<!DOCTYPE article PUBLIC "-//NLM//DTD JATS (Z39.96) Journal Archiving and Interchange DTD v1.0 20120330//EN" "JATS-archivearticle1.dtd">
<article xmlns:xlink="http://www.w3.org/1999/xlink">
  <front>
    <journal-meta />
    <article-meta>
      <title-group>
        <article-title>Inherently Reciprocal Predicates - Do They Exist At All?</article-title>
      </title-group>
      <contrib-group>
        <contrib contrib-type="author">
          <string-name>Markéta Lopatková</string-name>
          <xref ref-type="aff" rid="aff0">0</xref>
        </contrib>
        <contrib contrib-type="author">
          <string-name>Václava Kettnerová</string-name>
          <xref ref-type="aff" rid="aff0">0</xref>
        </contrib>
        <aff id="aff0">
          <label>0</label>
          <institution>Charles University, Faculty of Mathematics and Physics, Institute of Formal and Applied Linguistics</institution>
          ,
          <addr-line>Malostranské nám. 25, Prague, 118 00</addr-line>
          ,
          <country country="CZ">Czech Republic</country>
        </aff>
      </contrib-group>
      <abstract>
        <p>Reciprocity and mutuality represent phenomena intensively studied both in linguistics and in formal semantics. Here we focus on inherently reciprocal predicates, i.e., predicates bearing the feature of mutuality in their lexical meaning. While there is a consensus generally accepted by linguists that such predicates most likely exist in all natural languages, a reliable criterion defining this type of predicates is still missing. The usual definition - adopted from formal semantics - appears to be too strong for natural language predicates, leading to the contraintuitive conclusion that such predicates do not exist at all. Here, inspired by [1], we redefine the criterion. The proposed weaker definition allows us to group predicates of a natural language in a more appropriate way, reflecting their syntactic behavior and thus enabling their adequate lexicographic description.</p>
      </abstract>
      <kwd-group>
        <kwd>eol&gt;mutuality</kwd>
        <kwd>symmetry</kwd>
        <kwd>reciprocity</kwd>
        <kwd>inherently reciprocal predicates</kwd>
        <kwd>lexical meaning</kwd>
      </kwd-group>
    </article-meta>
  </front>
  <body>
    <sec id="sec-1">
      <title>1. Motivation</title>
      <p>
        the relation between  and  is the same as the relation
between  and ."
Symmetry, mutuality, and reciprocity represent closely As attested cross-linguistically, most mutual situations
related concepts, which have been widely discussed by are explicitly expressed by a specific morphosyntactic
patboth linguists and logicians since the 1970s, see esp. [2, 3, tern, usually referred to as reciprocal constructions, where
4, 5] and an immense efort underlying two big collections mutuality afects two syntactic positions (as exemplified
of studies [
        <xref ref-type="bibr" rid="ref6 ref7">6, 7</xref>
        ]. In the domain of natural languages in (1a) and (2a,b); compare them with the non-reciprocal
(NL), these concepts are relevant for diferent layers of patterns used in (1b) and (2c)). Morphosyntactically,
rethe language description. Here, we reserve the term ciprocal constructions are characterized by the
pluralizamutuality for the semantic layer and reciprocity for the tion of the subject position where both participants in
syntactic and morphosyntactic layer. Put it diferently, the mutual relation are expressed2 and typically by the
when we refer to the semantic feature, we use the term presence of an anaphoric expression coreferring with the
mutuality, while referring to a set of means encoding expression in the subject (as the expression each other in
this semantic feature, we use the term reciprocity, as English (1a) and the reflexive personal pronoun
se/si/seproposed, e.g., in [
        <xref ref-type="bibr" rid="ref8">8</xref>
        ].1 be/sobě/sebou or the bipartite expression jeden – druhý in
      </p>
      <p>
        In the semantic characterization of predicates of nat- Czech, see examples (2a) and (2b), respectively).
