (13), the event of begging is partitioned into subevents, and the location of the event of begging is partitioned into sublocations by the N-denotation (door here is a metonym of house/household), and there is a bijective relationship between subevents and sublocations.
Paraphrasing Beck & von Stechow (2007): (13) is true of a begging event e iff the relevant division of the spatial extension of e is into doors, and each door was the location of a relevant subevent of e, and each relevant subevent of e took place at one of the doors. Informally:
| begging subevent1 | <---> | door1 |
| begging subevent2 | <---> | door2 |
| begging subevent3 | <---> | door3 |
| … | <---> | … |
| begging subeventn | <---> | doorn |
Formally (in a Neo-Davidsonian even semantics framework, specifically, adopting Beck & von Stechow's 2007 analysis of pluractionals):
| a. | [[-nkéd]]: | λP<e, t>.λCov.λR<e, <v, t>>.λy.λe: PART(Cov, e + y). |
| **[λy’.λe’. Cov(y’) & Cov(e’) & P(y’) & R(y’)(e’)](y)(e) | ||
| b. | [[ajtó]]: | λx.door(x) |
| c. | [[ajtó-nkéd]]: | λCov.λR<e, <v, t>>.λy.λe: PART(Cov, e + y). |
| **[λy’.λe’. Cov(y’) & Cov(e’) & y’ is a door & R(y’)(e’)](y)(e) | ||
| d. | [[a fráterek]]: | ιx.(λx.friar(x)) |
| e. | [[koldul]]: | λy.λx.λe.beg(e)&ag(e,x)&th(e,y) |
| f. | [[A fráterek ajtó-nkéd koldulnak.]]: | |
| λe.<e, C> ∈ **[λl’.λe’. Cov(l’) & Cov(e’) & door(l’) & beg(e’)&ag(e’, ιx.(λx.friar(x))) & loc(e’,l’)], | ||
| where PART(Cov, e+l) | ||
| Cov: the contextually salient division (partition) of the N-denotation (and the eventuality) into subparts | ||
Note already here that this has a quantificational flavour: all the relevant sublocations (doors) are such that a relevant subevent (of begging) took place in them: this is because what we have is a partition (non-overlapping subsets, with the union of them being equal to the whole).
Below, we map out the Old Hungarian empirical landscape of -nkéd-based pluractional adverbs in terms (i) types of eventualities, (ii) types of N-denotations and (iii) types of relations between the two.
2.1 Types of N-denotations
In Old Hungarian, the argument of -nkéd indicates a partition in terms of one of the following subtypes: by participant (agent, theme, goal etc.), by time or by location. In the case of participants, partitioning happens in terms of the subset-set relation or of the whole-part relation.
2.1.1 Partitioning by agent: typically, subsets of set of individuals
Agents being typically [+animate], the partitioning here happens into subsets of a set of individuals. Consider:
| az | nagy | prímát | elvégezvén | az | atyafiak | kapitulumba |
| the | big | prime | concluded | the | friars | chapter.into |
| fej-enkéd | bemennek | |||||
| head-plact | enter | |||||
| ‘having concluded the Prime, the friars enter the chapter one-by-one’ | ||||||
| (Apor C., mid-15th c., 1520) | ||||||
In (16), the event of the friars as a group entering the chapter is divided into subevents, and the agent of the event (the set of the friars concerned) is partitioned into subsets by the N-denotation. Fő ‘head’ being a metonym for ‘individual’, the subsets are singleton sets (each containing a single friar). Informally:
| entering subevent1 | <---> | friar1 |
| entering subevent2 | <---> | friar2 |
| entering subevent3 | <---> | friar3 |
| … | <---> | … |
| entering subeventn | <---> | friarn |
Formally (in a Neo-Davidsonian even semantics framework):
| [[Az atyafiak fejenként bemennek.]]: |
| λe.<e, C> ∈ **[λa’.λe’. Cov(a’) & Cov(e’) & friar(a’) & enter(e’) & ag(e’,a’)], |
| where PART(Cov, e+a) |
In Wu's (2023) framework (inspired by Champollion (2017)), analysing fő ‘head’ as a sortal noun meaning 1 person:
| ∃πx(e).∀e’ ∈ πx(e).enterα(e’) ∧ |agent(e’)| = 1 ∧ human(agent(e’)) |
The adverbial fej-enkéd (head-wise) contributes the meaning that there is a non-trivial partition (possibly but not necessarily temporal) of the event it modifies where each cell is an atomic entering event and the agent in each cell contains only one atomic part which is a human person. Since N(um)-nkéd is limited to mereological readings, in what follows, I will used the model of Beck & von Stechow (2007) for simplicity.
While in this particular case, world knowledge suggests that the subevents are temporally sequenced, this is not a burnt-in characteristic of -nkéd-based pluractionality (see below).
In addition to partitioning into singleton sets (N: fő ‘head’, one ‘egy’), partitioning into sets of higher cardinality (N: kettő ‘two’, ötven ‘fifty’, száz ‘hundred’) and into sets characterized by shared quality (N: talentum ‘talent’) or group membership (N: sereg ‘group’, rész ‘group, part’, fejedelmi ház ‘dynastic lineage’) is also attested. (Cf. Table 9 in the Appendix and the detailed and annotated list of attestations in the electronic supplement.)
Since in the case of egyenkéd ‘one-plact’ and fejenkéd ‘fő-plact’, the subevents are distributed over individuals (i.e. singleton sets), it might be tempting to characterize egyenkéd and fejenkéd as distributivity operators. However, it is important to remember that the semantics of egyenkéd and fejenkéd is compositionally derived from the semantics of the nouns fő ‘head, i.e. individual’ and egy ‘one’ and of the pluractional suffix -nkéd. This means that staying true to Occam's razor, it is unwarranted to posit a separate category (‘distributivity operator’) for them (see Section 3 for a more detailed discussion).
2.1.2 Partitioning by theme or goal: subsets of set of individuals or mereological part-whole relation
Themes and goals may be [+animate] or [−animate], and therefore, partitioning in terms of subsets of individuals and the part-whole relation are both attested. Consider first:
| ha | valami | könyvet | lát | vala | ottan | oda | fut | vala | és | a | több |
| if | some | book | see | past | there | there.to | run | past | and | the | more |
| gyermökök | módjára | igen | hányja | vala | levelönként | ||||||
| children | in.fashion.of | very | leaf.through | past | page-plact | ||||||
| ‘if he saw a book there, he ran up to it and like the other children, he studied it intensively page by page’ | |||||||||||
| (Debrecen C., 1519) | |||||||||||
In (20), the event of Saint Thomas Aquinas studying the book is partitioned into subevents, and the theme of the event (the book as a whole) is partitioned by the N-denotation (levél ‘leaf, letter, page’) into pages.
Informally:
| reading subevent1 | <---> | page1 |
| reading subevent2 | <---> | page2 |
| reading subevent3 | <---> | page3 |
| … | <---> | … |
| reading subeventn | <---> | pagen |
Formally:
| [[Szent Tamás hányja a könyvet levelenkéd.]]: |
| λe.<e, C> ∈ **[λth’.λe’. Cov(th’) & Cov(e’) & page(th’) & study(e’) & ag(e’, aquinas) & th(e’,th’)], where PART(Cov, e+th) |
(20) exemplifies the partitioning of the theme in terms of the mereological part-whole relation, with pages being the physical constituents of a book. In our corpus, in addition to levél ‘page’, partitioning is attested in terms of íz ‘small body part’, tag ‘member, body part’, folt ‘bit’ (a cognate of Modern Hungarian falat ‘bite’), apró ‘tiny’ and ige ‘word’ (qua constituent part of a whole text).
Below, consider an example of a goal being partitioned in terms of a subset of a group of individuals:
| [mely] | eredet | bűn | szálla | fej-enkéd | mi | reánk |
| which | origin | sin | descended | head-plact | us | onto |
| ‘the aforementioned original sin descended upon us one by one’ | ||||||
| (Tihany C., 1530–1532) | ||||||
In (23), the event of the descent of original sin on humanity is divided into subevents, and the goal of the event (the set of humans) is partitioned into subsets by the N-denotation. Fő ‘head’ being a metonym for ‘individual’, the subsets are singleton sets (each containing a single human). Formally:
| [[Eredet bűn száll fejenkéd mi reánk.]]: |
| λe.<e, C> ∈ **[λg’.λe’. Cov(g’) & Cov(e’) & speaker+(g’) & descend(e’) & th(e’, original.sin) & goal(e’,g’)], where PART(Cov, e+g) |
2.1.3 Partitioning by time: subintervals (i.e. subsets) of the time of the eventuality
Partitioning in terms of the timespan of eventuality can also be modelled in terms of the subset-set relation. Representing time as a straight line, the temporal extension of a temporally uninterrupted eventuality corresponds to a line segment on this straight line, i.e. the set of all the timepoints falling within that line segment. Consider:
The temporal extension of an interrupted eventuality (one that can be divided into temporally distinct uninterrupted subevents) corresponds the the union of line segments (with each line segment corresponding to the temporal extension of the uninterrupted subevents):
(26) is not taken to mean that Mary worked in an uninterrupted fashion from Monday 00:00 to Friday 23:59. It is can be uttered truthfully if it is the case that each day of the week is such that a (possibly interrupted) working subevent took place on that day.
When -nkéd is applied to a noun denoting a time interval (in our corpus: kor ‘time interval of unspecified length’, szempillantás ‘blink of an eye’, nap ‘day’, hónap ‘month’ or esztendő ‘year’), the eventuality is partitioned into subeventualities, and the time (temporal extension) of the eventuality is partitioned in terms of the N-denotation. Consider:
| nap-onkéd | azért | siránkozik | és | szepeg | vala | [Sámson | felesége] |
| day-plact | therefore | lament | and | weep | past | Samson | wife.3sg |
| ‘therefore the wife of Samson kept lamenting and crying from day to day’ | |||||||
| (Jordánszky C., 1516–1519, Judges 14:17) | |||||||
The interrupted event of lamenting and crying is being partitioned into subevents such that each subevent corresponds to a daily chunk of the temporal extension of the event. To see how this works, consider a simple scenario where the person concerned lamented and cried for a week, and two times per day: there was a morning weeping & crying uninterrupted subevent and an afternoon weeping & crying uninterrupted subevent:
| event = subeventMoAM + subeventMoPM + subeventTuAM+subeventTuPM + … + subeventSuPM |
In (27), this event is partitioned in a way that that each subevent is such that its temporal extension is contained within a day, with the days taken together forming a partition of the whole week (the contextually given salient time period of which the temporal extension of the event forms a subset).
