ArticlePDF Available

A questionnaire-based study of impersonalization in Romanian and English With special attention to passivization

Authors:

Abstract and Figures

This paper is the first contrastive study of impersonalization in Romanian and English. Taking an acceptability judgment approach, we describe the functional potential in all impersonal uses of not only the pronouns ‘one’, ‘you’ and ‘they’ but also the lesser studied passive. We find inter alia: a similar division of labor in the languages between ‘you’ and ‘they’ for contexts paraphrasable as, respectively, ‘everyone’ and ‘someone/some people’; a wider range of uses for pro-dropped ‘they’ than for its overt counterpart, as hypothesized in previous research; and a preference in English, but not Romanian, for passives to ‘they’ especially in contexts like ‘they’ve stolen my wallet!’, where the referent is entirely unidentifiable and likely to be singular. Level of identifiability and number, each of which has been suggested in a separate semantic map as necessary for capturing impersonalization, are also shown to interact, supporting a proposal to combine them in one map.
Content may be subject to copyright.
A questionnaire-based study of impersonalization in Romanian and English
With special attention to passivization
Valentin Rădulescu and Daniël Van Olmen
Lancaster University
1 Introduction
The study of impersonalization has a well-established tradition. The phenomenon, exemplified
by one in (1), is often subsumed under the functional domain of agent defocusing, which covers
a variety of constructions across languages that shift attention away from the agent in the clause
(e.g. Shibatani 1985; Myhill 1997). They include, inter alia, middle constructions and passives
like (2a) and (2b) respectively. Agent defocusing is motivated by a variety of factors. An agent
may, for instance, be demoted to an oblique when the patient is more topical (see Myhill 1997:
805), which appears to be the case in (2b) with its demonstrative-marked patient-subject. It is
also possible that an agent is omitted altogether when its reference is obvious from the context
(see Shibatani 1985: 831). Importantly, as argued by Sansò (2006) among others, such factors
affect the choice of construction in a language and languages can differ in the construction that
they use under similar conditions.
(1) One shouldn’t say things like that. (Myhill 1997: 802)
(2) a. The book sells well. (Sansò 2006: 242)
b. That book was written by Mary. (Myhill 1997: 803)
The precise conditions of agent defocusing that constitute the topic of this article are when the
agent is generic (e.g. Sansò 2006: 242-245) more or less paraphrasable as ‘everybody’ or, in
the case of (1), ‘nobody or when the agent is a particular but unknown (set of) individual(s)
(e.g. Myhill 1997: 806-822) roughly equivalent to ‘someone’ or ‘some people’, as in (3). The
constructions under investigation are one in (1) and the third person plural in (3), as well as the
second person singular in (4a) and the agentless passive in (4b), intended here as the respective
counterparts of (1) and (3).
1
And the languages of study, finally, are English and Romanian.
(3) They’ve taken the radio. (Myhill 1997: 802)
(4) a. You shouldn’t say things like that.
b. The radio has been taken.
In the following sections, we will first describe our approach to the particular agent defocusing
conditions at issue, then introduce the four constructions mentioned above and, lastly, discuss
our interest in the two languages within this context.
1.1 Impersonalization
The phenomena in (1), (3) and (4) are not always grouped together in the literature. Kitagawa
and Lehrer (1990: 742), for example, reserve the label “impersonal” for generic cases and clas-
sify cases with “specific individuals [that] … are not identified, or identifiable, by the speaker”
as “vague”. We, however, will follow Gast and van der Auwera (2013), who observe that many
1
Unless specified otherwise, the examples are constructed by the authors.
languages employ the same construction for (1) and (3). Dutch men ‘one, they’ in (5) is a case
in point (see also Giacalone Ramat and Sansò 2007 on so-called ‘man’-pronouns).
(5) a. Dat soort ding-en moet men niet zeg-gen.
DIST.DEM.SG sort thing-PL must.SG.PRS MEN NEG say-INF
‘One shouldn’t say things like that.’
b. Men heeft de radio gestolen.
MEN have.3SG.PRS DEF.SG.M radio steal.PST.PTCP
‘They’ve taken the radio.’
The different agent defocusing conditions can thus, in their view, be considered instances of a
single domain of “impersonalization”, which they define as “the process of filling an argument
position of a predicate with a variable ranging over sets of human participants without estab-
lishing a referential link to any entity from the universe of discourse” (Gast and van der Auwera
2013: 124).
In their map of impersonal uses, given in Figure 1, Gast and van der Auwera (2013) call
contexts paraphrasable as ‘everyone’ “universal” (UNI) and contexts paraphrasable as ‘some-
one, some people’ “existential” (EXI). Based on a survey of impersonal constructions in Euro-
pean languages, the map makes additional distinctions, though.
Figure 1: Adapted version of Gast and van der Auwera’s (2013: 141) semantic map of impersonalization
In the vein of, among others, Kitagawa and Lehrer (1990: 749) and Sansò (2006: 268), it sets
apart universal uses with an “internal” (INT) perspective, like (4a), from those with an “exter-
nal” (EXT) one, like (6a). The former is characterized by Gast and van der Auwera (2013: 139)
as follows: “A ‘center of consciousness’ (e.g. the speaker or hearer) identifies, or is identified,
with the set of referents under discussion.” In (6b), for instance, the statement pertains to any-
one, the interlocutors included, who happens to be or can imagine themselves in Bali. By con-
trast, (6a) is about the set of inhabitants of the country, typically excluding the speaker and the
addressee.
(6) a. In Bali, they eat dragonflies. (Gast and van der Auwera 2013: 144)
b. In Bali, you eat dragonflies.
Further differentiation within the universal-internal domain in Figure 1 rests on the parameters
of veridicality i.e. the state of affairs is (not) assumed to be true and of modality i.e. the
non-veridical state of affairs has an/no explicit marker of possibility or necessity (see Gast and
van der Auwera 2013: 137). The first parameter distinguishes (7a)
2
from (7b) and (7c): only
2
Another example of the use presented in (7a) is (6b).
living once is presented as an unqualified fact, making this context “veridical” (VER). The
second one separates (7b) and (7c). The “non-veridicality” (NVER) of (7b) is due to the occur-
rence of the modal (MOD) shouldn’t, which puts the state of affairs of drinking and driving in
the realm of the (un)necessary. In (7c), there may be no modal marker (NMOD) but the condi-
tional clause still renders drinking sour milk hypothetical and hence non-veridical. Crucially,
each distinction is motivated by at least one impersonal construction in the languages of Europe
that can occur in one use but not in another (see Gast and van der Auwera 2013: 140-154). The
parameters have also been shown to be relevant for language-specific descriptions, though.
Laberge and Sankoff (1979: 432), for example, indicate that French prefers tu ‘you’ to on ‘one’
in conditional contexts.
(7) a. You only live once.
b. You shouldn’t drink and drive.
c. What happens if one drinks sour milk?
(all from Gast and van der Auwera 2013: 138)
The existential domain in Figure 1 displays more fine-grained uses too. Gast and van der Au-
wera (2013: 140) distinguish definite or “semi-impersonal” contexts like (8a) from indefinite
or “truly impersonal” ones like (8b) and (8c). In (8a), the referents may not be explicitly iden-
tified but the action of raising taxes itself makes them partly accessible, since only the “corpo-
rate” (COR) entity of the government can undertake it. Examples (8b) and (8c), however, are
intended as applying to a particular set of human participants whose referents are inaccessible
or, in short, unknown. The difference between (8b) and (8c) lies in number: the set of individ-
uals in (8b) is necessarily plural” (PL) whereas that in (8c) is “neutral with regard to number”
(NN) and may consist of one person or more than one person.
(8) a. They’ve raised the taxes again.
b. They’ve surrounded us.
c. They’re knocking on the door.
(all from Gast and van der Auwera 2013: 152)
Cross-linguistic support for the distinctions comes from languages where the third person plu-
ral can be employed impersonally for (8a) but not (8b) and (8c) or for (8a) as well as (8b) but
not (8c). Unlike they, German sie ‘they’, for instance, can only really occur with a plural read-
ing (Gast and van der Auwera 2013: 144; see also Siewierska and Papastathi 2011: 596).
The third person plural has actually been the focus of a second map of impersonal uses,
given in Figure 2. In their study of nine European languages, the authors, Siewierska and Pa-
pastathi (2011), recognize the same universal-external and existential-corporate uses, as in (6a)
and (8a), but they add a “speech act verb” (SAV) use and make distinctions in the indefinite or
truly impersonal existential domain based not on number (8b) versus (8c) but on a param-
eter characterizable as “(un)knownness” (note that “anaphoric” in Figure 2 stands for personal
‘they’).
Figure 2: Adapted version of Siewierska and Papastathi’s (2011: 604) semantic map for ‘they’-impersonals
The extra context, involving a speech act verb, is exemplified in (9) and marks information as
coming from some unidentified source (see also Myhill 1997: 811). Siewierska and Papastathi
(2011) regard it as a direct off-shoot from anaphoric ‘they’ in their map, unrelated to the other
impersonal uses.
3
This decision is driven by languages like French and Finnish: the third person
plural in French can serve a variety of impersonal purposes but is not acceptable in (9) whereas
its Finnish counterpart cannot be employed universal-externally or existentially but can occur
with speech act verbs (see Siewierska and Papastathi 2011: 585, 591). This separate status of
the use in (9) is the reason why Gast and van der Auwera (2013) do not include it in their map.
Still, the fact that, in many languages, markers dedicated to the expression of impersonalization
in general can appear with speech act verbs too, like Dutch men in (10), suggests that studies
of impersonalization, like the present one, should take the context into account.
(9) They say he met vampires in the Black Forest. (Siewierska and Papastathi 2011: 585)
(10) Men zeg-t dat hij in het Zwarte Woud vampier-en
MEN say-3SG.PRS COMP 3SG.M.NOM in DEF.SG.N Black Forest vampire-PL
ontmoet heeft.
meet.PST.PTCP have.3SG.PRS
‘They say he met vampires in the Black Forest.’
For the indefinite existential realm, Siewierska and Papastathi (2011) follow Cabredo Hofherr
(2006), who distinguishes three uses that vary in the type or level of (un)knownness (see Creis-
sels 2008a: 5-6 too). In a “vague” (VAG) context like (11a), the speaker knows that a particular
(group of) individual(s) carried out some action but is unable or unwilling to characterize them
(the term “vague” will henceforth be used in this sense and not in Kitagawa and Lehrer’s 1990
sense, mentioned earlier). In an “inferred” (INF) context like (11b), the speaker deduces from
the situation (a smell here) that an essentially unknown event took place and that some unspec-
ified person/people is/are responsible for it. In a “specific” (SPE) one like (11c), finally, some-
thing happens at an exact place and time, typically the speaker’s here and now in this case, a
knock on the door. They may therefore be aware of who is causing it but abstain from identi-
fying them.
(11) a. They’ve found his bike in the back of a barn.
b. They’ve been frying chips here. [I can smell it.]
c. They’re knocking on the door. [It might be Mary.]
(all from Siewierska and Papastathi 2011: 581)
3
As Siewierska and Papastathi (2011: 585) point out, it differs from the existential ones in that it is “typically not
episodic, i.e. do[es] not refer to a specific event” and it “typically cannot be substituted by someone, as the referent
of they clearly corresponds if not to the whole human race then to some group of people at a given time and place”.
These distinctions are again motivated by cross-linguistic differences, in the distribution of the
third person plural: in some languages, ‘they’ is restricted to universal-external and existential-
corporate uses; in others, it may also be existential-vague but not -inferred or -specific; and in
yet other languages, it can occur in existential-vague and -inferred but not -specific contexts or
in existential-vague and -specific but not -inferred ones. Syrian Arabic would be an example
of the last type of language and French an example of a language with universal-external, ex-
istential-corporate and existential-vague uses only for its third person plural (Cabredo Hofherr
2006: 245; see also Siewierska and Papastathi 2011: 596).
Figures 1 and 2 differ in the inclusion of the universal-internal and speech act verb uses
and in the differentiation within the indefinite existential ones. The map in Figure 3, proposed
by Van Olmen and Breed (2018a, 2018b), merges the two, to capture the “complete” range of
impersonal contexts. It appends Siewierska and Papastathi’s (2011) speech act verb use to Gast
and van der Auwera’s (2013) map and combines the former’s parameter of (un)knownness and
the latter’s parameter of number, resulting in existential-vague/inferred/specific-plural and ex-
istential-vague/inferred/specific-number-neutral uses. The reason for bringing the parameters
together is that, at least for the third person plural in West Germanic, they have been shown to
interact. In Van Olmen and Breed’s (2018b: 822-838) study of Afrikaans, Dutch and English,
the acceptability of ‘they’, as well as its usage (in a completion task), decreases from vague/in-
ferred/specific-plural to vague/inferred/specific-number-neutral contexts and from vague-plu-
ral/number-neutral to inferred-plural/number-neutral and specific-plural/number-neutral ones.