ural languages (NL predicates henceforth), the notion
of a situation plays a key role. Each NL predicate de- (1) a. Aisha and Pedro pinched each other. ([
        <xref ref-type="bibr" rid="ref8">8</xref>
        ])
notes a situation (i.e., a state or an action) characterized b. Aisha pinched Pedro.
by a set of participants and relationships between these
participants. The situation denoted by the predicate can
be syntactically structured in various ways. According
to Haspelmath [
        <xref ref-type="bibr" rid="ref8">8</xref>
        ]: "A mutual situation can be defined
as a situation with two or more participants (, , . . . )
in which for at least two of the participants  and ,
(2) a. Dva doplavou současně, začnou spolu zuřivě zápasit,
štípat se a navzájem se topit. (SYN v10)3
b. Dva doplavou současně, začnou spolu zuřivě zápasit,
štípat jeden druhého a navzájem se topit.
‘The two swim up at the same time, and begin to
wrestle furiously, pinch each other and drown each other.’
c. . . . plavec A štípá plavce B, . . .
‘. . . the swimmer A pinches the swimmer B, . . . ’
ITAT 2023: Information Technologies – Applications and Theory, 2023
$ lopatkova@ufal.mf.cuni.cz (M. Lopatková); However, we can observe that some NL predicates
exket0tn0e0r0o-v00a0@2u-3f8a3l.3m-9f.c6u1n1i.(cMz .(VL.oKpaettktnoevráo)v;á0)000-0001-9694-1304 press mutuality between two of their participants without
(V. Kettnerová) any overt morphosyntactic marking. For example,
sen© 2023 Copyright for this paper by its authors. Use permitted under Creative Commons License tence (3a) expresses that Evka ‘Eve’ and jeden Angličan
1SCPWrEooUorckReshdominpgsIhSetpN:/c1e6u1r3-w-0sa.o7r3g utAChttEorirbUustR,iohnWo4.0woInretekvrnseahrti,oonspaul (bPCsCruoBmYce4e.0e).dalilntghse(sCeEcUonRc-eWptSs.ourngd)er the term ‘one of the Englishmen’ are in the mutual relation, ‘they
reciprocity, which is used for both meaning and forms expressing 2Less frequently, the object position is involved, as, e.g., in example
this meaning. Others use the term symmetry instead, esp. in the (4) below.
realm of formal semantics [
        <xref ref-type="bibr" rid="ref1 ref9">9, 1</xref>
        ]; here we avoid this term (as possible) 3SYN refers to the Czech National Corpus, https://www.korpus.cz/.
since natural language predicates rarely denote fully symmetric Examples without sources are made-up, illustrating alternative
situations, as illustrated below. syntactic structures.
play together / against each other’, even though there
is no overt reciprocal marking; the mutual
interpretation is supported by the fact that (3a) implies reciprocal
construction (3b). Similarly, sentence (4a) expresses the
mutual relation between humor ‘humor’ and drsné scény
‘harsh scenes’ (‘they alternate each other’) without the
presence of reciprocal marking, compare (4a) with
reciprocal construction (4b). Typological studies have revealed
that – despite not being numerous – such predicates
appear most likely in all natural languages [
        <xref ref-type="bibr" rid="ref10">10</xref>
        ]. We will
call them inherently reciprocal predicates here [
        <xref ref-type="bibr" rid="ref11 ref12">11, 12</xref>
        ].4
predicates
predicates
that can
express
mutuality
predicates that
cannot express
mutuality (Type C.)
inherently reciprocal predicates
≈ expressing mutuality both
in reciprocal constructions and
in constructions without over reciprocal marking
(Type A.)
predicates expressing mutuality only in
reciprocal construction (Type B.)