Informally:
| subevent1 = subeventMoAM + subeventMoPM | <---> | day1 (Monday) |
| subevent2 = subeventMoAM + subeventMoPM | <---> | day2 (Tuesday) |
| subevent3 = subeventMoAM + subeventMoPM | <---> | day3 (Wednesday) |
| … | <---> | … |
| subevent7 = subeventMoAM + subeventMoPM | <---> | day7 (Sunday) |
Formally:
| [[Sámson felesége nap-onkéd siránkozik.]]: |
| λe.<e, C> ∈ **[λt’.λe’. Cov(t’) & Cov(e’) & lament(e’) & ag(e’, wife-of-samson) & time(e’) ⊆ t’ ], where PART(Cov, e+t) |
Note that the above means that every day within the time period concerned is such that a relevant subevent took place within it. This is in effect very similar to universal quantification over days, and indeed, as we will see later on in the section about New Testament concordances, where one translator applies nap-onkéd, another translator often applies the bona fide universal quantifier expression such as minden napon ‘every day-sup’. This, however, is not sufficient to justify an analysis of nap-onkéd and its ilk as bona fide universal quantifier expressions, since their quantificational import can be compositionally derived from the semantics of their elements: the pluractional suffix -nkéd and the N-expression denoting a time interval.
Consider a further example below:
| kor-onkéd | dagályosok | voltatok |
| time.interval-plact | boastful | be.past.3pl |
| ‘you have always been boastful’ (lit. ‘you were boastful from minimal time interval to minimal time interval’) | ||
| (Jordánszky C., 1516–1519, Deut 31:27) | ||
Koronkéd is frequently used in Old Hungarian texts (see Table 8 below and the electronic supplement), and the analysis of the texts in terms of context and concordances makes it clear that its contribution is almost identical to that of a universally quantified temporal expression such as minden kor-on ‘always, lit. every time.interval-sup’. While in Modern Hungarian, kor means ‘age (both in the sense of lengthy time period and the age of an individual)’, in Old Hungarian, it had a less restricted meaning of ‘time (period) of an unspecified length’. The semantics of (31) can be derived in the standard way: the state of being boastful is partitioned into substates, and the temporal extension of being boastful is also partitioned into time subintervals such that each substate corresponds to a time subinterval. Formally:
| [[kor-onkéd dagályosak voltatok]]: |
| λs.<s, C> ∈ **[λt’.λs’. Cov(t’) & Cov(s’) & boastful(s’) & th(s’, addressee+) & time(s’) ⊆ t’ ], where PART(Cov, s+t) |
The universal import is clear here as well: if every time subinterval of unspecified length is such that the state held in it, then it follows that the state held for the totality of the time period.
2.1.4 By location: sublocations (i.e. subsets) of the location of the eventuality
In cases where the N in an N-nkéd expression refers to a location, the eventuality is partitioned into subeventualities, and the location of the eventuality is partitioned into sublocations by the N-denotation, and there is a bijective relationship between subeventualities and sublocations. For a detailed example, consider (13) above. The nouns involved in such constructions in our corpus are the following: ház ‘house’, ajtó ‘door’, ország ‘country’, tartomány ‘province’, város ‘town’, falu ‘village’, utca ‘street’, gyülekezet ‘congregation’. (See Table 11 and the electronic supplement for details.)
2.2 Types of eventualities
The eventualities concerned in -nkéd constructions involve events (such as in (20) above) and also states (such as in (7) above).
Focusing on events, we can identify two strategies of the construction of the event-subevent structure which can be roughly characterized as top-down vs. bottom-up. Consider:
| és | kik | ott | valának | mindenekre | fej-enkéd | szálla | [a | Szentlélek] |
| and | who | there | were | all.unto | head-plact | descended | the | Holy.Spirit |
| ‘the Holy Spirit descended upon each of them who were there’ | ||||||||
| (Érsekújvár C., 1529–1531) | ||||||||
Here, arguably, the starting point is the single event of the descent of the Holy Spirit on the apostles at the first Pentecost which is then divided up into subevents in terms of the partition of the theme argument: the event-subevent structure is constructed in a top-down fashion.
The other strategy is exemplified below:
| és | az | ő | szülei | felmennek | vala | esztendőnként | Jeruzsálembe |
| and | the | he | parent.3sg.pl | go.up | past | year-plact | Jerusalem.into |
| húsvét | napjára | ||||||
| Passover | day.3sg.sub | ||||||
| ‘and his parents went up to Jerusalem every year for the day of Passover’ | |||||||
| (Pesti Bible, 1536, Lk 2:41) | |||||||
Here, the starting point is a set of similar but distinct events (going up as a family to Jerusalem for the occasion of the Passover in Year 1, going up as family to Jerusalem for the occasion of the Passover in Year 2, etc.), from which a superevent of ‘going up as a family to Jerusalem for the occasion of the Passover’ is constructed in a bottom-up fashion. And it is this superevent that is then (re)partitioned into subevents.
As expected, the dividing line between these two strategies is blurred. Consider (35) below:
| ennek | utánna | űtőlök | fej-enkéd | búcsút | vőn |
| this | after | from.them | head-plact | farewell | took |
| ‘thereafter he bid farewell to them one by one’ | |||||
| (Kazinczy C., 1526–1541) | |||||
Whether (35) is to be analysed as (i) a single event of saying farewell to a group being divided into subevents or (ii) as several events of saying farewell to one person each being collated into a superevent, is very much debatable. However, for our analysis, this distinction is not crucial: either way, in the end, we have a structure of one event being partitioned into subevents.
A similar distinction can be made when it comes to states. Consider:
| [az | üdvözültek] | igen | méltóságossok: | mert | istennek | mind | fejenkéd |
| the | beatified | very | dignified | because | God.dat | all | head-plact |
| leányi | és | fiai | |||||
| daughter.3sg.pl | and | son.3sg.pl | |||||
| ‘[the saints] possess a high dignity, as they are all the daughters and sons of God’ | |||||||
| (Sándor C., first quarter of the 16th c.) | |||||||
The state of being God's child holds of each person separately: from these distincts states, a super-state of the group as a whole being in the state of being God's children is constructed in a bottom-up fashion, and it is this superstate that is being repartitioned into states.
Consider (37) below (also (31) above):
| nap-onkéd | ő | szüleit | tiszteli | vala |
| day-plact | she | parent.pl.3sg.acc | respect | past |
| ‘he was being respectful to his parents from day to day (= every day = always)’ | ||||
Here, the starting point is the uninterrupted state of being respectful to the parents, which is then divided into substates in a top-down fashion.
2.3 Types of eventuality-N-denotation relations
In Old Hungarian -nkéd-based pluractionality, the relationship between the set of subeventualities and the set of sub-N-denotations has to be a bijection, but otherwise, it is unrestricted. E.g. there is no requirement that the subeventualities be temporally ordered. Consider (5) above, reproduced here as (38):
| az | nagy | prímát | elvégezvén | az | atyafiak |
| the | big | prime | concluded | the | friars |
| kapitulumba | fej-enkéd | bemennek | |||
| chapter.into | head-plact | enter.3pl | |||
| ‘having concluded the Prime, the friars enter the chapter one-by-one’ | |||||
| (Apor C., mid-15th c., 1520) | |||||
Here, the subevents are temporally ordered. However, while we can make this inference due to our world knowledge (a group of people typically cannot enter through a door simultaneously), this is not part of the semantics of the construction. Consider (11) above, reproduced here as (39):
| [mely] | eredet | bűn | szálla | fej-enkéd | mi | reánk |
| which | origin | sin | descend.past.3sg | head-plact | us | onto |
| ‘the aforementioned original sin descended upon us one by one’ | ||||||
| (Tihany C., 1530–1532) | ||||||
Here, the subevents take place simultaneously (or at least, they are not temporally ordered).
In addition to the standard pluractional construction discussed above, -nkéd in Old Hungarian also had the potential to encode pluractional comparisons (broadly construed, cf. Beck 2012) in the case of time-denoting Ns. Consider:
| nap-onkéd | kedig | gyűl | vala | az | híveknek | ő | számuk |
| day-plact | then | increase.3sg | past | the | faithful.dat | 3sg | number.3pl |
| ‘The number of the faithful increased from day to day.’ | |||||||
| (Jordánszky C., 1516–1519, Acts 5:13) | |||||||
Such pluractional comparisons were the most frequent with nap ‘day’ (out of altogether 272 nap-onkéd constructions, 34 were pluractional comparisons) and less frequent with kor ‘unspecified interval’ (1/195) and esztendő ‘év’ (0/40). This probably has to do with the fact that day-to-day is the scale at which most natural changes are readily observable.11
Note that a large class of pluractional comparisons (broadly construed) can be collapsed into simple pluractionals or even universal quantification, exemplified below:
| és | nap-onként | számmal | bővelkednek | vala | (Heltai Bible, 1565) |
| and | day-plact | number.with | grow.3pl | past | |
| ‘and from day to day, their number increased’ | |||||
| és | minden | napon | az | ő | számuk | |
| and | every | day.on | the | 3sg | number.3pl | |
| növekedik | vala | (Sylvester Bible, 1541) | ||||
| increase.3sg | past | |||||
| ‘and every day, their number increased’ | ||||||
| (Acts 16:5) | ||||||
Put simply, if (i) each subevent is such that it denotes an increase (or decrease) along the relevant scale and (ii) the subevents are temporally ordered and (iii) the starting value on the relevant scale for subevent t equals the closing value for subevent t−1, then it follows that there is increase taking place over the course of the whole event (which itself denotes an increase).