Figure 3: Van Olmen and Breed’s (2018b: 840) combined semantic map of impersonalization
The twelve impersonal uses in Figure 3 will be central to the rest of the present article, which
will hereafter employ their abbreviations.
1.2 Impersonal constructions
The constructions under examination in this study are one, the second person singular, the third
person plural and the passive. ‘One’-pronouns can be found in a variety of languages. Spanish,
with uno ‘one’, can serve as an example (e.g. Fernández 2013: 93-94) and so can English, as
(1) repeated here as (12) shows. Romanian has no such pronoun, though.
(12) One shouldn’t say things like that. (Myhill 1997: 802)
In general, ‘one’-pronouns are known to be more or less restricted to UNI-INT uses (e.g. Gast
and van der Auwera 2013: 149; Fenger 2018: 298). Yet, they can also be employed “for refer-
ence to the first person, especially if the speaker wants to represent his/her behaviour as a result
of general rules” (van der Auwera et al. 2012: 47), like the politician in (13). This extended use
is, however, beyond the scope of the present article.
(13) One doesn’t want to set quotas. One doesn’t want to set diktats, but one does want to
maintain a dialogue. (van der Auwera et al. 2012: 48)
English one is typical of formal language and has been said to invite the addressee “to identify
with the group at issue” (Haas 2018a: 174) less than you. It has also been suggested that one
tends to occur in NVER contexts (e.g. Moltmann 2010: 463-466; Rudolf 2016: 115), although
this correlation has become less pronounced over time (Haas 2018b).
The impersonal use of the second person singular is a common phenomenon in language
(e.g. Laberge and Sankoff 1979 for French; De Hoop and Tarenskeen 2015 for Dutch; Jensen
and Gregersen 2016 for Danish) and is attested in English and Romanian too, as you in (7a)
repeated here as (14) and ai ‘you have’ in (15) make clear (the overt subject pronoun tu ‘you’
only appears for, say, emphasis, contrast or disambiguation in Romanian and does not normally
allow an impersonal reading; see Pană Dindelegan 2013: 104-111).
(14) You only live once. (Gast and van der Auwera 2013: 138)
(15) Ai doar o viață.
have.2SG.PRS just INDF.F.SG life
‘You only live once.’
In the literature, considerable attention has been paid to the way in which personal ‘you’ turns
impersonal and to the question whether the difference in interpretation is semantic or pragmatic
in nature (e.g. Malamud 2012; Gast et al. 2015). What is important here is that the central role
of the addressee persists in the impersonal second person singular in that it has been argued to
essentially be limited to UNI-INT uses across languages (Gast and van der Auwera 2013: 146).
Such uses typically include the addressee among the set of referents or at least encourage them
to imagine themselves as belonging to the set. The “lingering” reference to the speaker’s inter-
locutor gives ‘you’ compared to alternative impersonal constructions a “sense of informal
camaraderie” (Kitagawa and Lehrer 1990: 752) and of “recruiting involvement, here empa-
thetic, from the addressee” (Stirling and Manderson 2011: 1583). Much has been written about
other possible effects (e.g. O’Connor 1994: 45 on distancing; Kluge 2016: 515-518 on polite-
ness) but, though interesting, they fall outside of the present article’s scope. We will also not
consider the usage of the second person singular to describe personal experiences, as in (16).
For a range of potential reasons (see Stirling and Manderson 2011: 1590-1598), the speaker
chooses you here instead of I to somehow generalize her own specific sensation.
(16) I knew the bandage was there, you could feel it and everything.
(Stirling and Manderson 2011: 1593)
What the UNI-INT part of the map in Figure 3 is concerned with for ‘you’ is different contexts
where “structural knowledge” (see Kitagawa and Lehrer 1990: 748) is described. The speaker
conveys not what happens to themselves but what holds for anyone or what commonly befalls
(or can/should befall) anyone in a situation. The fact that the parameters differentiating these
contexts, i.e. (N)VER and (N)MOD, may indeed affect the impersonal second person singular
is evident from research mentioned earlier (e.g. Laberge and Sankoff 1979 in Section 1.1).
The third person plural functions as an impersonal construction in numerous languages
(e.g. Fernández 2013: 101-104 on Spanish; Data-Bukowska 2018 on Swedish; Van Olmen and
Breed 2018a: 17-19 on Afrikaans). English and Romanian are among those languages, as they
in (3) repeated here as (17) and au ‘they have’ in (18) make clear.
(17) They’ve taken the radio. (Myhill 1997: 802)
(18) Au furat radio-ul.
have.3PL.PRS steal.PST.PTCP radio-DEF.SG.M
‘They’ve taken the radio.’
Impersonal ‘they’ is often characterized as a feature of informal language (e.g. Kitagawa and
Lehrer 1990: 746) and is generally found in non-UNI-INT contexts only (e.g. Siewierska and
Papastathi 2011; Gast and van der Auwera 2013: 144). Exceptions exist, however. According
to Cabredo Hofherr (2006: 246), in Russian and Hebrew, for instance, the third person plural
may also receive a UNI-INT interpretation. There is variation within the non-UNI-INT domain
too, of course, as amply shown in the discussion in Section 1.1. Languages diverge not only in
(not) employing ‘they’ for SAV but also in its (non-)restriction to the semi-impersonal UNI-
EXT and EXI-COR uses or if it can occur in truly impersonal EXI contexts its (in)compat-
ibility with NN readings or different forms of (un)knownness. The explanation for such differ-
ences is argued, by Siewierska and Papastathi (2011: 600) among others, to lie in the degree to
which the definite plural prototype” of ‘they’ as a personal pronoun has bleached in languages.
UNI-EXT and EXI-COR, for instance, can be regarded as close to the prototype as the set of
referents is plural and quite definite (see Siewierska and Papastathi 2011: 583). In (6a), here as
(19a), they refers to an entire group of people that in Bali makes accessible, i.e. the country’s
inhabitants. In (8a), repeated as (19b), the state of affairs itself renders the collective entity of
referents partially accessible, i.e. the government. EXI-INF and EXI-SPE, by contrast, may be
considered as further removed from the prototype. Not only is “referent identification purely
situational” (Siewierska and Papastathi 2011: 584) in (11b) and (11c), here as (19c) and (19d),
i.e. based on some smell in the former and the speaker’s here and now in the latter; they seems
NN in the examples in that it may refer to one or more than one person.
(19) a. In Bali, they eat dragonflies. (Gast and van der Auwera 2013: 144)
b. They’ve raised the taxes again. (Gast and van der Auwera 2013: 152)
c. They’ve been frying chips here. (Siewierska and Papastathi 2011: 581)
d. They’re knocking on the door. (Siewierska and Papastathi 2011: 581)
Siewierska and Papastathi (2011) also observe that, on the whole, languages with an overt third
person plural pronoun (e.g. French) exhibit a narrower range of uses, more in keeping with the
prototype, than languages without one or, put differently, with a pro-dropped one (e.g. Italian).
4
To account for these facts, they appeal to the established “relationship between morphophono-
logical reduction and semantic bleaching” (Siewierska and Papastathi 2011: 600): less reduced,
overt ‘they’ is more resistant to appearing in contexts that are not definite or plural. There are
exceptions, however. Polish, for one, is found to pattern “with the non-pro-drop languages in
apparently lacking inferred and specific 3pl IMP[ersonal]s and disfavoring vague ones” (Siew-
ierska and Papastathi 2011: 602).
The (agentless) passive is known to serve as an impersonal construction in countless lan-
guages (e.g. Shibatani 1985; Creissels 2008a: 9-10). The English one can be used in this way
too and so can the Romanian one, as in (4b) repeated here as (20) and (21).
(20) The radio has been taken.
4
Henceforth, we will use pro-drop when an impersonal pronoun is conveyed solely through verb agreement, as
in (18), and “overt for its realization as a separate word, as in (19).
(21) Radio-ul a fost furat.
radio-DEF.SG.M have.3SG.PRS be.PST.PTCP steal.PST.PTCP
‘The radio has been taken.’
The passive with (or without) an impersonal interpretation is often said to be a relatively formal
phenomenon (e.g. Kitagawa and Lehrer 1990: 746). Moreover, Sansò’s (2006: 245-263) in-
depth study of five European languages reveals that it may occur in UNI contexts but is clearly
a minority option there. Interestingly, Van Olmen and Breed (2018b: 819) independently find
that “H[uman]I[mpersonal]P[ronoun]s are very strongly preferred to other forms of imperson-
alization in universal” uses in West Germanic. No such obvious preference appears to exist for
the other contexts, however, which raises the question whether the passive is perhaps favored
more in that domain. According to Siewierska (2008: 12), this is indeed true for SAV in British
English: she claims that it is said that is more common than they say that. The passive is also
argued to be preferred more generally to the third person plural when an unidentified agent is
almost definitely a single person (Siewierska 2008: 16; cf. the NN distinction in Section 1.1.).
Yet, to our knowledge, the matter of how the passive tends to fare in languages in more partic-
ular uses such as EXI-COR or EXI-VAG-PL versus EXI-INF/SPE-PL has so far not been ex-
amined in much detail.
5
Languages may possess other impersonal constructions than those described here. ‘Man’-
pronouns, deriving from a noun meaning ‘man, human being’ and exemplified in (5) and (10),
are a case in point. Both English and Romanian lack such a pronoun, however (see Siewierska
2011). The first person plural, as in (22), is occasionally mentioned too (e.g. Siewierska 2004:
211) but seems to have a very low frequency in languages (see Van Olmen and Breed 2018b:
818 on Afrikaans, Dutch and English). It is therefore not included in the present study (see also
Kitagawa and Lehrer 1990: 744-745 on the ambiguity between the different interpretations of
we).
(22) Avem doar o viață.
have.1PL.PRS just INDF.F.SG life
‘We only live once.’
Another construction found in many languages is the reflexive impersonal (e.g. D’Allessandro
2007; Creissels 2008a: 8-9; Siewierska 2008: 18-21). It exists in Romanian, as the UNI-EXT
context with reflexive se in (23) shows, and is a conventional way of expressing impersonali-
zation in the language (e.g. Dobrovie-Sorin 1998; Pană Dindelegan 2008).
(23) În Anglia, se conduce pe stâng-a.
in England REFL drive.3SG.PRS on left-DEF.F.SG
‘In England, they drive on the left.’
It will nevertheless not be considered here (and neither will any other possible impersonal con-
structions), for the contrastive reason that it has no genuine or even approximate equivalent in
English, unlike the passive.
6
5
The factors discussed here are the ones relevant for the present study but they are not the only ones affecting the
passive’s potential use as an impersonal constructions. Languages where passivization requires the promotion of
an object to subject, for example, simply do not have the passive as an option for intransitive clauses (cf. Creissels
2008a: 9-10).
6
Note, for transparency's sake, that reflexive impersonals were included in the questionnaire given to our Roma-
nian participants but that we leave the analysis of this data for further research.
1.3 Languages
Impersonalization has attracted significant interest the last decade. Especially impersonal pro-
nouns have been the subject of much research. Recent work on English, for instance, includes
Rudolf (2016), Haas (2018a) and Hall (2020). Other Germanic languages that have been stud-
ied are Afrikaans, Danish, Dutch, German and Swedish (e.g. Coussé and van der Auwera 2012;
De Hoop and Tarenskeen 2015; Gast 2015; Jensen and Gregersen 2016; Fenger 2018; Van Ol-
men and Breed 2018a). Romance has received considerable attention too, with publications on
French (Creissels 2008b), Portuguese (Posio and Vilkuna 2013) and Spanish (Fernández 2013)
among others. Romanian, however, seems to have escaped notice.
No comprehensive account of the usage of impersonal pronouns in Romanian exists, to
our knowledge. Reference grammars such as Cojocaru (2003) and Gönczöl-Davies (2008) do
not even discuss them. Manea (2012: 70) does mention a generic-passive expression that was
used in an unspecified older Romanian”, illustrated in (24), but does not expand on it.
(24) Spun că (el) ar fi zis… (Manea 2012: 70)
say.3PL.PRS COMP 3SG 3.COND be.INF claim.PST.PTCP
‘They say that he (would have) claimed…’
Yet, spun ‘they say’ is not actually a passive. Rather, it is an instance of Siewierska and Papas-
tathi’s (2011) SAV function of the third person plural. The questions thus arise as to whether
it is still acceptable nowadays and whether Romanian ‘they’ can appear in the other UNI-EXT
and EXI uses identified for the third person plural in the cross-linguistic research. In the same
vein, Pană Dindelegan’s (2008: 147) chapter in the Romanian Academy’s grammar, about pas-
sivization and impersonalization, points out the existence of a generic second person singular
and a generic first person plural but only gives examples of the latter. It remains to be seen how
acceptable such pronouns are in the different contexts distinguished for the UNI-INT realm in
the literature on other languages.