(3) a. Spatřil jsem Evku v hedvábných bílých plavkách, jak
hraje ping-pong proti jednomu Angličanovi a druhý
jim soudcuje. (SYN v10, modified)
‘I saw Evka in a silk white swimsuit playing
pingpong against one of the Englishmen and the other
judging them.’
b. Spatřil jsem Evku a jednoho Angličana, jak proti sobě
hrají ping-pong a druhý jim soudcuje.
‘I saw Evka and one of the Englishmen playing
pingpong against each other and the other judging them.’
shown that predicates intuitively classified as inherently
reciprocal ones require a less degree of linguistic marking
in reciprocal constructions, see esp. [
        <xref ref-type="bibr" rid="ref13">13</xref>
        ] and for Czech
[
        <xref ref-type="bibr" rid="ref14">14</xref>
        ]. Thus the ‘inherently reciprocal’ feature appears to
be a syntactically relevant semantic feature, and as such,
it should be captured in a syntactic lexicon.
      </p>
      <p>
        There is a commonly cited definition of predicates
expressing mutuality, adopted from formal semantics; we
discuss it in Section 2. We argue that this definition is
(4) a. A stejně tak neváhá bleskurychle střídat humor s fully satisfied only by reciprocal constructions. In
convelmi drsnými scénami, . . . (SYN v10) trast, inherently reciprocal predicates in constructions
‘And he doesn’t hesitate to alternate humor with very without overt reciprocal marking, although they express
harsh scenes at lightning speed, . . . ’ mutuality as well, does not reliably meet this criterion, as
b. A stejně tak neváhá bleskurychle střídat humor a discussed in Section 3.5 As a result, based on this
definivelmi drsné scény, . . . tion, an inventory of NL predicates expressing mutuality
‘And he doesn’t hesitate to alternate humor and very can be established – but it is not capable to single out
harsh scenes with each other at lightning speed, . . . ’ those predicates that are inherently reciprocal from this
(5) a. Avšak každé válce předcházejí manévry, . . . (SYN inventory. To identify inherently reciprocal predicates,
v10) we thus propose a modified version of the given criterion,
‘However, every war is preceded by maneuvers, . . . ’ on the basis of which we introduce more subtle
classifib. *Válka a manévry se vzájemně předcházejí. cation of predicates expressing the relation of mutuality
‘*The war and maneuvers precede each other.’ (Section 4). Lastly, we introduce how this classification
As a result, NL predicates fall into three diferent may be applied in a syntactic lexicon, namely in the
vagroups with respect to mutuality, as shown in Figure 1: lency lexicon VALLEX (Section 5).
on the first level, we can distinguish those predicates
that can express mutuality (as the verb štípat ‘pinch’ in 2. Reciprocal (Symmetric)
(1), (2), the verb hrát ‘play’ in (3), and the verb střídat
‘alternate’ in (4)) and those that cannot. The latter can Predicates in Formal Semantics
be exemplified, e.g., by the verb předcházet ‘to precede’,
compare (5a) and (5b). Those predicates that can express A formal definition of reciprocity can be found in a
nummutuality further split into those that express mutual- ber of studies from formal semantics, one of the most
ity only in reciprocal constructions (exemplified by the often cited ones is given in [
        <xref ref-type="bibr" rid="ref2">2</xref>
        ]. Here the following
converb štípat ‘pinch’ in (1)) and those that beside reciprocal dition of strong reciprocity for the so-called elementary
constructions create constructions expressing mutuality reciprocal sentences  is introduced ( denotes a set
without explicit morphosyntactic marking, referred here of cardinality at least 2,  is a relation on  × , and 
to as inherently reciprocal predicates (as illustrated by is a reciprocal element, e.g., each other):
the verb hrát (proti někomu) in (3) and the verb střídat in
(4)). (∀,  ∈ ) ( ̸=  → (, )) (*)
      </p>
      <sec id="sec-1-1">
        <title>Distinguishing the three groups of NL predicates with</title>
        <p>
          respect to mutuality is not an end in itself: it has been
Thus, as paraphrased in [
          <xref ref-type="bibr" rid="ref4">4</xref>
          ], for specific substitutions
of values  and  (,  ∈ ) for the variables  and :
4The terms (inherently) symmetric(al) predicates (as, e.g., in [
          <xref ref-type="bibr" rid="ref1 ref4">4, 1</xref>
          ]) or
lexical reciprocals [
          <xref ref-type="bibr" rid="ref6">6</xref>
          ] are also used for the same type of predicates.