This means that such instances (where the eventuality being partitioned is an event denoting an increase or decrease) can be modelled using the standard analysis of pluractionality (discussed above): using Occam's razor, they do not necessitate the stipulation of a separate semantic type and as such, they might be called pseudo-[pluractional comparisons].
Note however that this unified analysis cannot be extended to cases where the eventuality is a state, such as:
| Mari | napról | napra | jobban | érzi | magát. | (Modern Hungarian) |
| Mary | day.from | day.to | better | feel.3sg | herself.acc | |
| ‘Mary is feeling better from day to day.’ | ||||||
Here, Mary's level of well-being is constant within each state: the increase in well-being takes place in-between the states.
Strikingly, in the Old Hungarian Corpus, the vast majority of -nkéd-type pluractional comparisons (26 out of 28) are clearly of the pseudo sort, and involve verbs denoting events of increase or decrease: növekedik ‘grow’ (8 counts); gyarapodik ‘grow, increase’ (6); elfogyatkozik ‘diminish’ (2); bővelkedik ‘grow, widen, increase’ (2); and gyűl ‘increase’, sokasodik ‘increase in number’, sokasít ‘increase in number (tr.), teljesedik ‘grow’, szaporít ‘increase in number, multiply’, megnehezedik ‘deteriorate health-wise’, elasz ‘wither’, megárad ‘swell’ (1 each).
There are as few as altogether two instances that at first sight might count as true pluractional comparisons in our corpus. Consider first:
| hogy | az | ti | szeretetetek | innét | tova | és | nap-onként | nagyobban |
| that | the | you | love.2pl | here.from | further | and | day-plact | more |
| kimutassa | magát | az Isten | ismeretében | |||||
| manifest.imp.3sg | itself.acc | the God | knowledge.3sg.in | |||||
| ‘that your love from now onwards and day by day in a greater fashion should manifest itself in the knowledge of God’ | ||||||||
| (Sylvester Bible, Fil 1:9) | ||||||||
Curiously, however, naponkéd appears to be an insertion by the translator, as neither the Greek original, nor the Vulgate contains any corresponding element. In fact, it appears to have been a fairly common strategy especially of Sylvester and Heltai (but to a more limited extent, Jordánszky and Károli as well) to use nap-onkéd as a kind of intensifier in the case of events denoting increase/decrease (often in addition to the adverb μᾶλλον (καὶ μᾶλλον), translated as inkább ‘more’). (See Table 4 in the concordances section, also Table 8 in the Appendix, as well as the electronic supplement.) This means that one might with some justification propose an adverbial semantics for these intensifier-naponkéds, different from the general pluractional analysis, something akin to ‘continuously, without interruption’.12
Consider the second potential case below:
| És | nap-onkéd | különb | különb | csodák |
| and | day-plact | various | various | miracles |
| lésznek | vala | az | szent | testnél |
| be.3pl | past | the | holy | body.at |
| ‘And from day to day, different miracles happened at the holy body.’ | ||||
| (Érdy C., 1526) | ||||
Különb in general is ambiguous between two readings: ‘better’ and ‘different, various’. However, a survey of the way különb-különb is used in the Érdy Codex clearly shows that the authors of this Codex exclusively use különb-különb in the sense of ‘different, various’, consider e.g.:
| különb | különb | nyelvő | népek | megértették | mondását |
| various | various | language.of | peoples | understand.past.3pl | saying.3sg.acc |
| ‘speakers of various languages managed to comprehend what he had said’ | |||||
| (Érdy C., 1526) | |||||
Here, it is clear that there is no degree comparison being made among the various languages. This suggests that (44) probably simply means that various miracles happened from day to day, i.e. no comparison is being made in the sense of the miracles becoming more and more excellent each day. In conclusion, since there is only one very spurious case, it is safe to say that in all probability, -nkéd in Old Hungarian did not encode true pluractional comparisons.
3 Grammaticalization pathways taken and not taken
3.1 No reinterpretation of koronkéd, naponkéd and esztendőnkéd as universal quantifier
As we have seen above, naponkéd ‘day-plact’, koronkéd ‘time.interval-plact’ and esztendő-nkéd ‘year-plact’ did have a universal flavour. However, this can be straightforwardly derived from the meaning of the nouns concerned and the general semantics of -nkéd as a pluractional suffix. As Beck (2012) has argued, pluractional adverbial modification entails universal quantification over the members of the partition, which, being a cover, entails all relevant subsets. This means that the truth conditions that we arrive at are necessarily consistent with a universal adverbial quantification reading too (the so-called constant entailment effect, cf. also Beck & Gergel 2015). This means that staying faithful to the principal of Occam's razor, it is unjustified to posit a separate semantics for naponkéd, koronkéd and esztendőnkéd: our model for N-nkéd covers these too.
Note also that focusing on New Testament concordances, it becomes clear that with every translator, the bona fide universal quantifiers minden nap(on) and minden kor(on) coexist with naponkéd and koronkéd, sometimes even within the same sentence (see Table 4 later for tabulated data). Consider:
| nézem | vala | az | én | uramat | minden-kor-on, | |
| look.at.1sg | past | the | 1sg | Lord.1sg.acc | every-time.interval-on | |
| mert | kor-onkéd | én jogomra | áll | énnekem | ||
| because | time.interval-plact | 1sg right.hand.onto | stand.3sg | 1sg.dat | ||
| ‘I am looking at my Lord at all times, as he stands to the right of me at all times.’ | ||||||
| (Jordánszky C., 1516–1519, Acts 2:25) | ||||||
Note also that the interchangeability of a time-based pluractional with a temporal universal quantifier is not limited to -nkéd-constructions. Consider (47) below:
| nap-ról | nap-ra | az | igaz | lelket | álnokságos | téteményekkel |
| day-from | day-to | the | true | spirit.acc | deceitful | deed.pl.ins |
| kínozzák | vala | |||||
| torment.3pl | be.pst.3sg | |||||
| ‘From day to day, they vexed the righteous soul with unlawful deeds.’ | ||||||
| (Jordánszky C., 1516–1519) | ||||||
| nap-onként | aggasztalja | vala | az | ő | igaz | lelkét |
| day-plact | vex.3sg | be.pst.3sg | the | 3sg | true | spirit.3sg.acc |
| azoknak | gonosz | cselekedetén | ||||
| those.dat | evil | deed.3sg.on | ||||
| ‘From day to day, he vexed his righteous soul with their evil deeds.’ | ||||||
| (Sylvester Bible, 1541) | ||||||
| minden | nap-on | az | ő | igaz | lelkét […] | gyötri | vala |
| every | day-on | the | 3sg | true | spirit.3sg.acc | torment.3sg | be.pst.3sg |
| ‘Every day, he vexed his righteous soul […].’ | |||||||
| (Károli 1590) | |||||||
| ἡμέραν | ἐξ | ἡμέρας | ψυχὴν | δικαίαν | ἀνόμοις |
| day.acc | after | day.gen | soul.acc | righteous.acc | lawless.pl.dat |
| ἔργοις | ἐβασάνιζεν | ||||
| deed.pl.dat | torment.imp.3sg | ||||
| ‘From day to day, they tormented the righteous soul with their unlawful deeds.’ | |||||
| (2Pt 2:8) | |||||
Just as one would not argue based on this interchangeability that napról napra is in fact a universal quantifier, one also cannot cogently make the same argument wrt to naponkéd.
As we will see in Section 7, the frequency of time -nkéd expressions has decreased rapidly in the Early Middle Hungarian period. This is probably due to the spread and consolidation of bona fide quantifiers (cf. Bende-Farkas 2015). This is borne out by relative frequency data gained from the New Testament translation concordances (Table 1).13
Pluractionals vs quantifiers
| Bible translation | Date | naponkéd ‘day by day' | minden nap(on) ‘every day' | napról napra ‘from day to day' | mindenkoron ‘every time' | – | total | naponkéd ‘day by day' | minden napon ‘every day' |
| Jordánszky Codex | 1516–1519 | 12 | 3 | 1 | 1 | 5 | 22 | 55% | 14% |
| Pesti Bible | 1536 | 3 | 2 | 3 | 8 | 38% | 25% | ||
| Sylvester Bible | 1541 | 14 | 8 | 5 | 27 | 52% | 30% | ||
| Heltai Bible | 1565 | 15 | 7 | 5 | 27 | 56% | 26% | ||
| Károli Bible | 1590 | 2 | 16 | 9 | 27 | 7% | 59% | ||
| Káldi Bible | 1626 | 2 | 14 | 1 | 10 | 27 | 7% | 52% |
3.2 No reinterpretation of fejenkéd, egyenkéd as distributive operator
While fejenkéd ‘head-plact’ and egyenkéd ‘one-plact’ did have a contribution similar to distributive operators, it is unnecessary (and thus, unjustified), to analyse them as such. As we have discussed above, their quasi-distributive import can be compositionally derived from the meanings of fő ‘head’ and egy ‘one’ and the semantics of the pluractional suffix -nkéd. This means that staying faithful to the principle of Occam's razor, it is unjustified to posit a separate semantics for fejenkéd and egyenkéd: our model for N-nkéd covers these too.
Note also that generally speaking, just because two constructions have the same or similar import in terms of distributivity does not mean that they have the exact same semantics. Consider the original and different translations of the same locus (Mt 20:9):
| minden-ik | felvevé | az | ő | pénzét | (Pesti Bible) |
| everyone-dist | collect.past.3sg | the | 3sg | money.3sg.acc |
| ki-ki | mind | fizetésül | vén | egy-egy | garast | (Sylvester B.) |
| who-who | all | wage.as | take.past.3sg | one-one | penny.acc |
| azokis | fej-enként | tiz | pénzt | vőnek | (Károli Bible) |
| those.too | head-plact | ten | money.acc | take.past.3pl |
| ἔλαβον | ἀνὰ | δηνάριον |
| receive.aor.ind.act.3pl | each | denarius.acc |
| ‘they each received one denarius’ | ||
| (Mt 20:9) | ||
In the original Koine Greek, distributivity is expressed via the preposition ἀνὰ (each). Károli renders this using an -nkéd-construction, while Sylvester applies a reduplicated indeterminate pronoun (Bende-Farkas 2015), whereas Pesti uses the distributive suffix -ik (É. Kiss & Tánczos 2018). While each strategy results in roughly the same meaning, it would not be justified to assign the same formal semantics to these elements. Similar diversity is exhibited in Jn 16:32 and Rom 12:5 (see the electronic supplement).