In the present article, we will map the functional potential, i.e. the degree of (in)compat-
ibility with various uses, of Romanian ‘they’ and ‘you’ in particular (for ‘we’, see Section 1.2)
and contrast it to that of the impersonal pronouns in English. The inclusion of English allows
a comparison of Romanian to a widely known language with a typical division of labor between
a UNI-INT second person singular and a non-UNI-INT third person plural. It will also enable
us to check the findings of a previous, more limited study of the English pronouns (Van Olmen
and Breed 2018b). Besides pronominal forms of impersonalization, this article will investigate
the passive, whose usage we have argued has not yet been thoroughly investigated in light
of all the contexts in Figure 3. In short, we primarily aim to answer the following questions: (i)
what is the functional potential of the impersonal pronouns as well as the passive in Romanian
and English?; and (ii) how do the impersonal constructions and the languages compare to each
other? In addition, our study may shed more light on two other issues raised above. On the one
hand, Romanian can function as a non-Germanic testing ground for the interaction attested in
Afrikaans, Dutch and English between number and (un)knownness in the truly impersonal EXI
domain (see Section 1.1). On the other hand, comparing Romanian and English ‘they’ can serve
to check the hypothesized impact of the third person plural’s pro-dropped versus overt nature
on its usage (see Section 1.2). In other words, we also seek to answer the following secondary
questions: (i) do the parameters of number and (un)knownness interact with one another (in the
expected way) in Romanian?; and (ii) does the pro-dropped third person plural exhibit a(n ex-
pected) wider range of less definite/plural impersonal uses than its overt counterpart?
The remainder of the article is structured as follows. Section 2 will introduce our meth-
odology: the choice for an acceptability judgment questionnaire approach (2.1), the design of
our questionnaire (2.2), details about its distribution and participants (2.3) and some statistical
background (2.4). In Section 3, we will present and discuss our findings, first for the UNI-INT
uses (3.1) and then for the non-UNI-INT ones (3.2), and subsequently summarize them (3.3).
Section 4, finally, is our conclusion.
2 Methodology
7
2.1 Acceptability judgment questionnaires
Acceptability judgments are usually regarded as signs of grammaticality. Yet, Bard et al. (1996:
33) argue that they “need not be one-to-one reflections of grammaticality” since they “may be
based, for example, on estimated frequency of usage, on conformity to a prescriptive norm or
a prestigious register, or on degree of semantic and pragmatic plausibility”. This suitability in
specific semantic and pragmatic contexts is what is at play in previous acceptability judgment
questionnaires for impersonalization, such as Siewierska and Papastathi (2011) and Van Olmen
and Breed (2018a, 2018b). They ask native speakers to rate the acceptability of different con-
structions in sentences that conclude short scenarios triggering particular impersonal interpre-
tations.
This approach has a number of advantages, despite the constructed nature of the language
(a drawback that a corpus study would not suffer from). If formulated carefully, such scenarios
allow researchers to exclude personal readings of constructions like ‘you’ in advance (cf. Siew-
ierska and Papastathi 2011: 587-588 on some of the issues in distinguishing personal and im-
personal ‘they’ in corpora). Consider the UNI-INT-NVER-NMOD context in (25): the fact that
the speaker is replying to their brother and it is their cousin who is going away makes a personal
interpretation of you rather unlikely here.
(25) You and your brother are helping your female cousin pack her suitcase for her trip to
England. When you hand her the brand new raincoat that you bought her the day be-
fore, your brother asks you why you bought that in particular. You reply: “If you go to
England, it is necessary to pack a raincoat. (Van Olmen and Breed 2018a: 7)
The set-up of these questionnaires also provides more control over the parameters differentiat-
ing the various uses in Figures 1 to 3. The distinction between PL and NN contexts, for instance,
may not always be so clear in authentic language data. By contrast, one can quite easily develop
scenarios that prompt an unambiguous PL or NN reading (see Section 2.2), making it possible
to test the impact of the parameter of number accurately. Moreover, the approach enables re-
searchers to check, in a straightforward way, a construction’s potential to occur in contexts that
may not actually arise in, say, a corpus due to their general infrequency. For example, EXI uses
are known to be much less common than UNI-ones, for ‘man’-pronouns at least (e.g. Zifonun
2001 on German; Fonesca-Greber and Waugh 2003 on French). Similarly, Siewierska and Pa-
pastathi’s (2011: 590) corpus study reveals that EXI-INF and EXI-SPE cases of ‘they’ are ex-
tremely rare.
The present questionnaire differs from the earlier ones in several regards. Siewierska and
Papastathi’s (2011) is primarily concerned with the third person plural and pays little attention
to other impersonal constructions. This focus means that it only contains scenarios for the UNI-
EXT, SAV and EXI uses in Figure 2. It also does not feature contexts that combine the param-
eters of number and (un)knownness (e.g. EXI-VAG-PL, EXI-INF-NN; see Section 2.2) to test
7
The methodology described here closely resembles that of Van Olmen and Breed (2018a, 2018b).
their interaction. Furthermore, Romanian is not amongst the languages under investigation and
the findings for English are based on just” ten participants. Van Olmen and Breed’s (2018a,
2018b) questionnaires do examine all the uses in Figure 3 but only look at impersonal pronouns
in Afrikaans, Dutch and English. The present one brings in Romanian, adds the passive as one
of the constructions to be assessed and has many more English-speaking participants than Van
Olmen and Breed’s (2018b) “mere” 23.
2.2 Questionnaire design
Every part of the questionnaire starts with a short scenario that prompts a particular impersonal
reading of the test items in its final sentence. The scenario in (26) can serve as an example. It
is also representative of the other scenarios in describing fairly informal situations and in often
inviting the reader to take the perspective of the speaker. The reason for this type of scenario
is that the use of especially impersonal pronouns appears to be more typical of everyday lan-
guage (see Siewierska and Papastathi 2011: 593).
(26) UNI-EXT
Your friend is making a list of the official languages in every country. Based on your
recent traveling experience, you tell her:
In Brazil, one speaks Portuguese. 1 2 3 4 5
“In Brazil, you speak Portuguese.” 1 2 3 4 5
“In Brazil, they speak Portuguese.” 1 2 3 4 5
“In Brazil, Portuguese is spoken.” 1 2 3 4 5
Participants are asked to go through each scenario and rate the different options for ending it
on a scale from 1 to 5, with 1 standing for ‘very unacceptable’ and 5 for ‘very acceptable’. We
are aware that the choice of a Likert scale is not uncontroversial. For one thing, it assumes that
there are (only) five clear-cut levels of acceptability. Still, a supposedly better alternative like
magnitude estimation (see Bard et al. 1996) has been criticized too in recent years (e.g. Sprouse
and Almeida 2012) and has even been shown to produce basically the same results as the Likert
scale (see Van Olmen and Breed 2018b: 803 for details). The latter method also has the ad-
vantage that participants may be expected to be familiar with it.
As (26) makes clear, the options provided for English in each scenario are the dedicated
impersonal pronoun one, the second person, the third person plural and the passive. For Roma-
nian, every scenario includes the second person singular and the third person plural marked by
verbal inflection only, the overt third person plural and the passive, as in (27).
(27) UNI-INT-NVER-MOD
Prietena ta a luat lecții de spaniolă de doar două săptămâni, când tu îi ceri să traducă
un text pentru tine. Ea e un pic supărată și îți răspunde:
‘Your friend has only been taking Spanish classes for two weeks when you ask her if
she could translate a text for you. She is a little annoyed and says:’
“Nu se poate învăţa o limbă străină în două săptămâni.” 1 2 3 4 5
‘“You cannot learn a foreign language in two weeks.”’
“Nu pot învăţa o limbă străină în două săptămâni.” [cu sensul de ei, nu eu] 1 2 3 4 5
‘“They cannot learn a foreign language in two weeks.”’ [meaning they, not I]
“Ei nu pot învăţa o limbă străină în două săptămâni.” 1 2 3 4 5
‘“They cannot learn a foreign language in two weeks.”’
“Nu poți învăţa o limbă străină în două săptămâni.” 1 2 3 4 5
‘“A foreign language cannot be learned in two weeks.’”
Three comments are in order. First, the third person plural can have the same verb form as the
first or the third person singular in Romanian. When this syncretism makes a sentence ambig-
uous, the intended meaning is indicated in brackets at the end, as with [cu sensul de ei, nu eu]
‘[meaning they, not I]’ in (27). Second, as previously mentioned, overt subject pronouns only
occur in a pro-drop language like Romanian for, say, emphasis, contrast or disambiguation and
do not ordinarily tolerate impersonal interpretations (see Pană Dindelegan 2013: 104-111). The
rationale behind nevertheless inserting a clause with overt ei ‘they’ in (27) and not one with
overt tu ‘you (singular)’ has to do with the comparison of the third person plural’s functional
potential in a pro-drop versus a non-pro-drop language (see Sections 1.2 and 1.3). If Romanian
and English are found to differ in this respect in the hypothesized manner, the expected low
acceptability of overt ei will highlight that the dissimilarities are due to the pro-drop character
of ‘they’ as an impersonal pronoun in Romanian. Third, since the present questionnaire looks
at the passive as well as at impersonal pronouns, all active sentences with impersonal pronouns
need to be and are transitive in both languages, as (26) and (27) show. This requirement ensures
that an equivalent passive can be added as an option.
Our questionnaire contains twelve scenarios in total, one for each impersonal use distin-
guished in Van Olmen and Breed’s (2018b) semantic map in Figure 3, repeated here as Figure
4 (see Section 1.1 for its make-up and abbreviations).
Figure 4: Van Olmen and Breed’s (2018b: 840) combined semantic map of impersonalization
The three leftmost uses are exemplified in (27) to (29).
8
They all involve UNI quantification in
that they are about people in general. Impersonal pronouns in these contexts can thus be para-
phrased as ‘everyone’ or ‘anyone’. The three uses also share an INT perspective, which typi-
cally means that speaker and/or addressee are among the people in general. In (27), for instance,
the statement about not being able to learn a language in two weeks is intended as applying to
anyone, including the speaker and their friend. The differences between (27) to (29) lie in ve-
ridicality i.e. the proposition is (not) assumed to be true and modality i.e. the non-veridical
proposition has an/no explicit marker of possibility or necessity. The former distinguishes (29)
from (27) and (28): not wearing trainers at a formal party is presented as an unqualified fact of
life here, making the sentence VER. The NVER nature of (27) is due to the occurrence of MOD
can’t, which sets learning a language in two weeks in the realm of (im)possibility. In NMOD
(28), there may be no modal marker but the conditional subordinate clause still renders eating
rotten eggs hypothetical and therefore NVER.
8
For reasons of space, the remaining examples in this section will be in English only. Their Romanian equivalents
can be found in the online questionnaire: https://forms.gle/cB7okX7Jpoe9UdkJ8. To access the English question-
naire, see: https://forms.gle/9Y2bd7xVrYD1yQZY9.
(28) UNI-INT-NVER-NMOD
Your sister cracks open two eggs and pours their contents into a bowl to make an ome-
lette. A bad smell is coming from the bowl. Your sister frowns at the eggs and then says
to you:
What happens if one/you/they eat/eats rotten eggs?
What happens if rotten eggs are eaten?
(29) UNI-INT-VER
Your partner tells you they want to buy a new pair of trainers for your son to wear at his
graduation ball. You completely disagree and retort:
One/you/they never wear/wears trainers at a formal party.
Trainers are never worn at a formal party.
Like (27) to (29), the scenario in (26) introduces a UNI context: people at large in Brazil are
said to speak Portuguese. Unlike (27) to (29), it takes an EXT rather than an INT point of view.
As their friend is listing the official languages of countries, the speaker’s contribution makes
sense only as a statement about the inhabitants of Brazil and not as one about anyone, including
the interlocutors, who happens to be or can imagine themselves in the country. In short, (26) is
an instance of the UNI-EXT use in Figure 4.
The next two uses in Figure 4, from left to right, are tested with (30) and (31). The SAV
context in (30) features the speech act verb say and marks information as originating from some
unidentified source. The context in (31) involves EXI quantification, i.e. presuming the exist-
ence of a distinct (set of) individual(s) responsible for the event described: the rising prices are
attributed not to people in general but to a particular, though unidentified, entity.
(30) SAV
Your colleague is about to go on holiday and visit a renowned fortress. You have heard
that the fortress is crawling with ghosts, and you tell your colleague:
One/you/they say/says that the fortress is haunted.