        </p>
      </sec>
      <sec id="sec-1-2">
        <title>5This paper follows the study [15], where an interested reader can</title>
        <p>ifnd a more detailed linguistic observations.
(, ) ↔ (, ). [4, p. 273] relate this asymmetry to “power, control,
ini</p>
        <p>
          This condition is satisfied, e.g., by sentence (6) with tiative, or involvement”, Panevová [
          <xref ref-type="bibr" rid="ref16">16</xref>
          ] stresses the role
 identifying a set of people in the house,  being the of homogeneity of the involved participants when
disrelation denoted by the binary predicate to know (i.e., cussing (8).
know(X, Y ) = X knows Y ), and  being the reciprocal
element each other. In Czech, the role of the reciprocal (8) a. Radotín se sloučil s Prahou před 30 lety. ([
          <xref ref-type="bibr" rid="ref16">16</xref>
          ])
element is taken on by the reflexive pronoun, as se in ‘3R0aydeoatírns (aagsom.’ all city near Prague) merged with Prague
(2a) and sobě in (3b), or by the bipartite expression jeden b. ?? Praha se sloučila s Radotínem před 30 lety. ([
          <xref ref-type="bibr" rid="ref16">16</xref>
          ])
– druhý ‘each other’, as in (2b); all sentences (2a), (2b) ?? ‘Praha merged with Radotín 30 years ago.’
and (3b) represent reciprocal constructions, see Section
1, and fulfill the criterion (*).
        </p>
      </sec>
    </sec>
    <sec id="sec-2">
      <title>3. Source of Asymmetry with NL</title>
    </sec>
    <sec id="sec-3">
      <title>Predicates</title>
      <p>
        (6) People in this house know each other. ([
        <xref ref-type="bibr" rid="ref5">5</xref>
        ])
(7) John met Mary at the station. ([
        <xref ref-type="bibr" rid="ref4">4</xref>
        ])
      </p>
      <sec id="sec-3-1">
        <title>Gleitman at al. [17] conducted an extensive study of</title>
        <p>Despite the prerequisite of the presence of reciprocal the asymmetry in constructions of inherently reciprocal
marker , some reciprocal constructions without such predicates (symmetrical in the authors’ terminology, e.g.
marker satisfy this condition as well, as (4b). In these similar, equal, meet, near). The series of psycholinguistic
reciprocal constructions, the pluralized subject or direct experiments revealed that experimental subjects
interobject, in which both participants involved in mutuality pret the pairs of sentences with swapped participants
are expressed, is a suficient marker of the mutual relation as, e.g., (9a) and (9b), in a diferent way, although they
between participants. assess the respective predicates as symmetrical, i.e.,
bear</p>
        <p>However, as exemplified in (3a), (4a), and (7), inherent ing mutuality in their meaning (inherently reciprocal
reciprocal predicates can express mutuality not only in re- in our terminology). The results of their study thus
reciprocal constructions but also in constructions without vealed a paradoxical situation when inherently reciprocal
any overt morphosyntactic marking (i.e., without the plu- predicates create constructions with the asymmetrical
ralized subject or direct object and without the reflexive interpretation. Consequently, these constructions do not
pronoun or the expression jeden – druhý). Distinguish- met the condition (**).