Finally, note that egy-enkéd ‘one-plact’ can actually co-occur with a bona fide distributivity operator, which is further evidence against it being a distributivity operator:
| minden-ik-et | egy-enkéd | üdvözlé |
| everyone-dist | one-plact | greet.past.3sg |
| He greated each of them one-by-one. | ||
| (Teleki C., 1525–1531) | ||
As we will see in Section 7 below, the frequency of egy-enkéd / fej-enkéd ‘one-plact’ has decreased rapidly in the Early Middle Hungarian period. This is probably explained by the fact that the functional load of these elements as markers of distributivity has decreased in the very same period due to (i) the emergence of bona fide universal quantifiers which are inherently distributive (Szabolcsi 1997 a.o.) and (ii) the reinterpretation of the floating quantifier mind ‘all’ as inherently distributive (as argued by Bende-Farkas 2019, mind ‘all’ has not been distributive in Old Hungarian, however, it is undisputably distributive in Modern Hungarian).
There are altogether 95 instances of egy-enkéd / fej-enkéd in the corpus, and 18 of these are cases where there is also a floating quantifier, such as in (23) above, reproduced here as (50):
| [az | üdvözültek] | igen | méltóságossok: | mert | istennek | mind | fejenkéd |
| the | beatified | very | dignified | because | God.dat | all | head-plact |
| leányi | és | fiai | |||||
| daughter.3sg.pl | and | son.3sg.pl | |||||
| ‘[the saints] possess a high dignity, as they are all the daughters and sons of God’ | |||||||
| (Sándor C., first quarter of the 16th c.) | |||||||
Once mind had been reinterpreted as inherently distributive, the distributive import of one-pluractionals became redundant.
3.3 The possible reinterpretation -nkéd as distributive operator in minden N-nkéd constructions
Sentences such as (51) below represent a challenge to our account. Consider:
| de | ez | áldozatokban | minden | esztendő-nként | emlékezet | leszen |
| but | this | sacrifices.in | every | year-plact | remembrance | will.be |
| az | bűnökről | |||||
| the | sins.from | |||||
| (Sylvester Bible, 1541, Heb 10:3) | ||||||
The suffix -nkéd typically applies to bare nominals or numerals, and vanishingly rarely to adjective+noun combinations (such as apró darab-onkéd ‘tiny piece-plact’), to adjectives (such as apró-nkéd ‘tiny-plact’) and to possessed nouns (nemzetség-ük-önkéd ‘tribe-3sg-plact’). However, there are altogether 20 cases (out of a total of 724 -nkéd-constructions) where -nkéd applies to a universal quantifier+noun sequence: minden nap-onkéd ‘every day-plact’ (8), minden esztendő-nkéd ‘every year-plact’ (6), minden város-onkéd ‘every town-plact’ (2), minden gyülekezet-enkéd ‘every congregation-plact’ (1), minden íz-enkéd ‘every bit-plact’ (1), minden ország-onkéd ‘every country-plact’ (1), minden sereg-enkéd ‘every group-plact’ (1).
Minden+N is of a higher type (minimally <<e,t>,t>) than N (<e,t>), which makes the integration of these instances into our general account non-trivial. At least three possible accounts present themselves: (i) the type-lowering account, (ii) the translation interference account and (iii) the -nkéd as distributivity marker account.
It has been known for a long time that in Hungarian (and other languages as well), universal quantifier phrases can be interpreted at a lower type instead of or in addition to their ‘native’ type (Szabolcsi 1997), so that for example, minden város ‘every town’ can be interpreted as denoting its own maximal witness set (in essence, having the same denotation as bare város ‘town’). It is thus possible that a phrase such as minden város ‘every town’ is type-ambiguous, and when combining with -nkéd, a suffix that only accepts operands of type <e,t>, it is this lower type that is being activated. However, this account leaves open the question as to why exactly this happens: if [[N-nkéd]] = [[minden N-nkéd]], why use the more complex expression in the first place?
This is where considering the potential interference of translation is relevant. As we will see below in the section on New Testament concordances, it is in general not the case that instances of -nkéd are the results of overly faithful translations of the original. However, specifically in case of minden N-nkéd, the influence of the original seems to be present. Consider first the texts where minden N-nkéd is attested (Table 2).
Attestations of minden N-nkéd
| Text | Date | Count | Translation of |
| Érdy C. | 1526 | 4 | Legenda Aurea |
| Apor C. | 1485 | 1 | Bible |
| Jordánszky C. | 1516–1519 | 3 | Bible |
| Sylvester Bible | 1541 | 1 | Bible |
| Károli Bible | 1590 | 3 | Bible |
| Heltai Bible | 1565 | 2 | Bible |
| Thewrewk C. | 1531 | 1 | Hortulus animae, Antidotarius animae |
| Érsekújvár C. | 1529–1531 | 2 | Legenda Aurea, Gesta Romanorum a.o. |
| Peer C. | 16C1Q | 1 | ? |
| Letters | 1540 | 2 | none |
It is striking that the vast majority of the texts are translations. Focusing on New Testament concordances (see the electronic supplement), it is also clear that in every case where one or more of the translations has minden N-kéd, the Greek or the Latin original contains an element corresponding to every (Table 3).
Concordances of minden N-nkéd
| Locus | minden N-nkéd | Transl. | Greek | Vulgate |
| Acts 14:23 | minden gyülekezet-enkéd every congregation-plact | Heltai | κατ’ ἐκκλησίαν thru congr.acc | per singulas ecclesias thru each.pl.acc congr.pl.acc |
| Acts 15:36 | minden város-onkéd every city-plact | Jord., Heltai | κατὰ πόλιν πᾶσαν thru city.acc every.acc | per universas civitates thru all.pl.acc city.pl.acc |
| Acts 20:23 | minden város-onkéd every city-plact | Károli | κατὰ πόλιν thru city.acc | per omnes civitates thru every.pl.acc city.pl.acc |
| Lk 2:41 | minden esztendő-nkéd every year-plact | Károli | κατ’ ἐνιαυτὸν thru year.acc | per omnes annos thru every.pl.acc year.pl.acc |
| Heb 10:3 | minden esztendő-nkéd every year-plact | Jord., Sylvester, Károli | κατ’ ἐνιαυτὸν thru year.acc | per singulos annos thru each.pl.acc year.pl.acc |
In contrast, note e.g. that καθ’ ἡμέραν / cotidie ‘daily, every day’, which crucially lack an element corresponding to every, is never translated as minden naponkéd ‘every day-plact’. The only attested translations in the New Testament are naponkéd (25) and minden nap(on) (38) and minden kor(on) (1).
This provides relatively strong (if circumstantial) evidence that minden N-nkéd is probably an interference from translation, made possible by the availability of the type lowering strategy discussed above.
A possible alternative to this type lowering account would be to assume that minden N-nkéd expressions are bona fide universal quantifier phrases, and -nkéd has the function of a distributive suffix. Such an account would not lack plausibility: note e.g. that in Heb 10:3, where Jordánszky, Sylvester and Károli apply minden esztendő-nkéd ‘every year-plact’, Káldi uses minden-ik esztendőben ‘every-dist year.ine’, i.e. the distributive suffix -ik ‘dist’. The low frequency of the construction and the fact that a purported -nkéd distributive suffix has no trace elsewhere, however, disfavours this analysis.
4 Excursus: the effect of translation, or rather, lack of (-nkéd ≠ κατά ≠ per)
The electronic supplement contains a detailed tabulation of all the New Testament locuses where at least on of the translations contained an N-nkéd expression, together with the Greek original and the Vulgate version. A comparison of these makes it clear that -nkéd-expressions are emphatically not verbatim translations of similarly constructed Greek or Latin originals (κατά- and per-expressions). By way of example, consider the locuses where at least one translator used naponkéd ‘day-plact’.14
Translations and Greek originals
 |
 |
While καθ’ ἡμέραν ‘through day.acc.f.sg’ looks similar to nap-onkéd ‘day-plact’; it is not the case that καθ’ ἡμέραν is mechanically translated into nap-onkéd, in fact, every translator applies minden nap-on ‘every day-on’as an alternative strategy. It is also not the case that nap-onkéd is used exclusively as a translation of καθ’ ἡμέραν. (Considering the potential effect of the Vulgate, it can be pointed out that the main counterpart of naponkéd is cotidie ‘daily’, which lacks any element potentially corresponding to -nkéd.)
The tenuousness of a strict structural correspondence between the suffix -nkéd, and the prepositions κατά ‘through’ and per ‘throughout’ is further underlined by the fact that kor-onkéd ‘always’ is a translation of διὰ παντός ‘always, lit. through all.gen.m.sg’ (4), ἀεὶ ‘always, lit. age.loc.m.sg’ (2) and πάντοτε ‘always, lit. every sometimes’ (1), none of which contains any trace of κατά. While the Latin counterpart, semper ‘always’, diachronically contains the element -per (sem(el) ‘once’ + per ‘throughout’), it is dubious that this was transparent to the translator (note that the form sem is not attested in Classical Latin, -per is suffixal and not prepositional, and the meaning is also not particularly transparent (‘once throughout’ -> ‘always’).
Similarly, egy-enkéd ‘one-plact’ and fej-enkéd ‘head-plact’ also serve as the translation of a plethora of original expressions, many of which do not contain κατά or per at all (see electronic supplement).