It is said that the fortress is haunted.
(31) EXI-COR
You go to the petrol station and notice the higher price for petrol. When you get home,
you tell your daughter:
“One/you/they have/has raised the petrol prices again.”
“The petrol prices have been raised again.”
What differentiates (31) from the other EXI uses in (32) to (37) is that the referents may be
deemed partly known here (as in UNI-EXT). The event itself hints at the typically institutional
or, in Figure 4’s terms, COR(porate) plural-like entity behind it. To be more precise, the rising
petrol prices can only really be the fault of the government or the oil companies/suppliers.
The last six uses combine the parameters of (un)knownness and number. The former has
three values i.e. VAG(ue), INF(erred) and SPE(cific) and the latter two i.e. PL(ural) and
N(umber-)N(eutral). To explain their combinations, let us consider (32) to (34). In (32) as well
as (33), the speaker knows that the event took place but cannot or will not identify the people
responsible. Both cases are, in other words, VAG. They differ in number, however, as lynching
can never be done alone while finding a wallet can, making (32) EXI-VAG-PL and (33) EXI-
VAG-NN. The context in (34) has its number value in common with (32): one can eat pizza on
one’s own. It is not VAG, though, but INF: the speaker does not actually know that someone
or some people ate pizza. The smell in the air just makes them hypothesize that this event
happened. Scenario (34) thus triggers an EXI-INF-NN interpretation.
(32) EXI-VAG-PL
Your friend asks you about the latest film you watched: “What happened to the protag-
onist in the end?”. You reply:
“One/you/they lynched him.”
“He was lynched.”
(33) EXI-VAG-NN
You ring up your friend to tell them that the wallet that they lost the day before has
been found. You say:
“One/you/they have/has found your wallet in the park.”
“Your wallet has been found in the park.”
(34) EXI-INF-NN
After finishing lectures for the day, you and your course friend are walking back to the
common kitchen to have a late dinner. When you get there, you notice a particular
smell in the air and tell your friend:
One/you/they have/has eaten pizza in here.
Pizza has been eaten in here.
In (35) to (37), finally, the scenarios for the three remaining uses are presented. The EXI-INF-
PL context in (35) shares with (34) that the event, its actual occurrence and the existence of the
people responsible for it are inferred from certain signs the litter in the park here. Unlike the
action of eating pizza in (34), however, that of having a party normally involves more than one
person. In scenarios (36) and (37), the events of sending money and making a call are happen-
ing in the here and now of the speaker. The individuals performing these actions are neverthe-
less not explicitly identified, making both contexts EXI-SPE. They differ in number, though:
(36) is PL as only multiple people can send money from all around the world while (37) is NN
since a phone call may but need not involve more than one person.
(35) EXI-INF-PL
Your friend comes to visit your hometown and you take him for a walk in the local
park. There, you see many empty bottles on the grass, together with some leftover food
and used napkins. You tell your friend:
“One/you/they have/has thrown a party here.”
“A party has been thrown here.”
(36) EXI-SPE-PL
You and your friend have created an online donations page for the environmental or-
ganization you’re working for. When you check the page the next day, you exclaim to
your friend: “Wow! Come and have a look! …
One/you/they are sending us money from all around the world.”
Money is being sent to us from all around the world.”
(37) EXI-SPE-NN
Your sister is in the shower when her phone starts ringing at the same time. You tell
your sister: “Hurry up! ...
One/you/they is/are calling you on the phone!”
You are being called on the phone!”
2.3 Implementation
The same scenarios were presented in the same random order in the Romanian and the English
questionnaire. Both were created in Google Forms and checked by a(nother) native speaker of
the language before distribution, resulting in a few minor corrections with no significant effect
on the comparability of the two questionnaires. After receiving ethics approval from Lancaster
University, we promoted links to our Google Forms via Facebook and Whatsapp, emphasizing
that participation was voluntary and anonymous. It was also made clear at the start of the ques-
tionnaire that some demographical data would be requested (i.e. age, gender, native language(s)
and language variety) and that participants could ask to have their answers removed within two
weeks of completion (based on their demographical information). In the instructions, we high-
lighted that: (i) there are no right or wrong answers; (ii) participants should rely on their own
linguistic intuition and not on what might be (in)correct according to certain norms; (iii) there
was no time limit; and (iv) participants should make full use of the five-point scale.
The English questionnaire was filled out in December 2019 January 2020 by 62 native
speakers, of which two were left out because they were over 35 years old.
9
For Romanian, in
the same period, 77 native speakers completed the questionnaire, of which 16 were excluded
because of age considerations and one because they did not follow the instructions of making
full use of the scale. About two thirds of the 18- to 35-year-old Romanian participants identified
as female and one third as male while, among the English ones, there was essentially a fifty-
fifty split between male and female, with one participant identifying as other. We have, how-
ever, no reason to believe that gender had an impact on our results. Furthermore, most partici-
pants did not identify as speaking any particular variety of English or Romanian. For English,
the majority of participants who did answer the question wrote British or British English”,
which was to be expected as the questionnaire was primarily distributed among the first au-
thor’s contacts at Lancaster University. For Romanian, there were not enough answers to de-
termine a majority. Since the link to the questionnaire was mainly disseminated to the first
author’s contacts in Romania, it is not unreasonable to assume that the non-respondents to the
question about language variety consider themselves speakers of standard Romanian. The
same holds for the English non-respondents, of course.
2.4 Statistics
We will give, for each impersonal construction in each scenario, the following descriptive sta-
tistics: (i) the mean, i.e. the average acceptability score of the construction in question in the
impersonal use in question; (ii) the standard deviation (std), i.e. the extent to which the scores
for the construction in question in the use in question vary between all participants.
To compare one impersonal construction in two contexts (e.g. the Romanian passive in
EXI-VAG-PL versus EXI-VAG-NN) or two constructions in the same context (e.g. you versus
they in UNI-EXT), two-sided t-tests will be used in order to avoid making any assumptions
about the direction of any difference (see Baayen 2008: 8). These t-tests will be of the so-called
dependent kind for findings from the same group of participants, i.e. when contrasting imper-
sonal constructions in the same language (e.g. you versus one in UNI-INT-VER). But they will
be of the independent kind for results from different groups of people, i.e. impersonal construc-
tions in English versus Romanian (e.g. you versus the second person singular in Romanian in
UNI-INT-VER) (see Rasinger 2013: 20). If data is compared multiple times with these tests,
9
We are not aware of any generational differences in the use of impersonal constructions. We simply wanted to
ensure that our Romanian and English participants are as comparable as possible.
we will apply a Bonferroni correction to prevent any type I errors (see Baayen 2008: 114). Put
differently, our standard threshold for significance of 0.05 used for the “simple” comparison
of two results (e.g. the Romanian passive versus pro-dropped ‘they’ in SAV) – will be divided
by the number of tests undertaken, to ensure that we do not give too much weight to the out-
come of one particular test. Examining one across the UNI-INT domain, for instance, requires
three contrasts (i.e. NVER-NMOD versus NVER-MOD, NVER-MOD versus VER and VER
versus NVER-NMOD) and, accordingly, our threshold for significance will be 0.017, which is
0.05 divided by three. In the remainder of the article, we will always refer to the threshold that
is required for the type of comparison at issue.
Finally, to evaluate the influence of one or two variables on an impersonal construction’s
acceptability, we will compute one- or two-way analyses of variance (ANOVAs; see Rasinger
2013: 210-217), for which our threshold for significance is 0.05. The impact of the parameters
of number and (un)knownness on the acceptability of the third person plural, for example, will
be tested with a two-way ANOVA and, to be more precise, one with repeated measures as the
data is from a single group of participants (see Baayen 2008: 264). Afterward, post-hoc Bon-
ferroni-corrected t-tests will be used to ascertain significant differences between specific con-
texts (e.g. they in EXI-SPE-PL versus EXI-SPE-NN or pro-dropped ei in EXI-VAG-PL versus
EXI-SPE-PL).
3 Results
As mentioned in Section 1.3, the present section will first look at the UNI-INT uses (3.1), then
discuss the non-UNI-INT ones (3.2) and, lastly, give a summary of the results (3.3).
3.1 Universal-internal uses
3.1.1 In general
Table 1 gives the means and standard deviations of the impersonal pronouns and the passive in
Romanian and English in all UNI-INT contexts. Note that, for convenience’s sake, tu here and
elsewhere stands for the second person singular expressed by verbal inflection, not for the overt
pronoun.
Table 1: Descriptive statistics of the UNI-INT uses
Use
Romanian
tu
ei (pro-drop)
ei (overt)
Passive
one
You
they
Passive
UNI-INT-NVER-NMOD
mean
4.90
1.58
1.45
1.63
2.80
4.73
1.63
3.05
std
0.35
0.94
0.83
0.88
1.26
0.61
0.80
1.05
UNI-INT-NVER-MOD
mean
4.75
1.37
1.18
4.50
2.90
4.43
1.40
3.97
std
0.54
0.74
0.62
0.83
1.30
0.79
0.72
0.84
UNI-INT-VER
mean
4.13
1.32
1.15
3.87
2.93
4.30
2.23
4.00
std
1.21
0.70
0.40
1.14
1.35
1.05
1.14
0.84
Let us look at the third person plural in both languages first. The numbers for they, pro-dropped
ei and overt ei are so low in all three uses in Table 1 that we can probably safely reject them as
UNI-INT impersonal constructions without any statistical analysis. They’s average score across
these contexts is 1.75, pro-dropped ei’s 1.42 and overt ei’s 1.26, suggesting that the third person
plural is, essentially, unacceptable in the UNI-INT domain. Romanian ‘they’ is, in this regard,
in keeping with what we know from most other languages (see Section 1.2, where only Russian
and Hebrew are mentioned as exceptions to the third person plural’s characteristic incompati-
bility with UNI-INT uses). ‘They’ will therefore not be considered any further in this section
(but see Section 3.2).
The remaining data is presented in a more accessible way in Figure 5. Green is used for
Romanian and orange for English whereas full lines with round markers indicate ‘you’, dashed
lines with square markers the passive and the double line with a triangular marker ‘one’.
Figure 5: Distribution of the means of ‘you’, ‘one’ and the passive for the UNI-INT uses
3.1.2 ‘You’ and ‘one’
‘You’ is, in absolute terms, the most acceptable impersonal construction in both Romanian and
English. Tu and you also do not differ much from one another: only in UNI-INT-NVER-MOD
is there a small but significant difference in acceptability between the two (4.75 versus 4.43;
p<0.05). The way in which they vary across the three uses is very similar too: both are slightly
but substantially less acceptable in UNI-VER than in at least one of the other universal-internal
contexts (4.13 versus UNI-INT-NVER-MOD 4.75 and UNI-INT-NVER-NMOD 4.90 for Ro-
manian; 4.30 versus UNI-INT-NVER-NMOD 4.73 for English; p<0.017 in each case). The
reasons for these minor dissimilarities are not entirely clear to us. Still, the fact that both tu and
you score highest for UNI-INT-NVER-NMOD could be argued to be in line with Laberge and
Sankoff’s (1979: 432) findings for French, where tu is preferred (to on) in conditional contexts
(see Section 1.1). Notwithstanding the small differences, the overall picture of the second per-
son (singular) as an all-purpose UNI-INT impersonal construction is consistent with previous
research (see Section 1.2). One explanation for its suitability for this domain of impersonaliza-
tion lies in its features as a personal pronoun. ‘You’ is normally used to refer to the addressee.
When it occurs in a generic sentence not immediately or solely relevant for that person, they
will still be invited to transpose themselves into the situation through mental simulation, pro-
ducing an INT perspective (see Gast et al. 2015).
Of the constructions that can appear in UNI-INT contexts in English, one can be argued
to be the least acceptable. Its scores, averaging 2.87, are significantly lower than those of you
and the passive, averaging 4.49 and 3.67 respectively, in all uses (p<0.017 in all cases), except
for the latter’s UNI-INT-NVER-NMOD score (see Section 3.1.3). This result confirms an ear-
lier one about one compared to you by Van Olmen and Breed (2018b: 815), who attribute it to
its “formal character … clash[ing] with the colloquial nature of the scenarios”. Note also that
one’s acceptability is unaffected by the type of UNI-INT context, as a one-way ANOVA shows
1
1.5
2
2.5
3
3.5
4
4.5
5
UNI-INT-NVER-NMOD UNI-INT-NVER-MOD UNI-VER
one
you
Eng passive
tu
Rom passive
(p>0.05). This fact supports Haas’s (2018b) claim that any correlation between the use of one
and non-veridicality has weakened/disappeared (see Section 1.2).
3.1.3 The passive, versus ‘you’
For the passive, we will first examine the UNI-INT-NVER-MOD and UNI-INT-VER contexts.