ing inherently reciprocal predicates thus requires such As the asymmetry manifests itself just in constructions
a criterion that identifies mutuality of participants also of inherently reciprocal predicates without overt
reciproin the constructions that are not overtly marked. cal marking (i.e., in constructions with the afected
partic</p>
        <p>
          As the first step, we adopt the concept of a mutual ipants distributed into separate syntactic positions, as in
situation, as proposed in [
          <xref ref-type="bibr" rid="ref8">8</xref>
          ], see Section 1, and relate it to (3a) or (7)), the authors [
          <xref ref-type="bibr" rid="ref17">17</xref>
          ] put forward the hypothesis
a predicate. We will say that predicate  with participants that it is the asymmetry of syntactic constructions which
 and  satisfies the condition of mutuality if and only if causes the asymmetric interpretation of inherently
rethe situation denoted by the predicate involves (at least) ciprocal predicates. In particular, the syntactic position
two diferent participants and for the values ,  of these of participants (i.e., their distribution in the subject and
participants, the following equivalence holds: object positions) imposes their interpretation as either
Figure (Variant) or Ground (Referent): the subject
posi (, ) ↔  (, ) (**) tion determines the status of a participant as Figure while
The condition (**) now covers also (4a) as we can object positions as Ground.
consider this construction and the construction with For example, in (9a) with the inherently reciprocal
swapped participants (i.e., the pairs střídat humor s velmi predicate similar, expressing comparison of two entities,
drsnými scénami ‘to alternate very harsh scenes with hu- North Korea in subject is interpreted as Figure, i.e., the
mor’ and střídat velmi drsné scény s humorem ‘to alternate compared entity, and China in the object position as
humor with very harsh scenes’) as paraphrases with the Ground, i.e., the more prominent entity to which the
forsame truth conditions. mer one is compared. With swapped participants in (9b),
        </p>
        <p>
          However, the situation is complicated by the fact that their interpretation with respect to the Figure/Ground
in many cases constructions of inherently reciprocal pred- distinction is reversed, which underlies interpretive
dificates without overt reciprocal marking show a certain ferences between (9a) and (9b). The ofered analysis –
asymmetry between participants, as pointed out by some that asymmetry of (9a) and (9b) is brought about by the
authors, see esp. [
          <xref ref-type="bibr" rid="ref12 ref16 ref17 ref4">16, 17, 4, 12</xref>
          ]. This asymmetry can be ex- asymmetric syntactic structuring of the sentences – is
emplified by the pairs in (8), where only (8a) with Radotín also supported by the fact that the asymmetry disappears
in subject is fully acceptable. In such cases, the condition in constructions with both participants structured in the
(**) can be hardly applied. While König and Kokutani same way, i.e., in the single subject position, as in (9c).
(9) a. North Korea is similar to Red China. ([17, p. 322])
b. Red China is similar to North Korea. ([17, p. 322])
c. North Korea and Red China are similar to each other.
        </p>
        <p>([17, p. 326])</p>
        <p>The same interpretive diferences between
constructions with swapped participants formed by inherently
reciprocal predicates are attested in Czech as well, as
illustrated, e.g., by predicates denoting comparison, see Moreover, in constructions of inherently reciprocal
examples (10a) and (10b) with the adjective srovnávaný, predicates with participants distributed in diferent
synand space relations, see examples (11a) and (11b) with the tactic positions, the asymmetry between participants can
verb sousedit, and by action predicates, see examples (12a) be further stressed by context (e.g., by an adverbial
modiand (12b) with the verb setkat se. In all these examples, fying the verb), and thus the mutual relation between the
the diference between sentences a. and b. can be ex- involved participants can be even blocked. For example,
plained on the basis of the Figure/Ground interpretation although the verb bojovat ‘to fight’ can be considered
of the afected participants, implied by their syntactic as an inherently reciprocal verb in all examples in (14),
positions in the sentences. only (14d) with the involved participants structured in
the same position of the (plural) subject is fully
symmetrical. While (14b) and (14c) are more or less exchangeable
in most contexts, the modal verb chtít ‘to want’ in (14a)
introduces asymmetry to the construction.