And finally, note that that many of the texts containing -nkéd-expressions are not translations at all.15
5 Alternative expressions of pluractionality in Old Hungarian
-nkéd-suffixation was by no means the only way to express pluractionality in Old Hungarian. Competing alternatives include:
N-ról N-ra ‘from N to N’:
| nap-ról | nap-ra | az | igaz | lelket | álnokságos | (Jordánszky C., 1516–19) |
| day-from | day-onto | the | true | spirit.acc | deceitful | |
| téteményekkel kínozzák | vala | |||||
| act.pl.ins torment.3pl | be.pst.3sg | |||||
| nap-onként | hamis | cselekedetekkel | (Káldi Bible, 1626) | ||
| day-plact | fake | act.pl.ins | |||
| kínozzák | vala | az | igaz | lelket | |
| torment.3pl | be.pst.3sg | the | true | spirit.acc | |
| ‘From day to day, they vexed the righteous soul with unlawful deeds.’ | |||||
| (2Pt 2:8)16 | |||||
N szerte ‘across N’
| [nem] | szűnnek | vala | meg […] | hirdetni | (Sylvester Bible, 1541) |
| not | cease.3pl | be.pst.3sg | prt | proclaim.inf | |
| az | Jézus Krisztust | az | templomban | és | ház-szerte. |
| the | Jesus Christ.acc | the | temple.ine | and | house-across |
| nem | szűnnek | vala | meg | a | Jézus | (Heltai Bible, 1565) | ||
| not | cease.3pl | be.pst.3sg | prt | the | Jesus | |||
| Krisztusnak […] | hirdetésétől | a | templomba | és | ház-onként. | |||
| Christ.dat | proclamation.3sg.abl | the | temple.ine | and | house-plact | |||
| ‘They did not cease to preach Jesus Christ in the temple and from house to house.’ | ||||||||
| (Acts 5:42)17 | ||||||||
N-ek szerte ‘across N-pl’
| ennak okáért | hagyálak | tégedet | Krétában, | (Sylvester Bible, 1541) |
| therefore | leave.pst.1sg>2 | you.acc | Crete.ine | |
| hogy […] | városok | szerte | püspököket | válassz |
| that | town.pl | across | bishop.pl.acc | elect.subj.2sg |
| ezért | hagyálak | tégedet | Krétában, | (Heltai Bible, 1565) |
| therefore | leave.pst.1sg>2 | you.acc | Crete.ine | |
| hogy […] | város-onként | véneket | rendelnél | |
| that | town-plact | elder.pl.acc | order.cond.2sg | |
| ‘I left you in Crete for this purpose, that you ordain elders town-by-town.’ | ||||
| (Tit 1:5)18 | ||||
N szerint ‘according to N’
| köszönj | te | is | mi | barátainknak | (Jordánszky C., 1516–19) | |
| greet.imp.2sg | you | too | 1pl | friend.poss.pl.1pl.dat | ||
| személy | szerint | |||||
| person | according.to | |||||
| köszöntsed | a | mi | barátainkat | (Károli Bible, 1590) |
| greet.imp.2sg | the | 1pl | friend.poss.pl.1pl.acc | |
| fej-enként | ||||
| head-plact | ||||
| ‘You should likewise greet our friends one-by-one.’ | ||||
| (3Jn 1:15)19 | ||||
N-ek szerint ‘according to N-pl’
| ő | utat | teszen | vala | (Munich C., 1466) | |
| 3sg | road.acc | make.3sg | be.past.3sg | ||
| városok | és | kastélyok | szerint | prédikálván […] | |
| town.pl | and | village.pl | according.to | preach.part |
| ő | jár | vala | város-onként | (Károli Bible, 1590) |
| 3sg | go.3sg | be.past.3sg | town-plact | |
| és | falu-nként, | prédikálván […] | ||
| and | village-plact | preach.part | ||
| ‘He went from town to town and from village to village, preaching […]’ | ||||
| (Lk 8:1)20 | ||||
reduplication: Num (és) Num ‘Num (and) Num’
| hívá | a | tizenkettőt, | és | kezdé | (Munich C., 1466) |
| call.pst.3sg | the | twelve.acc | and | begin.pst.3sg | |
| azokat | ereszteni | kett-en | és | kett-en | |
| those.acc | let.inf | two-sup | and | two-sup |
| előhívá | az | ő | tizenkét | tanítványit, | (Pesti Bible, 1536) |
| prt.call.pst.3sg | the | 3sg | twelve | disciple.poss.pl.3sg.acc | |
| kezdé | őket | kibocsátani | kettő-nként | ||
| begin.pst.3sg | 3pl.acc | prt.emit.inf | two-plact | ||
| ‘He called the twelve apostles and sent them on their ways two-by-two.’ | |||||
| (Mk 6:7)21 | |||||
Providing a more detailed map of all things pluractional in Old Hungarian is beyond the scope of this paper.
6 The modern landscape: fragmentation
While our focus in this paper has been Late Old to Early Middle Hungarian, we offer a few preliminary descriptive remarks on the contour of N(um)-nkéd expressions in Modern Hungarian. For a formal semantic analysis of -nkéd22 in Modern Hungarian, we refer the reader to the recent work of Coppock (2024).
6.1 Time -nkéd expressions
N-nkéd here is idiosyncratically, lexically constrained, in competition with the suffix -nta (a pluractional suffix limited to time expressions):
| a. | (három) másodperc-enként | |
| / (három) *másodperc-ente | ‘second by second, every 3 seconds’ | |
| b. | (három) perc-enként / (három) *perc-ente | ‘minute by minute, every 3 minutes’ |
| c. | (három) órá-nként / (három) *órá-nta | ‘hour by hour, every 3 hours’ |
| d. | (három) ?nap-onként / (három) nap-onta | ‘day by day, every 3 days’ |
| e. | (három) *het-enként / (három) het-ente | ‘week by week, every 3 weeks’ |
| f. | (három) *hav-onként / (három) hav-onta | ‘month by month, every 3 months’ |
| g. | (három) év-enként / (három) év-ente | ‘year by year, every 3 years’ |
Consider also times of day:
| a. | reggel-enként / reggel-ente | ‘morning by morning’ |
| b. | esté-nként / *esté-nte | ‘night by night’ |
| c. | délután-onként / *délután-onta | ‘afternoon by afternoon’ |
| d. | hétfő-nként / *hétfő-nte | ‘each Monday’ |
| e. | hétvégé-nként / *hétvégé-nte | ‘weekend by weekend’ |
| f. | *nyar-anként / ?nyar-anta | ‘each summer’ |
| g. | *ősz-önként / *ősz-önte | ‘each autumn’ |
Kor-onként survives, but in the changed sense of kor as a longer period or epoch:
| változtattunk […] | bizonyos | szövegeken […] | ||
| change.pst.1pl | certain | text.pl.sup | ||
| hiszen | minden | darab | bizonyos | koronként |
| since | every | piece | certain | period-plact |
| meg | kell, | hogy | újuljon. | |
| prt | must.3sg | that | renew.subj.3sg | |
| ‘We changed certain parts of the text as every play has to be renewed after a considerable time has passed.’ | ||||
6.2 Location -nkéd expressions
These appear to be somewhat limited23 in Modern Hungarian:
| Régen | a | tejesember | ház-ról | ház-ra | járt. |
| earlier | the | milkman | house-ela | house-sub | go.past.3sg |
| *Régen | a | tejesember | ház-anként | járt. |
| earlier | the | milkman | house-plact | go.past.3sg |
| ‘The milkman used to go from house to house.’ | ||||
| Falu-ról | falu-ra | változik | a | részvételi | arány. |
| village-ela | village-sub | change.3sg | the | participation | rate |
| ?Falv-anként | változik | a | részvételi | arány. |
| village-plact | change.3sg | the | participation | rate |
They are freely available with unit expressions:
| Kilométer-enként | / | 100 méter-enként | állt | egy | rendőr. |
| kilometre-plact | 100 meter-plact | stand.pst.3sg | a | police.officer | |
| ‘A police officer was stationed at every kilometre / 100 metres.’ | |||||
| *Sark-onként | / | *Ház-anként | állt | egy | rendőr. |
| corner-plact | house-plact | stand.pst.3sg | a | police.officer | |
| intended: ‘A policeman was stationed at every corner / house.’ | |||||
6.3 Part-whole -nkéd expressions
These are freely available in Modern Hungarian:
| Ház-anként | bontot-t-ák | le | a | lakótelepet. |
| house-plact | dismantle.pst.3pl | prt | the | estate.acc |
| ‘The housing estate was dismantled house by house.’ | ||||
| Szál-anként | fogom | kitépni | a | maradék | hajamat. |
| strand-plact | will.1sg | prt.tear.inf | the | remaining | hair.1sg.acc |
| ‘I will tear out my remaining hair strand by strand.’ | |||||
| Versszak-onként | tanultam | meg | a | memoritert. |
| verse-plact | learn.pst.1sg | prt | the | poem |
| ‘I learnt the poem verse by verse.’ | ||||
6.4 Participant-related subset-set -nkéd-expressions
These are possible but more limited than in Old Hungarian. (Possibly because the floating quantifier mind ‘all’ and universal quantifiers of the minden-paradigm are inherently distributive so the functional load of egy-enkéd ‘one-plact’ is lower.)