The construction scores relatively high in these uses in both languages, with a small but signif-
icant drop in acceptability from UNI-INT-NVER-MOD to UNI-INT-VER in Romanian (from
4.50 to 3.87; p<0.017) but not in English (3.97 and 4.00 respectively; p>0.017). Furthermore,
in UNI-INT-NVER-MOD, both passives are less acceptable than ‘you’ (p<0.05 in both cases)
whereas, in UNI-INT-VER, they are actually on par with the second person (singular) (p>0.05
in both cases).
10
This finding indicates that the supposed preference for impersonal pronouns
and relative aversion to passives in the UNI-INT domain (see Section 1.2) may at least partially
be affected by veridicality or modality. This idea seems to be confirmed by our data for UNI-
INT-NVER-NMOD. The passive is significantly less acceptable here than in the other two
UNI-INT uses (p<0.017 in all cases) and than ‘you’ (p<0.05 in both cases) in Romanian and
English, with much lower scores of 1.63 and 3.05 respectively. Interestingly, Gast (2015) too
finds, in his corpus study of the English translated and original equivalents of German man,
that combining veridicality and modality interacting with quantification (alongside the original
or translated status of man) predicts the choice of man’s English counterpart the best. Unfortu-
nately, the concrete impact of these variables is not discussed in detail. Still, Gast’s (2015: 22)
association plot for (non-)veridicality suggests, for one, that, in NVER (versus VER) contexts,
there are fewer passives and more instances of you than expected in view of the overall distri-
bution of impersonal constructions. Our data may be pointing in the same direction, in that the
acceptability of English and Romanian ‘you’ is significantly higher than that of the passive in
the two NVER uses, providing some supporting evidence for the aforementioned dispreference
for passives in this domain (see also Sansò 2006: 245-263), but not in the UNI-INT-VER one.
Our figures would then also signal a potentially even stronger preference for the second person
(singular) to the passive in UNI-INT-NVER-NMOD (versus -MOD) contexts.
A closer look at the UNI-INT-NVER-NMOD scenario reveals that veridicality and mo-
dality are of course not the only factors at play in the competition between ‘you’ and the pas-
sive. Our intuition is that the latter’s scores for this use would be much higher if its subject, the
active object, was more animate and/or more definite, i.e. if the scenario tested a sentence like
‘what happens if animals are treated inhumanely?’ instead of ‘what happens if rotten eggs are
eaten?’ (see Section 2.2). The inanimate and indefinite nature of the active object here seems
especially problematic in Romanian, given the fairly large difference in the passive’s accepta-
bility between this language (1.63) and English (3.05) (p<0.05). It is important to add, however,
that the UNI-INT-NVER-MOD and UNI-INT-VER scenarios contain inanimate and indefinite
passive subjects too i.e. a foreign language’ in ‘a foreign language cannot be learned in two
weeks’ and ‘trainers’ in ‘trainers are never worn at a formal party’ respectively and that, in
these contexts, Romanian and English on the one hand and the passive and ‘you’ on the other
are not too dissimilar. It is unclear to us why such passive subjects would be less problematic
here and we leave the question for further research.
3.2 Non-universal-internal uses
3.2.1 In general
10
The English passive does outperform one in both UNI-INT-NVER-MOD (3.97 versus 2.90) and UNI-INT-
VER (4.00 versus 2.93) (p<0.05 in either case).
Table 2 provides the means and standard deviations of the impersonal pronouns and the passive
in Romanian and English in all non-UNI-INT uses.
Table 2: Descriptive statistics of the non-UNI-INT uses
Use
Romanian
English
tu
ei (pro-drop)
ei (overt)
passive
one
you
they
passive
UNI-EXT
mean
2.10
3.53
2.60
2.40
2.57
3.10
4.82
3.57
std
1.19
1.27
1.17
1.17
1.13
1.24
0.43
1.01
EXI-COR
mean
1.05
4.37
2.27
4.30
1.48
1.22
4.77
4.55
std
0.29
0.92
1.02
0.87
0.75
0.67
0.46
0.65
EXI-VAG-PL
mean
1.03
4.63
2.50
4.67
1.32
1.08
4.25
4.67
std
0.18
0.64
1.28
0.68
0.62
0.33
0.95
0.73
EXI-VAG-NN
mean
1.17
4.45
2.47
4.55
1.47
1.03
3.18
4.78
std
0.59
0.75
1.20
0.77
0.75
0.18
1.20
0.45
EXI-INF-PL
mean
1.20
3.73
2.35
3.27
1.40
1.20
2.83
4.37
std
0.82
1.13
1.18
1.38
0.72
0.58
1.21
0.76
EXI-INF-NN
mean
1.48
3.78
2.18
2.37
1.40
1.33
3.53
4.12
std
1.13
1.26
1.21
1.25
0.76
0.73
1.03
0.85
EXI-SPE-PL
mean
1.03
4.30
2.05
3.15
1.27
1.08
4.02
4.57
std
0.18
0.98
1.06
1.27
0.58
0.28
0.98
0.65
EXI-SPE-NN
mean
1.00
2.87
1.75
3.78
1.20
1.03
2.37
4.10
std
0.00
1.33
0.91
1.25
0.44
0.18
1.23
1.12
SAV
mean
1.45
1.98
2.53
2.22
2.20
1.85
4.55
3.97
std
1.10
1.05
1.26
1.22
1.13
1.04
0.70
0.99
It is clear from the numbers that ‘you’ and ‘one’ do not function as non-UNI-INT impersonal
constructions. Tu’s scores range from 1.00 to 2.10, you’s from 1.03 to 3.10 and one’s from 1.20
to 2.57. These results are not surprising as the second person (singular) and ‘one’-pronouns are
known to be restricted to the UNI-INT domain (see Section 1.2). In fact, their highest scores in
Table 2 are all for UNI-EXT and the likely reason for this incongruity is that certain participants
read an INT perspective into (26), repeated as (38).
(38) UNI-EXT
Your friend is making a list of the official languages in every country. Based on your
recent traveling experience, you tell her:
“In Brazil, one/you/they speak/speaks Portuguese.”
“In Brazil, Portuguese is spoken.”
Although the scenario strongly favors an EXT point of view (see Section 2.2), it is not incon-
ceivable that, when actually seeing the sentences with ‘you’ and ‘one’, some participants some-
how tried to include themselves and people at large in the interpretation (e.g. ‘if one is in Brazil,
one is forced to communicate in Portuguese’) (see also Van Olmen and Breed 2018b: 823). In
short, notwithstanding the slightly higher scores for UNI-EXT, the overall numbers justify not
considering ‘you’ and ‘one’ any further here.
The same holds for overt ei, whose scores vary from 1.75 to 2.60 in Table 2. It is found
to be substantially less acceptable than pro-dropped ei, averaging 3.76 compared to 2.26, in all
non-UNI-INT contexts (p<0.05 for every t-test). This result shows that, in a pro-drop language
like Romanian, it is the pro-dropped third person plural that serves as an impersonal pronoun
and that impersonal they needs to be compared to pro-dropped and not overt ei. The only
exception is SAV: overt ei is significantly more acceptable in this use than its pro-dropped
counterpart (p<0.05). It is tempting to attribute this finding to the “controversial” status of SAV
as an impersonal context. As discussed in Section 1.1, in some languages, ‘they’ cannot be
employed impersonally but may occur in SAV while, in other languages, it has a range of
impersonal uses but does not appear in SAV. Siewierska and Papastathi (2011) therefore regard
SAV as a direct off-shoot from personal ‘they’ in their semantic map and as unrelated to UNI-
EXT and the EXI uses (see Figure 2). Our Romanian data could be seen as supporting evidence
for this decision. However, both overt and pro-dropped ei score relatively low, with 2.53 and
1.98 respectively, suggesting that neither option is actually particularly acceptable in SAV.
Figure 6 shows the remaining data in a more accessible manner. Green, orange and the
dashed lines with square markers again stand for Romanian, English and the passive respec-
tively while the dotted lines with diamond markers represent the third person plural.
Figure 6: Distribution of the means of ‘they’ and the passive for the non-UNI-INT uses
3.2.2 ‘They’
Let us start with the third person plural. For English, a one-way ANOVA confirms what is
clear from Figure 6: the acceptability of they varies significantly in the non-UNI-INT domain
(p<0.05). Post-hoc Bonferroni-corrected t-tests reveal that it scores much higher for UNI-EXT
(4.82) and EXI-COR (4.77) than for the other existential uses (p<0.006 in all cases). These two
uses do not differ from SAV (4.55), though, whose score in turn is similar to that of EXI-VAG-
PL (4.23) (p>0.006 for every test) but dissimilar to those of the remaining EXI uses (p<0.006
in each case). Van Olmen and Breed (2018b: 827) too find that they is most acceptable in SAV
and in the two contexts where the referents are partly known (see Section 1.1). Our results thus
corroborate that the third person plural’s “overall speaker preference for [the so-called] semi-
impersonal uses” (Siewierska and Papastathi 2011: 599), where its personal pronoun traits of
definiteness and plurality can still be felt (see Section 1.2), manifests itself in English.
For the “purely” impersonal EXI contexts, a two-way ANOVA demonstrates that num-
ber, (un)knownness and their interaction all have a significant impact (p<0.05 in each case) on
the acceptability of they. Moreover, post-hoc Bonferroni-corrected t-tests show that the signif-
icant differences between particular uses follow the logic of the map in Figure 4: if differences
exist, ‘they’ will be more acceptable in PL than in NN contexts as the former are closer to its
personal trait of plurality and more acceptable in VAG than in INF or SPE contexts as the
1.00
1.50
2.00
2.50
3.00
3.50
4.00
4.50
5.00
they
Eng passive
ei (pro-drop)
Rom passive
former are arguably more definite than the latter two types with their situational identification.
To be more precise, they scores significantly higher: (i) in EXI-VAG-PL (4.25) than in EXI-
VAG-NN (3.18) and in EXI-INF-PL (2.83); (ii) in EXI-VAG-NN (3.18) than in EXI-SPE-NN
(2.37); and (iii) in EXI-SPE-PL (4.02) than in EXI-SPE-NN (2.37) (p<0.006 in all cases).
11
On
the whole, our results therefore confirm Van Olmen and Breed’s (2018b) finding that number
and (un)knownness interact in the way expected for they (see Section 1.1). The only exception
here, for which we have no ready explanation, is that they is substantially more acceptable in
EXI-INF-NN (3.53) than in EXI-INF-PL (2.83) (p<0.006).
12
Number and (un)knownness, as well as their interaction, affect Romanian ‘they’ too, as
a two-way ANOVA makes clear (p<0.05 for all), and our post-hoc t-tests show that they do so
entirely in the expected way. Pro-dropped ei has a significantly higher score: (i) in EXI-VAG-
PL (4.63) than in EXI-INF-PL (3.73); (ii) in EXI-VAG-NN (4.45) than in EXI-INF-NN (3.78)
and EXI-SPE-NN (2.87); and (iii) in EXI-SPE-PL (4.30) than in EXI-SPE-NN (2.87) (p<0.006
in all cases).
13
A comparison of these purely impersonal EXI uses to the three remaining non-
UNI-INT ones produces some remarkable results, though. EXI-COR (4.37) is one of four con-
texts which exhibit no substantial variation in the third person plural’s acceptability (p>0.006
in all cases) but which all score higher than any other context (p<0.006 in all cases). The uses
with which EXI-COR shares the top spot are EXI-VAG-PL (4.63), EXI-VAG-NN (4.45) and
EXI-SPE-PL (4.30) and not UNI-EXT (3.53) and SAV (1.98), like in many other languages
(e.g. Siewierska and Papastathi 2011: 596; Van Olmen and Breed 2018b: 838; see also Section
1.2). The fact that pro-dropped ei is less acceptable, though not unacceptable, in UNI-EXT
means that there is no general preference for semi-impersonal uses of ‘they’ in Romanian,
unlike in English. It also makes one wonder whether any other form of impersonalization is
favored in this context. The same question arises for SAV. The fact that the third person plural’s
acceptability in this use is very low (and significantly lower than in any other use; p<0.006 for
every t-test) suggests that Romanian is like French in having a range of impersonal contexts
for ‘they’ but not employing it to mark information as coming from an unidentified source (see
Section 1.1).
14
A number of (dis)similarities between pro-dropped ei and they are evident from the pre-
ceding paragraphs. They obviously differ in their acceptability in UNI-EXT and SAV relative
to the other uses but EXI-VAG-NN too occupies another position in Romanian than in English.