(13) a. . . . a kat . . . jednou ranou oddělil jeho hlavu od těla.</p>
        <p>(SYN v10)
‘. . . and the executioner separated his head from his
body with a single blow.’
b. . . . a kat jednou ranou oddělil jeho tělo od hlavy.
‘. . . and the executioner separated his body from his
head with a single blow.’
(14) a. . . . tento mladíček tvrdí, že by chtěl bojovat s</p>
        <p>Goliášem. (SYN v10)
‘. . . . . . this young man claims he would like to fight
Goliath’
b. . . . tento mladíček bude bojovat s Goliášem.
‘. . . this young man will fight Goliath.’
c. . . . Goliáš bude bojovat s tímto mladíčkem.
‘. . . Goliath will fight this young man.’
d. . . . tento mladíček a Goliáš spolu budou bojovat.
‘. . . this young man and Goliath will fight together.’
(10) a. Galský Taranis je "hromovládce", srovnávaný
Římany s Jupiterem. (SYN v10)
‘The Gallic Taranis is the "thunderlord", compared by
the Romans to Jupiter.’
b. Jupiter je "hromovládce", srovnávaný Římany s
galským Taranisem.
‘Jupiter is the "thunderlord", compared by the Romans
to the Gallic Taranis.’
(11) a. Nevelká stavba . . . přímo sousedí s domem Le
Corbusierova bratra Alberta Jeannereta a jeho rodiny. (SYN
v10)
‘The small building . . . is directly adjacent to the house
of Le Corbusier’s brother Albert Jeanneret and his
family.’
b. Dům Le Corbusierova bratra Alberta Jeannereta
a jeho rodiny . . . přímo sousedí s nevelkou stavbou.
‘The house of Le Corbusier’s brother Albert Jeanneret
and his family . . . is directly adjacent to the small
building.’</p>
        <p>
          To summarize, although inherently reciprocal predicates
are supposed to carry the feature of mutuality between
two of their participants in their lexical meaning (and
(12) a. Krátce před smrtí se setkal s Elizabeth Stoneovou, . . . thus their mutual interpretation should not depend on
(SYN v10) special syntactic constructions), structuring these
par‘He met Elizabeth Stone shortly before his death, . . . ’ ticipants in diferent syntactic positions results in the
b. Krátce před jeho smrtí se s ním setkala Elizabeth asymmetry, which is manifested by a meaning shift when
Stoneová, . . . the afected participants are interchanged in the involved
‘Elizabeth Stone met him shortly before his death, . . . ’ positions. Put it diferently, “symmetry is a property
Gleitman et al. [
          <xref ref-type="bibr" rid="ref17">17</xref>
          ] limited their attention to con- of lexical items and has no special syntax”, in terms of
structions of those inherently reciprocal adjectives and [
          <xref ref-type="bibr" rid="ref17">17</xref>
          ], constructional properties then bring about lesser or
verbs where the subject position is involved in mutuality. bigger asymmetry. As a result, criterion (**) relying on
However, similar interpretive diferences can be observed the preservation of truth conditions when the afected
also in constructions of those inherently reciprocal pred- participants permuted cannot be used for identification
icates where the participants in the mutual relation are of inherently reciprocal predicates. Otherwise, it would
distributed in two object positions, as illustrated by ex- lead to an inescapable conclusion that there are no
inamples (4) above and (13). herently reciprocal predicates at all, contradicting the
        </p>
        <p>Again, the interpretive diference between (13a) and generally accepted hypothesis.