| Az | óvodások | csoport-onként | levonultak | az | udvarra. |
| the | pre-schooler.pl | group-plact | prt.march.pst.3pl | the | court.sub |
| ‘The pre-schoolers walked down to the yard in groups.’ | |||||
| Poirot | egy-enként | elbeszélgetett | a | gyanúsítottakkal. |
| Poiro | one-plact | prt.talk.pst.3sg | the | suspect.pl.ins |
| ‘Poirot had a chat with the suspects one-by-one.’ | ||||
Participant-related subset-set -nkéd-expressions do not work very well with states:
| fej-enként […] | egymás | másikunknak | tagjai | vagyunk |
| head-plact | each.other | other.1pl.dat | member.poss.pl.3sg | be.1pl |
| ‘We are each of us members of one another.’ (Sylvester, Rom 12:5) | ||||
| *fej-enként / | *egy-enként | egymás | tagjai | vagyunk | (Mod. H.) | |
| head-plact | one-plact | each.other | member.poss.pl.3sg | be.1pl | ||
| intended: ‘We are each of us member of one another.’ | ||||||
| mind | egymás | tagjai | vagyunk | (Mod. H.) |
| all | each.other | member.poss.pl.3sg | be.1pl | |
| ‘We, every one of us, are members of one another.’ | ||||
In events, there appears to be an asymmetry: -nkéd works fairly well with themes but not with agents (though the presence of a distributive universal makes the appearance of a pluractional somewhat marked, cf. 68b):
| ennek utánna | űtőlök | fej-enkéd | búcsút | vőn | (Kazinczy C.) |
| thereafter | 3pl.ela | head-plact | farewell.acc | take.pst.3sg |
| ezután | egy-enként | elbúcsúzott | tőlük | (Modern H.) |
| thereafter | one-plact | prt.bid.farewell.pst.3sg | 3pl.ela | |
| ‘After this, he said farewell to them each.’ | ||||
| mindezekről | nem | kell | mostan | szólni | egy-enként | (Sylvester, Heb 9:5) |
| all.this.pl.ela | not | must.3sg | now | talk.inf one-plact | ||
| mindezekről | nem | kell | most | ?egy-enként | beszélni | (Modern H.) |
| all.this.pl.ela | not | must.3sg | now | one-plact | talk.inf | |
| ‘There is no need to talk about each of these things.’ | ||||||
| melyek | ha | egy-enként | megírattatnának | (Heltai Jn 21:25) |
| which.pl | if | one-plact | prt.write.pass.cond.3pl |
| amiket | ha | egy-enként | megírnának | (Modern H.) |
| what.acc | if | one-plact | prt.write.cond.3pl | |
| ‘which, if each of them was written down, …’ | ||||
| ti | is | egy-enként | azt | műveljétek | (Heltai, Ef 5:33) |
| 2pl | too | one-plact | that.acc | do.imp.2pl | |
| ‘therefore each of you should do this’ | |||||
| *ti | is | egy-enként | tegyétek | ezt | (Modern H.) |
| you | too | one-plact | do.imp.2pl | this.acc | |
| intended: ‘Therefore, each of you should do this.’ | |||||
The intended meaning of (70a) can be most naturally rendered using the distributive floating universal quantifier mind ‘all’.
6.5 Rate phrases
Strikingly, while in Old Hungarian, fej-enkéd ‘head-plact’ was a synonym of egy-enkéd ‘one-plact’, in Modern Hungarian, it is reserved for so-called rate phrases:
| A | katonák | kaptak | fej-enként | három | tucat | biztosítótűt. |
| the | soldier.pl | get.pst.3pl | head-plact | three | dozen | safety.pin.acc |
| ‘The soldiers received 3 dozen safety pins each.’ | ||||||
| Mindenki | kapott | fej-enként | három | szál | rózsát. |
| everyone | receive.pst.3pl | head-plact | three | piece | rose.acc |
| ‘Everyone received three roses per person.’ | |||||
This interpretation is available with locations as well as collective nouns:
| Szobánként / | tanszék-enként | beállítottunk | két | vázát. |
| room-plact | department-plact | prt.install.pst.1pl | two | vase.acc |
| ‘We installed two vases per room / department.’ | ||||
In the Old Hungarian Corpus, there are altogether 4 instances where a rate phrase interpretation of fej-enkéd is possible:
| fejenként | egy-egy | napi | művet | adának | őnekik | (Sylvester, Mt 20:9) |
| head-plact | one-one | daily | work.acc | give.pst.3pl | 3pl.dat | |
| ‘They gave them each one day's work.’ | ||||||
| azok | is | fejenként | tíz | pénzt | vőnek. | (Károli, Mt 20:10) |
| those | too | head-plact | ten | money.acc take.pst.3pl | ||
| ‘They too received 10 denarii each.’ | ||||||
| egyenként | az | kapuk | egy-egy | drágaköből | vannak | (Károli, Acts 21:21) |
| one-plact | the | door.pl | one-one | gem.ela | be.3pl | |
| ‘each gate is made of one (kind of) gem’ | ||||||
| Júdásnak | nemzetéből | fej-enkéd | és | ház-ankéd | (Jordánszky, Num 26:22) | ||
| Judas.dat | nation.ela | head-plact | and | house-plact | |||
| meg | számozván | lének | hetven | hat | ezeren | öt | százan |
| prt | count.part | be.pst.3pl | seventy | six | thousand.sup | five | hundred.sup |
| ‘From the tribe of Judah, 76.500 were enumerated by head and by house.’ | |||||||
Each case can be analysed in terms of our general account for -nkéd.
6.6 A complication: non-surjective pluractionals in Modern Hungarian?
In Modern Hungarian, some N-nkéd expressions receive a quasi-existential adverbial reading:
| Idő-nként | esni | fog. |
| time-plact | rain.inf | will.3sg |
| ✓ | ‘It will rain now and then.’ | |
| ✗ | ‘It will always rain.’ | |
| Hely-enként | élénk | lesz | a | nyugatias | szél. |
| place-plact | vivid | be.fut.3sg | the | westerly | wind |
| ✓ | ‘In some places, the west wind will be strong.’ | ||||
| ✗ | ‘Everywhere, the west wind will be strong.’ | ||||
| Az | idősek | eset-enként | félreértik | a | számukra |
| the | elderly.pl | case-plact | misunderstand.3pl | the | benefit.3pl.sub |
| kijelölt | kijárási | időszak | lényegét. | ||
| designated | curfew | period | essence.3sg.acc | ||
| ✓ | ‘The elderly sometimes (in some cases) misunderstand the essence of the designated curfew relaxation.’ | ||||
| ✗ | ‘The elderly always (in every case, from case to case) misunderstand the essence of the designated curfew relaxation.’ | ||||
| Alkalm-anként | a | dolgok | nem | mennek | terv | szerint.24 |
| occasion-plact | the | thing.pl | not | go.3pl | plan | according.to |
| ✓ | ‘Occasionally (in some occasions), things do not go to plan.’ | |||||
| ✗ | ‘Always (in all occasions, from occasion to occasion), it is the case that things do not go to plan.’ | |||||
There are two things to note here. First, this phenomenon is limited to this handful of Ns. So for example, órá-nként can only mean ‘every hour, from hour to hour’, and not ‘some hours, but not others’. Second, for each of these expressions, we can easily find instances with an overt modifier expressing an irregularity of frequency:
| A | nagytestű | kutyámat | bizonyos | időnként |
| the | large.bodied | dog.1sg.acc | certain | time-plact |
| beengedem, | hogy | átmelegedjen. | ||
| prt.let.1sg | that | prt.warm.subj.3sg | ||
| ‘From time to time, I let my dog inside so that she can get warm.’ | ||||
| Bizonyos | időköz-önként, | 5–8 | év-ente |
| certain | time.interval-plact | 5–8 | year-plact |
| célszerű | […] | munkahelyet | váltani. |
| expedient | workplace.acc | change.inf | |
| ‘After a certain time has elapsed, every 5 to 8 years, it makes sense to get a new job.’ | |||
| A | mérnökök | a | plafonra | bizonyos | hely-enként |
| the | engineer.pl | the | ceiling.sub | certain | place-plact |
| korongokat | is | felragasztottak. | |||
| disc.pl.acc | too | prt.glue.pst.3pl | |||
| ‘In some places, the engineers fastened discs to the ceiling.’ | |||||
| Egyes | helyenként | még | 80 km/h−1 | körüli | széllökés | is | várható. | |
| certain | place-plact | even | 80 km/h−1 | around | wind.gust | too | expected | |
| ‘In some places, as strong as 80 km/h−1 wind gushes are expected.’ | ||||||||
| Bizonyos | esetenként | előfordul | olyan, | hogy | az | ENTERSys | program |
| certain | case-plact | happen.3sg | such | that | the | ENTERSys | programme |
| elküldi | a | számlát, | de | valamiért | nem | kapunk | vissza |
| prt.send.3sg | the | invoice.acc | but | for.some.reason | not | get | prt |
| választ. | |||||||
| answer.acc | |||||||
| ‘In some cases, it happens that the ENTERSys programme sends out the invoice but for some reason, we do not get an answer back.’ | |||||||
| Létezik | olyan | beállítás, | hogy | ne | mutassa | |
| exist.3sg | such | setting | that | not | show.subj.3sg | |
| mindig | az | olajszintet, | csak | bizonyos | alkalmanként? | |
| always | the | oil.level.acc | only | certain | occasion-plact | |
| ‘Is there a setting where it does not always show the oil level, only sometimes? | ||||||
Note that such overt modification of -nkéd-expressions is productive in general:
| Órán-ként | ellenőrizni | kell | a | hőmérsékletet. |
| hour-plact | check.inf | must.3sg | the | temperature.acc |
| Néhány | órán-ként | ellenőrizni | kell | a | hőmérsékletet. |
| some | hour-plact | check.inf | must.3sg | the | temperature.acc |
So the most straightforward analysis is that időnként/helyenként/esetenként/alkalmanként have incorporated a silent modifier expressing irregularity, which can optionally be spelled out overtly. This is more economical than the alternative, the stipulation that there exists a non-surjective -nkéd2.
7 Conclusion
Based on a detailed exploration of evidence from the Old Hungarian corpus, we argued that N(um)-nkéd pluractional adverbials in Late Old and Early Middle Hungarian were (i) mereological-only, (ii) they could be associated with the agent, theme, time or location of the eventuality, (iii) they modified states as well as events and (iv) they could not instantiate pluractional comparisons across substates. These findings are an important addition to the emerging cross-linguistic typology of pluractional adverbials, especially in terms of the mereological-scalar dichotomy: in addition to (i) context and (ii) the type of the N(um)-denotation, (iii) the morphosyntactic makeup of the pluractional also has to be taken into account. Adopting a diachronic approach also enabled us to shed light on a somewhat neglected aspect of pluractional adverbials: their functional load, especially in terms of the division of labour vis-à-vis universal quantifiers (‘day-by-day’ vs. ‘every day’) and distributive operators (‘all the boys one-by-one’ vs. ‘each boy’). By observing changes playing out in the Late Old Hungarian to Early Middle Hungarian as evidenced in corpora, we showed that the development and spread of bona fide universal quantifiers and of the partitive-distributive suffix -ik indeed happened in tandem with a sharp reduction of the frequency of the relevant types of pluractional adverbials.