It yields one of the highest scores for pro-dropped ei, yet one of the lowest ones for they. What
the two forms share is relatively high scores in EXI-COR, EXI-VAG-PL and EXI-SPE-PL and
considerably lower ones in EXI-INF-PL, EXI-INF-NN and EXI-SPE-NN. However, to answer
the question raised in Section 1.3 does a pro-dropped third person plural fare better than an
overt one in less definite and less plural contexts? a more direct comparison is required. The
results of such an undertaking seem to confirm Siewierska and Papastathi (2011: 600) hypoth-
esis. They only scores significantly higher than pro-dropped ei in the two semi-impersonal uses,
11
EXI-VAG-PL (4.25) versus EXI-SPE-PL (4.02) on the one hand and EXI-VAG-NN (3.18) versus EXI-INF-
NN (3.53) on the other hand do not differ in English.
12
We considered the possibility of participants just finding the EXI-INF-PL scenario of having a party in the park
somewhat strange, leading them to give lower scores. Their much higher scores for the passive in this context are,
however, at odds with this explanation. We also gave thought to the possibility of participants regarding the EXI-
INF-NN scenario of eating pizza in a common kitchen as an activity almost necessarily involving more than one
person, prompting them to rate they higher than perhaps expected. A party is still a more clearly PL activity in our
view, though.
13
EXI-VAG-PL (4.63) versus EXI-VAG-NN (4.45) on the one hand and versus EXI-SPE-PL (4.30) on the other
and EXI-INF-PL (3.73) versus EXI-INF-NN (3.78) do not differ in Romanian.
14
Manea (2012) was quoted in Section 1.3 as saying that the SAV use of ‘they’ was attested in “older Romanian”.
We are not in any position to evaluate this claim. Still, if it is true, the data from our 18- to 35-year-old participants
indicates that it has disappeared from the language.
i.e. UNI-EXT and UNI-COR, and in the inherently plural SAV use (p<0.05 in every case). Pro-
dropped ei, by contrast, has a substantially higher acceptability than they in the “less plural”
contexts of EXI-VAG-NN and EXI-SPE-NN and the “less definite” plural contexts of EXI-
VAG-PL and EXI-INF-PL (p<0.05 in all cases). No differences exist in EXI-SPE-PL and EXI-
INF-NN. Overall, the more reduced form of ‘they’ can therefore be said to be more bleached
semantically, in faring better in uses that are further removed from its personal pronoun fea-
tures.
3.2.3 The passive, versus ‘they’
Let us now turn to the passive. A one-way ANOVA for English indicates that its acceptability
varies substantially in the non-UNI-INT domain (p<0.05). Post-hoc t-tests point out that two
uses in particular stand out, in a negative sense. The passive scores significantly lower (p<0.006
in all cases) in UNI-EXT (3.57) than in all other contexts, except for EXI-SPE-NN (4.1) and
SAV (3.97), and in SAV than in the other contexts, excluding the two aforementioned ones
and EXI-INF-NN (4.12). A plausible explanation for UNI-EXT’s status is that the passive is
essentially vague with respect to perspective, which has already been argued to be somewhat
of a problem with this use (see the discussion of the scores of ‘one’ and ‘you’ in Table 2). The
original subject might be taken to be you/one, in which case the passive receives an INT inter-
pretation, or they, in which case it gets an EXT one. Speakers may prefer to avoid this ambi-
guity by employing an alternative impersonal construction. SAV’s status is interesting given
Siewierska’s (2008: 12) assertion (see Section 1.2) that the passive is favored in this context in
British English. A comparison with they is needed, though, to assess this claim properly. An-
other observation for which such an undertaking seems relevant is that the passive exhibits no
clear preference for more definite and/or more plural uses. EXI-VAG-NN (4.78) is its top use,
for example, and semi-impersonal EXI-COR (4.55) only comes fourth (p>0.006, however).
More generally, despite other minor statistically significant differences, the passive can be said
to score consistently, and perhaps surprisingly, high in all contexts other than UNI-EXT and
SAV, ranging from 4.78 to 4.10 while they’s scores vary from 4.77 to 2.37 in these uses.
A first finding of the two strategies’ direct comparison is that UNI-EXT and SAV are
two contexts where they scores substantially higher (p<0.05). In the former (4.82 versus 3.57),
the pronoun has the advantage of making the EXT perspective explicit. Its higher acceptability
in the latter (4.55 versus 3.97) contradicts Siewierska (2008: 12): it is actually the third person
plural that is (slightly) favored in SAV. A possible explanation for this difference is the sce-
nario’s informality, which may be more compatible with the pronoun than with the passive (see
Section 1.2). A second outcome concerns EXI-COR, where the statistically significant differ-
ence between they (4.77) and the passive (4.55) is fairly minimal. It suits the pronoun because
of its semi-impersonal and plural character but it also fits well with the passive. The referents
can be identified through the event itself here, making the inclusion of a separate marker point-
ing at them perhaps less essential. A third result is that, in all remaining contexts, the passive
is significantly more acceptable than they (p<0.05 in every case), despite the former’s assumed
formal character (see Section 1.2). Thus, as suggested by Van Olmen and Breed (2018b: 839),
pronouns may not be a preferential construction in the purely impersonal EXI domain in Eng-
lish, unlike in the other areas of impersonalization (see also Section 3.1.3). A fourth and final
finding is that the differences between the passive and they are much larger in NN than in PL
contexts in EXI-VAG (NN 4.78 and 3.18; PL 4.67 and 4.25) and EXI-SPE (NN 4.10 and 2.37;
PL 4.57 and 4.02). This fact could be regarded as supporting Siewierska’s (2008: 16) idea (see
Section 1.2) that passives are favored over ‘they’ when the unidentified agent is almost cer-
tainly a single person.
Like its English equivalent, the Romanian passive shows substantial variation in the non-
UNI-INT domain, as a one-way ANOVA shows (p<0.05), and post-hoc t-tests reveal the “neg-
ative” prominence of the same two contexts. The passive scores significantly lower in UNI-
EXT (2.40) and SAV (2.22) than in all other uses (p<0.006 in all cases), except for EXI-INF-
NN (2.37). The passive’s low acceptability in the former can probably again be explained by
its perspectival ambiguity. Its score for the latter, together with our knowledge of ei in this
context, suggests that a construction other than the passive or third person plural the reflexive
impersonal (see Section 1.2) in our intuition is preferred for SAV in Romanian. As regards
the other uses, the passive is most acceptable not just in semi-impersonal EXI-COR (4.30) but
also in purely impersonal EXI-VAG-PL (4.67) and EXI-VAG-NN (4.45) (no substantial dif-
ferences exist between them). Its score in EXI-SPE-NN, though significantly lower than in the
preceding contexts (p<0.006 in all cases), is quite high too (3.78). However, it trails to such
low levels in EXI-INF-PL (3.27), EXI-SPE-PL (3.15) and EXI-INF-NN (2.37) that one cannot
but wonder whether pro-dropped and semantically bleached ei is preferred in these uses.
The explicit comparison needed to answer the question produces a fairly complex picture.
As expected, the third person plural is significantly more acceptable in UNI-EXT than the pas-
sive (3.53 and 2.40; p<0.05) because of its unambiguously EXT reading and, in SAV, the two
strategies seem equally unacceptable (1.98 and 2.22; p>0.05). For EXI-COR, pro-dropped ei
and the passive score similarly high (4.37 and 4.30; p>0.05), which may be attributed to their
comparable suitability for the use (see the argument for English). Moreover, they do not differ
in EXI-VAG-PL (4.63 and 4.67) and EXI-VAG-NN (4.45 and 4.55) either. In three of the four
remaining contexts, however, the third person plural has a substantially higher score than the
passive in EXI-INF-PL (3.73 and 3.27), EXI-INF-NN (3.78 and 2.37) and EXI-SPE-PL (4.30
and 3.15), to be precise (p<0.05 in each case). In other words, unlike in English, the pronoun
actually seems to be favored in these purely impersonal EXI uses a preference that is likely
facilitated by the less definite and less plural character of impersonal pro-dropped ei. Still, one
would probably assume the third person plural to be more acceptable in EXI-SPE-NN too then
but, for some reason, the passive scores significantly higher here (3.78 and 2.87; p<0.05). Note,
finally, that the above discussion features no evidence for a passive bias in number-neutral or
single contexts, unlike in English, and that, arguably, no treatment of the competition between
impersonal constructions in Romanian is complete without the reflexive impersonal (see Pană
Dindelegan 2008: 138, who states that it is generally preferred to the passive).
To conclude the present section, we will briefly compare the passive in the two lan-
guages. It will not come as a surprise that it scores significantly higher in English than in Ro-
manian in nearly every single use (p<0.05 in all cases). Contrasting the English passive to they
has already shown that it is a more generally accepted construction in this area, particularly in
the purely impersonal EXI domain and its NN contexts. The Romanian passive, by contrast,
tends to fare worse than pro-dropped ei in such uses. One possible explanation for the differ-
ences in the passive between the two languages is therefore that its acceptability (at least par-
tially) depends on the competition that it faces as an impersonal construction and thus on the
degree to which the third person plural is bleached as an impersonal pronoun in the language.
It is important to stress again in this regard that ‘they’ is not the only established alternative to
the passive in Romanian, unlike in English. The existence of the reflexive impersonal may well
play a role in the Romanian passive’s overall lower scores too, averaging 3.41 (compared to
4.30 for its English equivalent). It almost certainly accounts, for instance, for the difference in
SAV (2.22 versus 3.97), where the reflexive impersonal is probably the only option in our
view. An assessment of this construction’s impact in general will, however, have to be left for
future research.
3.3 Summary
As a way to sum up our results, Figures 7 to 12 present the varying acceptability of the imper-
sonal pronouns examined here and of the passive on Van Olmen and Breed’s (2018b) semantic
map (see Figure 4) in such a way that the darker a context is, the more acceptable the construc-
tion is there (one is not included here as, for our purposes, it can just be assumed to be a less
acceptable version of you).
Figure 7: Romanian ‘you’
Figure 8: Romanian ‘they’
Figure 9: The Romanian passive
Figure 10: English ‘you’
Figure 11: English ‘they’
Figure 12: The English passive
4 Conclusion
As mentioned in Section 1.3, the main questions that this article aims to answer are: (i) what is
the functional potential of the impersonal pronouns and the passive in Romanian and English?;
(ii) how do the impersonal constructions and the languages compare to each other?; (iii) do the
parameters of number and (un)knownness interact with one another (in the expected way) in
Romanian?; and (iv) does the pro-dropped third person plural exhibit a(n expected) wider range
of less definite/plural impersonal uses than its overt equivalent? We will now discuss these four
questions one by one.
Let us start with (iv). Our results clearly show that the answer to this question is positive.
Unlike its pro-dropped counterpart, overt ei is, in essence, unacceptable as an impersonal pro-
noun in Romanian. Pro-dropped ei is also more acceptable than they in contexts that are further
removed from the third person plural’s definite and plural personal pronoun prototype (com-
pare Figures 8 and 11). Romanian can thus be added to the list of languages supporting Siew-
ierska and Papastathi’s (2011) hypothesis based on the relation between formal reduction and
semantic bleaching that the more reduced ‘they’ is, the more likely it is to be able to occur in
less definite or less plural impersonal contexts. This fact makes Polish, the only pro-drop lan-
guage in their sample of nine with a limited range of impersonal uses for the third person plural,
look even more exceptional. An investigation into the reason why it occupies this special po-
sition thus seems warranted. Still, the hypothesis itself would probably also benefit from more
data from pro-drop versus non-pro-drop languages beyond Europe.
As to question (iii), our findings indicate that the answer is yes too. (Un)knownness and
number combine to influence the acceptability of both English and Romanian ‘they’ in the truly
impersonal EXI domain in the predicted way on the whole (see Figures 8 and 11). Pro-dropped
ei in EXI-VAG/SPE-PL/NN can serve as an example of an interaction between the parameters
that meets expectations. The VAG-SPE contrast itself is not sufficient to affect the pronoun’s
acceptability, as there is no difference between EXI-VAG-PL and EXI-SPE-PL, and neither is
the PL-NN contrast, as no difference exists between EXI-VAG-PL and EXI-VAG-NN. How-
ever, the combination of SPE and NN does produce a significantly lower acceptability of pro-
dropped ei, compared to: (i) EXI-SPE-PL in line with the prediction that ‘they’ fares worse
in NN contexts due to possible persistence of its personal sense of plurality; and (ii) EXI-
VAG-NN (and EXI-VAG-PL) in keeping with the prediction that it fares worse in contexts
relying on situational identification due to potential persistence of its personal sense of defi-
niteness. Our Romanian and English data therefore reinforce Van Olmen and Breed’s (2018b)
argument based on evidence from Afrikaans and Dutch as well as English that, to capture
the third person plural’s behavior in the truly impersonal EXI domain, the separate parameters
of (un)knownness and number in, respectively, Siewierska and Papastathi’s (2011) and Gast
and van der Auwera’s (2013) maps need to be brought together. In future research, this require-
ment for a combined semantic map should be tested for other languages of the Romance family
and others.