(13b) can be accounted for by the Figure/Ground
distinction: namely, the direct object position is associated
with the Figure interpretation and the indirect object in
the form of the prepositional case od+genitive with the
Ground interpretation.</p>
        <p>
          The usually cited condition of strong reciprocity (*) [
          <xref ref-type="bibr" rid="ref4">4</xref>
          ]  (, ) &amp;  (, ) (†)
takes this property as a property of the given predicate, So similarly to (**), the condition (†) makes is possible
i.e., it should be apply to all instances of this predicate – to consider a predicate together with a particular choice
to any lexical choice of the afected participants, in any of its participants involved in mutuality. The diference
syntactic construction, in any context.6 As such, it is too lies in the fact that (†) generalizes over individual triplets
strong for classification of NL predicates. On the other  (, ) and examines thus the property of the predicate
hand, the condition of mutuality (**), inspired by [
          <xref ref-type="bibr" rid="ref8">8</xref>
          ], is  itself (instead of assessing individual instances in
isoapplied to a predicate with a particular lexical choice of lation).
the afected participants (i.e., with a particular choice In opposition to the previous approaches, where
symof their values in [
          <xref ref-type="bibr" rid="ref4">4</xref>
          ] and (*)), that is, to the particular metry and asymmetry are treated as sharp concepts (and
instances of the predicate. As such, it is too weak, and thus reciprocal predicates must be symmetrical in all
hence not helpful in classifying NL predicates with re- their occurrences), in [
          <xref ref-type="bibr" rid="ref1 ref18">18, 1</xref>
          ] symmetry is treated as
spect to the feature of mutuality in their lexical meaning. a graded property. Still, the authors distinguish four
Neither of these criterion in isolation is thus suitable for distinct groups of predicates:
distinguishing the set of inherently reciprocal predicates. (i) predicates expressing mutuality for all instances
of participants (‘purely symmetric predicates’ in
4.1. The Concept of Symmetry As Graded their terminology), e.g., meet, similar, sibling;
Property: Symmetrical Instances (ii) predicates expressing mutuality, which, however,
can be blocked (‘mixed symmetric predicates’ in
their terminology; there are both symmetric and
asymmetric instances of the given predicates),
e.g., kiss, opposed to, friend;
(iii) predicates not expressing mutuality but enabling
symmetric instances (‘mixed asymmetric
predicates’ in their terminology), e.g., drawn, kill, hit;
(iv) predicates asymmetric for all their instances
(‘purely asymmetric predicates’), e.g., contain,
father, below.
        </p>
      </sec>
      <sec id="sec-3-2">
        <title>The problem with possible asymmetry of predicates in</title>
        <p>
          tuitively classified as inherently reciprocal predicates is
addressed by Partee [
          <xref ref-type="bibr" rid="ref18">18</xref>
          ] and Gleitman and Partee [
          <xref ref-type="bibr" rid="ref1">1</xref>
          ].
The authors show that the usual linguistic test
indicating "symmetrical" predicates,7 the logical definition, and
judgments ofered by native speakers are often in
contradiction. Gleitman and Partee [
          <xref ref-type="bibr" rid="ref1">1</xref>
          ] thus propose to consider
– in addition to the relations satisfying the condition of
strong reciprocity (*) as a property of the whole relation
, which must be valid for all instances/values of ,  –
also relations symmetrical only for some instances:
• A symmetrical instance of a relation  is a pair
,  such that (, ) &amp; (, ).
• An asymmetrical instance of a relation  is a pair
,  such that (, ) &amp; ¬(, ).