Conflict of Interest
Tamás Halm is a member of the Editorial Board of the journal and was not involved in the review process in any capacity.
Acknowledgements
We wish to thank our colleagues in the Implications of endangered Uralic languages for syntactic theory and the history of Hungarian project, as well as two anonymous reviewers and the editors for helpful comments and advice. The research reported here was supported by Grant NKFIH 129921 of the National Research, Development and Innovation Office of Hungary (Halm and Bende-Farkas), the Bolyai scholarship of the Hungarian Academy of Sciences and the Bolyai Plus scholarship of the New National Excellence Programme of the Ministry of Innovation and Technology of Hungary (Halm). The present work was also supported by the Faculty of Humanities and Social Sciences of Pázmány Péter Catholic University in the frame of project no. PPKE-BTK-KUT-23-2 (Halm). The first manuscript of this paper was prepared by the second author (Bende-Farkas). The substantially revised and augmented second and final version of the paper was prepared and submitted by the first author (Halm), with the prior agreement of the second author (Bende-Farkas).
References
Beck, Sigrid. 2012. Pluractional comparisons. Linguistics & Philosophy 35. 57–110.
Beck, Sigrid. 2021. Multiple events and ‘N preposition N’. In P. C. Hofherr and J. Doetjes (eds.) The Oxford handbook of grammatical number. Oxford: Oxford University Press.
Beck, Sigrid and Remus Gergel. 2015. The diachronic semantics of English again. Natural Language Semantics 23. 157–203.
Beck, Sigrid and Arnim von Stechow. 2007. Pluractional adverbials. Journal of Semantics 24(3). 215–254.
Bende-Farkas, Ágnes. 2015. The landscape of universal quantification in Old Hungarian. Acta Linguistica Hungarica 62(3). 223–261.
Bende-Farkas, Ágnes. 2019. The semantics of Old Hungarian floating mind ‘all’. Talk at Formal Diachronic Semantics 4, Ohio State University, Columbus, OH, November 15–16, 2019.
Braginsky, Pavel and Susan Rothstein. 2008. Vendlerian classes and the Russian aspectual system. Journal of Slavic Linguistics 16(1). 3–55.
Brasoveanu, Adrian and Robert Henderson. 2009. Varieties of distributivity: ‘One by one’ vs ‘each’. Semantics and Linguistic Theory, Vol. 19. 55–72.
Champollion, Lucas. 2017. Parts of a whole: Distributivity as a bridge between aspect and measurement. Oxford University Press.
Coppock, Elizabeth. 2024. Division and distributivity: The case of Hungarian -nként. Talk given at the Research Centre for Linguistics, Budapest, May 23, 2024.
Egedi, Barbara. 2014. A latin forrásszövegek hatása az ómagyar szintaktikai jelenségekre [The influence of Latin originals on syntactic phenomena in Old Hungarian]. In K. É. Kiss (ed.) Magyar generatív történeti mondattan [A generative diachronic syntax of Hungarian]. Budapest: Akadémiai Kiadó.
É. Kiss, Katalin and Orsolya Tánczos. 2018. From possessor agreement to object marking in the evolution of the Udmurt -jez suffix: A grammaticalization approach to morpheme syncretism. Language. 94(4). 733–757.
Henderson, Robert. 2013. Quantizing scalar change. Semantics and Linguistic Theory, Vol. 23. 473–492.
Simon, Eszter. 2014. Corpus building from Old Hungarian codices. In K. É. Kiss (ed.) The evolution of functional left peripheries in Hungarian syntax. Oxford: Oxford University Press.
Simon, Eszter and Bálint Sass. 2012. Nyelvtechnológia és kulturális örökség, avagy korpuszépítés ómagyar kódexekből [Language technology and cultural heritage: creating corpora from Old Hungarian codices]. Általános Nyelvészeti Tanulmányok XXIV. 243–264.
Szabolcsi, Anna. 1997. Strategies for scope taking. In A. Szabolcsi (ed.) Ways of scope taking. Dordrecht: Springer.
Vlášková, Lucia and Mojmír Dočekal. 2020. Scalar vs. mereological conceptualizations of the N-BY-N and NUM-BY-NUM adverbials. Proceedings of the Joint Workshop on Multiword Expressions and Electronic Lexicons. 27–31.
Wu, Nairan. 2023. The semantics of N-by-N adverbials. Proceedings of the Linguistic Society of America 8(1). 5536.
8. Appendix
8.1. Texts without any -nkéd-attestations
It is not surprising that most of the texts in the Miscellaneous Old Hungarian Texts section25 of the Old Hungarian Corpus lack any -nkéd-attestations: these are very short texts, often just a couple of lines or just a single line.
Focusing on the Old Hungarian Codices section, we can observe that 35 out of 47 have -nkéd-attestations. Of the 12 that lack -nkéd-attestations:
most of them are relatively short (Birk C., Bod C., Gyöngyösi C., Christina Legend, Pressburg C.), or very short (Máriabesnyő Fragment, Miskolc Fragment, Piry Membrane, Simor C.)
or belong to the Hussite Bible (Vienna Codex and Munich Codex): while the codices themselves date from the middle third of the 15th century and 1466, respectively, the texts are believed to date from the first half of the 15th century
or are old: the Jókai C. (around 1440, original may date from around 1370)
Two tentative explanations can be offered. It is possible that we are witnessing dialectal variation: the translators of the Hussite Bible conceivably spoke an -nkéd-less dialect. (Such an explanation is complicated by the fact that the Apor Codex, which also includes parts of the Hussite Bible, contains a single instance of naponkéd.)26 Alternatively, it is possible that -nkéd was in general a 15th-c. development and this is why older texts (stemming from the first half of the 15th century or the second half of the 14th century lack it).
8.2. The general picture
Frequencies (corpus size: number of tokens, originals, without punctuation marks)
| Period/Text | Time | Location | Subset-Set | Part-Whole | Total | Corpus size | Rel. frequency |
| pre-1476 | 134 575 | 0.000% | |||||
| 1476–1526 | 154 | 27 | 30 | 9 | 220 | 986 633 | 0.022% |
| 1526–1540 | 308 | 12 | 29 | 12 | 361 | 876 508 | 0.041% |
| Sylvester (1541) | 25 | 1 | 14 | 1 | 41 | 187 694 | 0.022% |
| Heltai (1565) | 20 | 7 | 9 | 36 | 156 519 | 0.023% | |
| Károli (1590) | 5 | 7 | 19 | 31 | 162 244 | 0.019% | |
| Káldi (1626) | 8 | 6 | 20 | 34 | 176 625 | 0.019% | |
| Total | 520 | 60 | 121 | 22 | 723 | 2 680 798 |
Note first that a direct comparison of the frequency of -nkéd-expressions over the whole time period is not possible for the simple reason that in general, the texts in each line/period differ in terms of size and also in terms of composition. However, a direct comparison is possible between the four Middle Hungarian Bible translations.27 Also, it is possible to examine how the share of different types of -nkéd-expressions changes over time:
Relative frequencies across types
| Period/Text | Time | Location | Subset-Set | Part-Whole | Total |
| pre-1526 | 70% | 12% | 14% | 4% | 100% |
| 1526–1540 | 85% | 3% | 8% | 3% | 100% |
| Sylvester (1541) | 61% | 2% | 34% | 2% | 100% |
| Heltai (1565) | 56% | 19% | 25% | 100% | |
| Károli (1590) | 16% | 23% | 61% | 100% | |
| Káldi (1626) | 24% | 18% | 59% | 100% |
The following general conclusions can be drawn:
The use of -nkéd expression to distribute the subevents over time shows a decreasing tendency. (Note that in Modern Hungarian, this use is even more limited.)
The use of -nkéd expressions to distribute the subevents over subsets of affected participants shows an increasing tendency. (In Modern Hungarian, this survives, although with a more restrictive semantics.)
Part-whole was an infrequent construction throughout the period. Its absence in the Bible translations is probably accidental (note the very low figures in the Old Hungarian texts, which are considerably larger), as this use survives in Modern Hungarian.
Looking at the relative frequencies of the types of -nkéd expressions, a more fine-grained picture emerges:
Relative frequencies by type
| Period/Text | Time | Location | Subset-Set | Part-Whole | Total |
| pre-1476 | |||||
| 1476–1526 | 0.016% | 0.003% | 0.003% | 0.001% | 0.022% |
| 1526–1540 | 0.035% | 0.001% | 0.003% | 0.001% | 0.041% |
| Sylvester (1541) | 0.013% | 0.001% | 0.007% | 0.001% | 0.022% |
| Heltai (1565) | 0.013% | 0.004% | 0.006% | 0.023% | |
| Károli (1590) | 0.003% | 0.004% | 0.012% | 0.019% | |
| Káldi (1626) | 0.005% | 0.003% | 0.011% | 0.019% |
The relative frequency of -nkéd-expressions rises from the second half of the 15th century until the middle of the 16th century, and a declines from then onwards.
Subset-set -nkéd expressions form an exception to this general trend: their frequency rises throughout the entire period.
The relative frequency of time -nkéd expressions drops rapidly at the end of the 16th century.
8.3. Time -nkéd-expressions
Frequencies of time-nkéd expressions
| Period/Text | kor ‘time interval' | Szempillantás ‘blink of an eye' | óra ‘hour' | nap ‘day' | éjnap ‘night and day' | Hónap ‘month' | esztendő‘year' | total |
| 1476–1525 | 73 | 1 | 71 | 1 | 8 | 154 | ||
| 1526–1540 | 121 | 7 | 158 | 22 | 308 | |||
| Sylvester (1541) | 1 | 1 | 19 | 1 | 3 | 25 | ||
| Heltai (1565) | 1 | 16 | 1 | 2 | 20 | |||
| Károli (1590) | 2 | 3 | 5 | |||||
| Káldi (1626) | 6 | 2 | 8 | |||||
| Total | 195 | 1 | 9 | 272 | 1 | 2 | 40 | 520 |
The most striking tendency here concerns kor-onkéd ‘always’, which practically disappears in the texts after 1540, while nap-onkéd persists (although with a decreasing frequency). This might be related to meaning change affecting kor, it becoming limited to meaning ‘an extended period of time’ or ‘age of a person’.