To answer question (ii), we first have to address (i). For Romanian, our findings reveal a
clear division of labor between the impersonal pronouns ‘you’ and ‘they’.
15
The former’s func-
tional potential lies in the UNI-INT domain and the latter’s in the non-UNI-INT domain (com-
pare Figures 7 and 8). Their respective incompatibility with the other sphere is, of course, not
surprising (e.g. Kitagawa and Lehrer 1990) and can be traced back to their features as personal
pronouns. The second person singular’s addressee-oriented nature clashes with uses excluding
the addressee as a possible referent; the third person plural’s focus on parties other than speaker
and addressee is at odds with contexts that include the speech participants as potential referents.
Within the UNI-INT domain, there is, all in all, little variation. The fact that the second person
singular’s highest level of acceptability is found in UNI-INT-NVER-NMOD can nevertheless
be said to be in line with Laberge and Sankoff’s (1979: 432) findings for French, where tu is
preferred (to on) in conditional contexts. Within the non-UNI-INT domain, Romanian ‘they’
seems unable to occur in SAV, which confirms Siewierska and Papastathi’s (2011) treatment
of this context as separate from the other ones. In these remaining uses (see Figure 8), the
pronoun may exhibit the effects of (un)knownness and number as mentioned earlier and es-
pecially in lower levels of acceptability in the EXI-INF and EXI-SPE-NN contexts but scores
rather consistently high, with no bias toward semi-impersonal UNI-EXT and EXI-COR (found
in other languages; see below). This fact can be attributed to its pro-dropped character, as dis-
cussed before. The passive’s functional potential, lastly, is not limited to the UNI-INT or non-
UNI-INT sphere in Romanian (see Figure 9). It does exhibit substantial variation, though. The
passive scores particularly low in UNI-EXT and SAV (and in UNI-INT-NVER-NMOD; see
below). The likely reason for the former fact is that passives are essentially vague with respect
to INT versus EXT point of view and speakers may prefer the perspectival clarity of the third
person plural. The latter fact, together with the equally low acceptability of ‘they’ in this use,
suggests that Romanian favors yet another construction for SAV, the reflexive impersonal. This
construction deserves to be studied in more detail in future research. Its functional range and
competition with the other impersonal constructions in the language could help explain some
of the passive’s other relatively lower acceptability levels in the non-UNI-INT domain.
Let us now answer question (i) for English and, at the same time, start answering (ii) too.
Our findings confirm the division of labor between UNI-INT you and one and non-UNI-INT
they (see Figures 10 and 11; cf. Van Olmen and Breed 2018b: 842). In this regard, English and
Romanian are thus similar. Within the UNI-INT domain, you is consistently more acceptable
15
The lack of an impersonal pronoun covering the complete range of uses makes Romanian similar to its Slavic
neighbors and Romance languages like Italian and Spanish but different from, for instance, French with on. The
absence of such a pronoun in English renders the language quite unique, though, among its closest neighbors and
relatives (e.g. German with man) (see Siewierska 2011: 71).
than one. This result is in line with earlier findings (see Van Olmen and Breed 2018b: 815) and
can be attributed to a clash between one’s formal character and the informality of our scenarios.
Contrary to certain claims in the literature (e.g. Moltmann 2010), one’s acceptability does not
seem to be affected by (non-)veridicality. This fact is, however, in keeping with Haas’s (2018b)
assertion that its impact on one has faded over time. You does not differ much from its Roma-
nian equivalent in acceptability terms (compare Figures 7 and 10). Like tu, it also scores highest
in UNI-INT-NVER-NMOD and can therefore be argued to support the aforementioned link
which merits further scrutiny between the impersonal second person singular and conditional
contexts. Within the non-UNI-INT domain, they displays substantial variation. Like its Roma-
nian counterpart and as discussed before, the pronoun shows the effects of the interacting pa-
rameters of (un)knownness and number in, inter alia, its low level of acceptability in EXI-SPE-
NN. Yet, it is typically less acceptable in these truly impersonal EXI uses than pro-dropped ei,
as a result of its overt status (see above). They is also generally less acceptable in these contexts
than in UNI-EXT and EXI-COR, where it actually fares better than pro-dropped ei (compare
Figures 8 and 11). So unlike Romanian ‘they’, it does have the “overall speaker preference for
semi-impersonal uses” that Siewierska and Papastathi (2011: 599) observe for the third person
plural in many languages. They also differs from pro-dropped ei in being very much acceptable
in SAV. The English passive, lastly, clearly has the functional potential to appear in UNI-INT
contexts but, on the whole, its scores for the non-UNI-INT domain are higher (see Figure 12).
The only real exception there is UNI-EXT, a feature that it shares with its Romanian equivalent
and can be attributed to its perspectival ambiguity. Unlike the Romanian passive, however, the
English one is an acceptable construction in SAV and scores fairly consistently high in the truly
impersonal EXI sphere (compare Figures 9 and 12). Part of the explanation for the latter fact
may lie in the stronger competition that the Romanian passive faces from the more bleached
third person plural in the language and probably also from the reflexive impersonal.
The previous paragraph has compared each impersonal construction in English to its Ro-
manian counterpart. To answer question (ii) fully, the present paragraph will discuss the intra-
linguistic relationships between the constructions. Within the UNI-INT sphere, the passive has
been said to score fairly high in both Romanian and English, with the exception of UNI-INT-
NVER-NMOD though, generally, still lower than ‘you’ in the two languages, except in UNI-
VER (compare Figures 7 and 9 and 10 and 12). These findings support the idea of a UNI-INT
preference for pronouns (e.g. Sansò 2006; Van Olmen and Breed 2018b) but show, at the same
time, the passive’s evident UNI-INT potential. They also suggest, consistent with Gast’s (2015)
exploratory findings, that the choice between ‘you’ and the passive is at least partially affected
by veridicality and/or modality. In the non-UNI-INT domain, UNI-EXT stands out as the use
with one of the lowest levels of acceptability for the passive in Romanian and English, as men-
tioned earlier. Our argument that its perspectival vagueness is the reason is reinforced by the
much higher acceptability of ‘they’ in the use (compare Figures 8 and 9 and 11 and 12). SAV
is of interest too in the two languages. The passive’s similarly low score to the third person
plural in Romanian indicates that the language favors yet another construction for the context.
The fact that, in English, the passive is less acceptable than ‘they’ here challenges Siewierska’s
(2008: 12) claim that the language prefers the passive for SAV. The informality of our scenario
has perhaps played a role in the results, though. More generally, the English passive is found
to score consistently high, not only in the semi-impersonal contexts but also in the truly imper-
sonal EXI ones. In fact, in the latter and their NN versions in particular, it outperforms they.
This finding lends support to Van Olmen and Breed’s (2018b) suggestion that pronouns may
not be the preferred construction in the non-UNI-INT domain and to Siewierska’s (2008: 16)
assertion that passives are favored in NN (or single) contexts, at least in English. The Romanian
passive’s acceptability varies much more in the non-UNI-INT domain. As pointed out before,
part of the explanation could be that it just has more competition in these contexts from the
more bleached third person plural (and the reflexive impersonal) than its English equivalent.
This article is by no means the final word on impersonal constructions in Romanian ver-
sus English. The constructions analyzed here should, for example, be compared to the reflexive
impersonal in Romanian. It would be a good idea to examine their actual usage in corpora too.
Such a study would also make it possible, for instance, to investigate the ways that the second
person singular is employed in contexts beyond “structural knowledge” (Kitagawa and Lehrer
1990: 748; see Section 1.2).
Abbreviations
2 second person
3 third person
COMP complementizer
COND conditional
DEF definite
DEM demonstrative
DIST distal
F feminine
INDF indefinite
INF infinitive
M masculine
N neuter
NEG negation
PL plural
PRS present
PST past
PTCP participle
REFL reflexive
SG singular
References
Baayen, R.H. 2008. Analyzing Linguistic Data: A Practical Introduction to Statistics Using
R. Cambridge: Cambridge University Press.
Bard, E.G., D. Robertson and A. Sorace. 1996. Magnitude estimation of linguistic acceptabil-
ity. Language 72(10, 32-68.
Cabredo Hofherr, P. 2006. “Arbitrary” pro and the theory of pro-drop. In P. Ackema, P.
Brandt, M. Schoorlemmer and F. Weerman (eds.), Agreement and Arguments. Oxford:
Oxford University Press, 230-257.
Cojocaru, D. 2003. Romanian Grammar. Durham, NC: Slavic and East European Language
Research Center, Duke University.
Coussé, E. and J. van der Auwera. 2012. Human impersonal pronouns in Swedish and Dutch:
A contrastive study of man and men. Languages in Contrast 12(2), 121-138.
Creissels, D. 2008a. Impersonal and related constructions: A typological approach. Manu-
script. Lyon: Lumière University Lyon 2.
Creissels, D. 2008b. Impersonal pronouns and coreference: The case of French on. Manu-
script. Lyon: Lumière University Lyon 2.
D’Alessandro, R. 2007. Impersonal Si Constructions: Agreement and Interpretation. Berlin:
De Gruyter.
Data-Bukowska, E. 2018. The third person plural impersonal in Swedish: A typological
account. Linguistica Copernicana 15, 163-193.
De Hoop, H. and S. Tarenskeen. 2015. It’s all about you in Dutch. Journal of Pragmatics 88,
163-175.
Dobrovie-Sorin, C. 1998. Impersonal se constructions in Romance and the passivization of
unergatives. Linguistic Inquiry 29(3), 399-437.
Fenger, P. 2018. How impersonal does one get? A study of man-pronouns in Germanic. Jour-
nal of Comparative Germanic Linguistics 21(3), 291-325.
Fernández, S.S. 2013. Impersonality in Spanish personal pronouns. In K.J. Kragh and J.
Lindschouw (eds.), Deixis and Pronouns in Romance Languages. Amsterdam: Benjamins,
87-107.
Fonesca-Greber, B. and L.R. Waugh. 2003. On the radical difference between the subject
personal pronouns in written and spoken European French. In P. Leistyna and C.F. Meyer
(eds.), Corpus Analysis: Language Structure and Language Use. Amsterdam: Rodopi,
225-240.
Gast, V. 2015. On the use of translation corpora in contrastive linguistics: A case study of im-
personalization in English and German. Languages in Contrast 15(1), 4-33.
Gast, V., L. Deringer, F. Haas and O. Rudolf. 2015. Impersonal uses of the second person
singular: A pragmatic analysis of generalization and empathy effects. Journal of Pragmat-
ics 88, 148-162.
Gast, V. and J. van der Auwera. 2013. Towards a distributional typology of human imper-
sonal pronouns, based on data from European languages. In D. Bakker and M. Haspelmath
(eds.), Languages Across Boundaries: Studies in the Memory of Anna Siewierska. Berlin:
De Gruyter, 119-158.
Giacalone Ramat, A. and A. Sansò. 2007. The spread and decline of indefinite man-construc-
tions in European languages: An areal perspective. In P. Ramat and E. Roma (eds.), Eu-
rope and the Mediterranean Linguistic Areas: Convergences from a Historical and Typo-
logical Perspective. Amsterdam: Benjamins, 95-131.
Gönczöl-Davies, R. 2008. Romanian: An Essential Grammar. London: Routledge.
Haas, F. 2018a. “You can’t control a thing like that”: Genre and change in modern English
human impersonal pronouns. In R.J. Whitt (ed.), Diachronic Corpora, Genre and Lan-
guage Change. Amsterdam: Benjamins, 171-194.
Haas, F. 2018b. Changing human impersonal pronouns in English: A corpus study. Paper at
Societas Linguistica Europaea 51, Tallinn.
Hall, D. 2020. The impersonal gets personal: A new pronoun in Multicultural London Eng-
lish. Natural Language & Linguistic Theory 38, 117-150.
Jensen, T.J. and F. Gregerson. 2016. What do(es) you mean? The pragmatics of generic sec-
ond person pronouns in modern spoken Danish. Pragmatics 26(3), 417-446.
Kitagawa, C. and A. Lehrer. 1990. Impersonal uses of personal pronouns. Journal of Prag-
matics 14(5), 739-759.
Kluge. B. 2016. Generic uses of the second person singular How speakers deal with refer-
ential ambiguity and misunderstanding. Pragmatics 26(3), 501-522.
Laberge, S. and G. Sankoff. 1979. Anything you can do. In T. Givón (ed.), Discourse and
Syntax. New York: Academic Press, 419-440.
Malamud, S. 2012. Impersonal indexicals: One, you, man and du. Journal of Comparative
Germanic Linguistics 15, 1-48.
Manea, C. 2012. Remarks on the passive voice in English and Romanian. Studii de Gra-
matică Contrastivă 18, 54-73.
Moltmann, F. 2010. Generalizing detached self-reference and the semantics of generic one.
Mind & Language 25(4), 440-473.
Myhill, J. 1997. Toward a functional typology of agent defocusing. Linguistics 35(5), 799-
844.
O’Connor, P.E. 1994. “You could feel it through the skin”: Agency and positioning in prison-
ers’ stabbing stories. Text 14(1), 45-75.
Pană Dindelegan, G. 2008. Construcții pasive și construcții impersonale. In V. Guţu Romalo
(ed.), Gramatica Limbii Române II: Enunțul. Bucharest: Editura Academiei Române, 133-
147.
Pană Dindelegan, G. (ed.). 2013. The Grammar of Romanian. Oxford: Oxford University
Press.
Posio, P. and M. Vilkuna. 2013. Referential dimensions of human impersonals in dialectal
European Portuguese and Finnish. Linguistics 51(1), 177-229.
Rasinger, S.M. 2013. Quantitative Research in Linguistics: An Introduction. London:
Bloomsbury.
Rudolf, O. 2016. Human Impersonal Strategies in English and Russian: A Comparative Cor-
pus Study. Jena: University of Jena PhD dissertation.
Sansò, A. 2006. ‘Agent defocusing’ revisited: Passive and impersonal constructions in some
European languages. In W. Abraham and L. Leisiö (eds.), Passivization and Typology:
Form and Function. Amsterdam: Benjamins, 232-273.
Shibatani, M. 1985. Passives and related constructions: A prototype analysis. Language
61(4), 821-848.
Siewierska, A. 2004. Person. Cambridge: Cambridge University Press.
Siewierska, A. 2008. Ways of impersonalizing: Pronominal vs. verbal strategies. In M.A.
Gómez González, J. Lachlan Mackenzie and E.M. González Álvarez (eds.), Current
Trends in Contrastive Linguistics: Functional and Cognitive Perspectives. Amsterdam:
Benjamins, 3-26
Siewierska, A. 2011. Overlap and complementarity in reference impersonals: Man-construc-
tions vs. third person plural-impersonal in the languages of Europe. In A. Malchukov and
A. Siewierska (eds.), Impersonal Constructions: A Cross-Linguistic Perspective. Amster-
dam: Benjamins, 57-90.
Siewierska, A. and M. Papastathi. 2011. Towards a typology of third personal plural imper-
sonals. Linguistics 49(3), 575-610.
Sprouse, J. and D. Almeida. 2012. Assessing the reliability of textbook data in syntax:
Adger’s Core Syntax. Journal of Linguistics 48(3), 609-652.
Stirling, L. and L. Manderson. 2011. About you: Empathy, objectivity and authority. Journal
of Pragmatics 43(6), 1581-1602.
van der Auwera, J., V. Gast and J. Vanderbiesen. 2012. Human impersonal pronoun uses in
English, Dutch and German. Leuvense Bijdragen 98, 27-64.
Van Olmen, D. and A. Breed. 2018a. Human impersonal pronouns in Afrikaans: A double
questionnaire-based study. Language Sciences 69, 1-29.
Van Olmen, D. and A. Breed. 2018b. Human impersonal pronouns in West Germanic: A
questionnaire-based comparative study of Afrikaans, Dutch and English. Studies in Lan-
guage 42(4), 798-846.
Zifonun, G. 2001. Man lebt nur einmal: Morphosyntax und Semantik des Pronomens man.
Deutsche Sprache 38(3), 232-253.
... Een mogelijke reden kan zijn dat deze context gelinkt is aan het hier en nu van een spreker en dus minder aanwezig is in geschreven teksten. Dit sluit aan bij de resultaten van Siewierska & Papastathi (2011, p. 590 van Rădulescu & Van Olmen (2021). Zij hebben onderzoek gedaan naar onpersoonlijke constructies in het Engels en het Roemeens. ...
Article
Full-text available
Abstract (English) In many (West) Germanic languages we see that personal pronouns can be used in impersonal contexts. These pronouns are often referred to as Human Impersonal Pronouns (HIPs). This is also the case for Afrikaans with the personal pronouns ‘jy’ and ‘hulle’. Since ‘('n) mens’ as a HIP in written standard Afrikaans has already received attention in Afrikaans corpus investigations, the aim of this investigation is to understand the use of ‘jy’ and ‘hulle’ in written standard Afrikaans. These pronouns can be seen as grammatical constructions, which can develop/change over time. Typology research on impersonalization has identified twelve impersonal contexts in which HIPs can be used. To investigate the use of ‘jy’ and ‘hulle’ it is important to have an understanding of impersonalization. For this project the Taalkommissiekorpus (2011) is used. The content of the corpus is divided into three subcategories, each representing a text genre: Fiction + non-fiction/non-academic (in short: books); academic and other formal texts (study guides/academic articles/minutes; in short: formal); and the third one is based on newspaper and magazine articles. Keywords: written Afrikaans, pronouns, impersonal use, language change
Article
Full-text available
Multicultural London English (MLE) has developed a new pronoun, man, with a number of interesting properties. It can be interpreted as any person and number combination (1sg, 1pl, 2sg, 2pl, 3sg, 3pl), and at first blush appears the allow a generic impersonal reading. One central strand of analyses of the ‘generic’ or ‘quasi-universal’ interpretation of impersonal pronouns involves treating them as featurally impoverished pronouns which act as variables bound by a generic operator either high in the clause (Moltmann 2006; Sigurðsson and Egerland 2009, a.o.) or at the top of the DP (Ackema and Neeleman 2018). What is interesting about MLE man is that it seems to allow a generic reading while being generally resistant to binding: the pronoun never behaves as a bound variable, and so it could not reasonably be assumed that a generic operator can bind it. I show, however, that a closer look at the facts reveals that the quasi-generic interpretation, and all other possible interpretations of the pronoun, can be explained instead by assuming that it has a featureless person head, which introduces the full lattice of possible referents including speaker and addressee (following Harbour 2016), and then allowing contextually determined subsets of that full lattice to be picked out by a choice function, modelled on the epsilon operator of von Heusinger (2004). This novel data from MLE suggests that generic-like interpretations can arise even where generic binding is not possible, but that the traditional strategy of generating these interpretations are still needed to capture the full typology of impersonal pronouns.
Article
Full-text available
This paper focuses on overt impersonal pronouns such as English one and Dutch men in eight Germanic languages (English, Frisian, Icelandic, Danish, Dutch, German, Norwegian and Swedish). Cinque (Linguist Inq 19:521–581, 1988), Egerland (Work Pap Scand Syntax 71:75–102, 2003), a.o., argued that there are two types of impersonal pronouns: one type that can occur in multiple syntactic positions but can only have a generic reading and another type that can have generic and existential readings but can only occur as an external argument. I show, based on novel data from ECM constructions, passives and unaccusatives, that it is not the syntactic position which restricts the distribution of men-type pronouns, but that it is case. English-type pronouns can occur with multiple cases, but can only have a generic inclusive reading. All Dutch-type pronouns can only occur with nominative case and can have multiple impersonal readings. Moreover, I show that Dutch and Swedish allow an existential reading when the pronoun is a derived subject (contra Cinque 1988; Egerland 2003). I propose a direction for this correlation between the different readings and case by assuming different feature make-ups for the pronouns, following Egerland (2003), Hoekstra (J Comp Ger Linguist 13:31–59, 2010), Ackema and Neeleman (A grammar of person. Linguistic inquiry monographs, MIT Press, Cambridge, 2018): one has ϕ\phi -features and, therefore, always needs to be obligatorily inclusive; men lacks this functional layer and, therefore, has no restriction on its readings. Furthermore, I propose that since men lacks a phi-layer, it is too deficient to project a KP, and thus can only occur with unmarked nominative case.
Conference Paper
Full-text available
Human impersonal pronouns introduce non-specific, often generic, participants into dis-course, avoiding the identification of specific individuals for different reasons. In contrast to many other European languages, English does not (anymore) possess an expression that is dedicated to the impersonal function. In this paper, I will present a corpus study that traces changes in the distribution of impersonal uses of the Modern English pronouns one, you, they and people (cf. [1]-[4]). (1) Also when sleeping in woods one wakes very soon, […]. [ARCHER, 1957maca.f8b] (2) You can’t understand a thing the bloody man says. [ARCHER, 1973trev.f8b] (3) The workhouse where they put me. They beat you there like a drum. [ARCHER, 1979pomr.d8a] (4) Well, you know, in spite of all, people do say he is clever. [ARCHER, 1899mart.d6b] There is substantial work on impersonal reference in Old, Middle and Early Modern English (Fröhlich 1951, Meyer 1953, Jud-Schmid 1956, Rissanen 1997, and Seoane Posse 2000), and it is clear that the demise of the dedicated impersonal man in late Middle English has led to some degree of reorganization among the remaining forms and also with respect to the divi-sion of labour between the latter and the passive construction (Jud-Schmid 1956, van Gelderen 1997, Los 2009, and Light & Wallenberg 2015). In order to trace more recent changes, all instances of one, you, they and people were ex-tracted from the ARCHER corpus (version 3.1, covering the period 1650 to 1999) and manu-ally coded as impersonal or personal. In addition to information about registers and dia-chronic stages, more specific lexical and morphosyntactic features of the data points have been identified in order to explore changes in usage conditions. These include the syntactic function of the pronoun and features of the clause, such as clause type, verb class and the presence of modal verbs. In addition, I categorized the semantics/pragmatics of each in-stance, checking whether the hearer was part of the group generalized over, whether the sentence described a generalizing or an episodic state of affairs and whether the speaker used the impersonal to hide self-reference behind an apparent generalization. Quantitative analyses of the data indicate that a number of changes in frequency and usage conditions of human impersonal pronouns have taken place. Inter alia, it can be observed that the well-known unpopularity of impersonal one in American English is already visible in the 17th-century data. In both British and American English, the frequency of impersonal one and you has risen after 1850. In addition, those subtypes of impersonal you that deviate more from its canonical use (simulation and self-reference) have become more common in these later stages, too, possibly as a consequence of extralinguistic developments (cf. Niel-sen et al. 2009 on similar developments in Danish). In the last part of the talk, the English findings will be compared to what is known about the diachrony of impersonal strategies in other European languages.
Chapter
This book brings together new work by leading syntactic theorists from the USA and Europe on a central aspect of syntactic and morphological theory: it explores the role of agreement morphology in the morphosyntactic realization of a verb's arguments. The authors examine the differences and parallels between nonconfigurational, pronominal- agreement languages; configurational languages which allow pronoun drop (for example, "Is coming" for "He is coming"); languages that allow pronoun drop in particular constructions only; and languages which always require overt syntactic determiner phrases as arguments. The book considers whether the morphological properties of agreement play a role in determining which of these types a language belongs to and how far languages differ with respect to the argumental status of their agreement and syntactic determiner phrases. The authors explore these and related issues and problems in the context of a wide range of languages. Their book will interest linguists at graduate level and above concerned with morphosyntactic theory, linguistic typology, and the interactions of syntax and morphology in different languages.
Article
Konstrukcja nieosobowa z zaimkiem trzeciej os. l.mn. w języku szwedzkim – ujęcie typologiczneW niniejszym artykule przeanalizowano konstrukcję nieosobową z zaimkiem 3 os. l.mn. de/dom ‘oni/one’ w języku szwedzkim. Zgodnie z ujęciem Siewierskiej (2008a) zaimek jest traktowany jako językowy środek przedstawiania agenta w dyskursie jako niereferencyjnego i nieokreślonego (lub nieostrego). Celem analizy jest ustalenie w jakim zakresie szwedzkie wypowiedzenia zawierające taki zaimek wpisują się w taksonomię nieosobowych znaczeń, ustaloną dla języków europejskich i szeroko dyskutowaną w badaniach typologicznych. Analiza danych językowych zaczerpniętych ze stron internetowych pozwala nakreślić bardziej wiarygodny obraz nieosobowego szwedzkiego de/dom. Jak zostaje wykazane zaimek ten posiada znacznie szerszy repertuar sensów niż zakłada się w szwedzkich gramatykach. Pokrywa on całe spektrum znaczeń określanych w badaniach z zakresu jako: uniwersalne (universal), korporacyjne (corporate), niejasne (vague existential) oraz specyficzne (specific existential). W funkcji nieosobowej występuje także w konstrukcji zawierającej czasownik mówienia säga. W przeprowadzonym badaniu nie potwierdzono natomiast znaczenia presuponowanej obecności sprawcy czynności (inferred existential).