8.4. Subset-set -nkéd expressions
Frequencies of subset-set-nkéd expressions
| Period/Text | fő ‘head' | egy ‘one' | numeral (kettő ‘two', ötven ‘fifty”, száz ‘hundred') | group noun (sereg ‘group', nemzetség ‘tribe', rész ‘group', fejedelmi ház ‘dynasty') | property (név ‘name', talentum ‘talent') | Total |
| 1476–1525 | 25 | 3 | 2 | 30 | ||
| 1526–1540 | 15 | 10 | 2 | 2 | 29 | |
| Sylvester (1541) | 5 | 6 | 3 | 14 | ||
| Heltai (1565) | 1 | 8 | 9 | |||
| Károli (1590) | 3 | 11 | 4 | 1 | 19 | |
| Káldi (1626) | 8 | 5 | 4 | 2 | 20 | |
| Total | 49 | 46 | 14 | 9 | 2 | 121 |
As we have seen above, the frequency of subset-set -nkéd expressions rises throughout the entire period. However, a closer look reveals that the fej-enkéd followed a different trajectory: its relative frequency increases until the middle of the 15th century and it starts to decrease from then onwards. This may be connected to the fact that fej-enkéd was, at some point, reinterpreted as a dedicated rate expression (a usage that survives in Modern Hungarian):
Relative frequencies of subset-set-nkéd expressions
| Period/Text | fő ‘head' | egy ‘one' | numeral (kettő ‘two', ötven ‘fifty”, száz ‘hundred') | group noun (sereg ‘group', nemzetség ‘tribe', rész ‘group', fejedelmiház ‘dynasty') | property (név ‘name', talentum ‘talent') | Total |
| 1476–1525 | 0.0025% | 0.0003% | 0.0002% | 0.0030% | ||
| 1526–1540 | 0.0017% | 0.0011% | 0.0002% | 0.0002% | 0.0033% | |
| Sylvester (1541) | 0.0027% | 0.0032% | 0.0016% | 0.0075% | ||
| Heltai (1565) | 0.0006% | 0.0051% | 0.0058% | |||
| Károli (1590) | 0.0018% | 0.0068% | 0.0025% | 0.0006% | 0.0117% | |
| Káldi (1626) | 0.0045% | 0.0028% | 0.0023% | 0.0011% | 0.0113% |
8.5. Location -nkéd-expressions
Frequency of location -nkéd expressions
| Period/Text | ház ‘house' | ajtó ‘door' | ország‘country' | tartomány‘province' | város‘town' | falu‘village' | utca‘street' | gyülekezet‘congreg- ation' | Total |
| pre-1526 | 21 | 4 | 1 | 1 | 27 | ||||
| 1526–1540 | 4 | 1 | 2 | 1 | 3 | 1 | 12 | ||
| Sylvester | 1 | 1 | |||||||
| Heltai | 4 | 2 | 1 | 7 | |||||
| Károli | 3 | 3 | 1 | 7 | |||||
| Káldi | 5 | 1 | 6 | ||||||
| Total | 38 | 5 | 2 | 1 | 10 | 1 | 2 | 1 | 60 |
Nothing much can be said here because of the low figures.
8.6. Part-whole -nkéd-expressions
Frequencies of part-whole-nkéd expressions
| Period/Text | íz ‘small body part' | tag ‘member, body part' | folt ‘bit' | apró darab ‘tiny piece' | apró ‘tiny (adjective)' | levél ‘leaf, page' | ige‘word' | Total |
| 1476–1525 | 3 | 2 | 3 | 2 | 1 | 1 | 9 | |
| 1526–1540 | 9 | 1 | 1 | 2 | 12 | |||
| Sylvester (1541) | 1 | 1 | ||||||
| Heltai (1565) | ||||||||
| Károli (1590) | ||||||||
| Káldi (1626) | ||||||||
| Total | 13 | 2 | 3 | 1 | 3 | 1 | 3 | 22 |
Nothing much can be said here because of the low figures. While the construction is not attested in most of the Middle Hungarian Bible translations, this probably does not indicate its absence in the period. Note that frequencies of this construction were always low (see Table 6), so its absence in a mid-size codex may be accidental. Note also that this construction is alive and well in Modern Hungarian.
N(um)-nkéd means that -nkéd combines with nominals as well as (cardinal) numerals.
This is reflected in the varied terminology: pluractional adverbials (Beck & von Stechow 2007), N-by-N adverbials (Henderson 2013; Wu 2023), the ‘N Preposition N’ construction (Beck 2021) a.o.
Partitioning means that (i) the subeventualities are non-overlapping and (ii) the sum (or union) of the subeventualities is equal to the eventuality.
While Beck & von Stechow (2007, 222) differentiate pluractional adverbials with a sequential interpretation from those with a divisional (i.e. non-sequential) interpretation, Wu (2023) posits that temporal sequencing is an in-built feature in the lexical entry of N(um)-by-N(um) pluractionals in English.
Here, I tried to use terms that are neutral, straightforward and empirically adequate.
This has been characterized as the participant-distributive reading by Wu (2023), a slightly misleading term as the possible range of N-denotations also includes the time and the location of the event, in addition to the participants (agent, theme and possibly other theta-roles).
For a different analysis in scalar terms, see Braginsky & Rothstein (2008).
In the corpus, -nkéd surfaces with a plethora of different spellings, because of (i) dialectal variation (e.g. the Sylvester Bible systematically has í in places where other authors/translators have é), (ii) diachronic change (-nkéd -> -nként) and (iii) lack of standardized orthography. We adapted our search strategy to this, covering a very wide range of potential spellings (see Appendix) and manually checking each hit. One potential confound was present in the form on the -ként ‘in the manner of’ suffix. In the vast majority of cases, -ként and -nkéd are trivial to tell apart. In a handful of cases, detailed analysis was needed. E.g. Munich Codex has something that might be read as minden-ként ‘by all means, in all ways’ (cf. minden-képp) or as mind-enkéd (all-plact). However, (i) since the association of -nkéd with a universal quantifier is vanishingly rare and (ii) the Munich Codex otherwise completely lacks -nkéd expressions (even in the concordances where other Bible translations have -nkéd-expressions in abundance), we analysed this as minden-ként.
Partitioning means that (i) the subeventualities are non-overlapping and (ii) the sum (or union) of the subeventualities is equal to the eventuality.
Note that in some Bible translations, nap-onkéd was sometimes used as an intensifier in the case of a growth verb (often in addition to) μᾶλλον (καὶ μᾶλλον) ‘more and more, increasingly’, typically translated as inkább ‘more’ (See Table 4 for details.)
Note e.g. that 24-7 in colloquial English seems to have undergone such a change, and acquired the meaning of ‘continuously’.
The table includes all the concordances where at least one of the Bible translations applied a temporal N-nkéd expression.
The following abbreviations are used: J = Jordánszky Codex (1516–1519), P = Pesti Bible (1536), S = Sylvester Bible (1541), H = Heltai Bible (1565), Kr = Károli Bible (1590), Kd = Káldi Bible (1626).
For a more general discussion of translation interference in Old Hungarian texts, cf. Egedi (2014).
Sylvester and Heltai have naponkéd ‘day-plact’, Károli has minden napon ‘every day-sup’. Cf. also 2Cor 4:16.
Jordánszky has minden házon ‘every house-sup’, Károli minden háznál ‘every house-ade’ and Káldi házonkéd ‘house-plact’. Cf. also Lk 8:1.
Károli and Káldi have városonkéd ‘town-plact’. Cf. also Acts 14:23, Acts 2:46, Acts 20:23 and Acts 8:3.
Sylvester has nevük szerint ‘name.poss.3pl according.to’, Heltai fejenkéd ‘head-plact’, Káldi nevenkéd ‘name-plact’. Cf. also Acts 21:19.
Jordánszky has városokon és falvakon ‘town.pl.sup and village.pl.sup’, Pesti városokon és falvakon által ‘across towns and villages’, Sylvester város és faluszerte ‘town and village across’, Heltai városok és faluk által ‘town.pl and village.pl across’, Káldi városokat és kastályokat ‘town.pl.acc and village.pl.acc’, cf. also Acts 14:23.
Jordánszky has ketten-ketten, Sylvester ketten ketten, Heltai ketten-ketten, Károli kettőnkéd, Káldi kettőnkéd. Cf. also Lk 10:1, Lk 9:14 and Mk 6:39–40.
As noted before, the pluractional suffix had many competing variants in Old Hungarian, of which only the form -nként survived into Modern Hungarian. In order to maintain conceptual clarity, we continue to refer to this suffix as -nkéd in our metalanguage even in our discussion of Modern Hungarian.
A reviewer finds sentences similar to (62b) fully acceptable and indeed, such sentences can be found in webcorpora reflecting current usage. Our take on this is that location-nkéd expressions are certainly more limited in Modern Hungarian than in Old and Middle Hungarian; however, the exact degree of this (and any dialectal variation) is still to be established by further research.
Eset-enként and alkalm-anként can also appear in rate phrases:
| Eset-enként | tízezer | forintba | kerül | a | szűrővizsgálat. |
| case-plact | ten.thousand | forint.ill | cost.3sg | the | screening |
| ‘The procedure costs ten thousands forints per case.’ | |||||
| Alkalmanként | 12 ezer | forintot | fizettem. |
| occasion-plact | 12 thousand | forint.acc | pay.pst.1sg |
| ‘I payed twelve thousand forints per occasion.’ | |||
The Apor Codex does include texts other than the Hussite Bible, however, this particular attestation is from the Hussite Bible part (the Psalms).
In the data below, only the New Testament part of the Bible translations is taken into account: for the simple reason that the Old Hungarian Corpus only includes the New Testament part of these translations.