        </p>
        <p>Predicates intuitively understood as inherently
reciprocal should fall under the first two types of predicates
in the above given classification. However, the existence
of ‘purely symmetric predicates’ seems to be
questionable since for each of these predicates, one can typically
ifnd contexts in which mutuality between the afected
participants is canceled.</p>
      </sec>
    </sec>
    <sec id="sec-4">
      <title>4. Revised Approaches to</title>
    </sec>
    <sec id="sec-5">
      <title>Symmetrical Predicates</title>
      <p>participants such that  and ,  satisfy the following
condition (†):</p>
      <p>
        Based on this consideration, we can then ask whether,
for a given predicate  , there is any instance ,  of its
6Though some authors point out that this condition is fully true
only for reciprocal constructions, see esp. [
        <xref ref-type="bibr" rid="ref4">4</xref>
        ] who refer to the
predicates satisfying this condition as to symmetric predicates:
“Perfect symmetry is only expressed by reciprocal constructions An interesting approach to the concept of predicates
exand this symmetry is often iconically signalled by the fact that the pressing mutuality can be found in [
        <xref ref-type="bibr" rid="ref19">19</xref>
        ] – Winter points
relevant semantic arguments are encoded by the same grammatical out that reciprocity as a linguistic phenomenon is often
7rTehlaetiaount.h”o[4rs, ph.a2v7e3]t.he so-called Litmus Test in mind, examining confused with symmetry as a concept belonging to the
pairs of constructions like John and Bill are similar/match/hug and domain of logic. To avoid this, he proposes to start (as in
*John is similar/match/hug where the constructions with plural sub- [
        <xref ref-type="bibr" rid="ref1">1</xref>
        ]) with those binary predicates that have also the
soject should express, roughly speaking, the mutual relation between called unary-collective forms (illustrated by examples as
John and Bill. A and B dated / are identical / are cousins.)8 and compare
We can see certain limitation here as (like in other logic-oriented
works) the authors do not distinguish between particular construc- 8Based on the provided examples, the unary-collective forms of
tions with a predicate and the predicate as an abstract unit bearing predicates correspond to those that satisfy the Litmus Test, see
some lexical properties as such (regardless of diferent syntactic footnote 7; i.e., they have plural subject and they express the mutual
construction it can create). relation between participants in the subject position.
      </p>
      <sec id="sec-5-1">
        <title>4.2. Unary-Collective Predicates And</title>
        <p>
          Their Binary Counterparts
5. Inherently Reciprocal Predicates
in Lexicographic Practice
them to their basic forms (i.e., to the forms illustrated
by A dated B / A is identical to B / A is B’s cousin.). The
author observes that for predicates that are symmetric
from the logical point of view (i.e., predicates with in- Let us repeat that inherently reciprocal predicates are
terchangeable arguments, cf. condition (*)), these two characterized by the possibility to express mutuality
betypes of constructions can serve as paraphrases with the tween participants in constructions that are not overtly
same meaning (contrary to asymmetric predicates). For marked, see esp. examples (3) and (4) in Section 1.
Howexample, in (16) with the asymmetric predicate talk to, ever, the mutuality condition (**) is met only by some
the first sentence “can be interpreted reciprocally, but instances of these predicates, as we have seen in Section
also distributively: Dan talked, and Sue talked as well.” 3. The existence of the instances violating this
condi[
          <xref ref-type="bibr" rid="ref19">19</xref>
          ]; the two constructions cannot be thus assessed as tion leads to the conclusion that – contrary to linguistic
paraphrases (contrary to all the sentences in (15) which intuition – natural languages do not have any
inherhave only the reciprocal interpretation). ently reciprocal predicates. This well known limitation
inspired researchers to seek for a weaker but reliable
(15) Sue and Dan dated. ↔ Sue dated Dan. criterion, as discussed in Section 4, making it possible
(↔ Dan dated Sue.) to examine not only behavior of predicates in
individ(16) Sue and Dan talked. ual constructions but (i) abstract from concrete syntactic
↮ Sue talked to Dan. &amp; Dan talked to Sue. constructions and (ii) relax the requirement of symmetry
to individual instances of predicates, as in [
          <xref ref-type="bibr" rid="ref1 ref19">1, 19</xref>
          ]. Only
        </p>
        <p>Inspired by this observation, the author define the such a criterion is applicable in lexicographic description,
criterion of plain symmetry (‡), which is applied to an which takes a predicate as a language unit representing
unary-collective predicates  over sums of (singular) all its syntactic constructions.
entities in  (with cardinality 2 (or more)) and a binary These considerations led us to define inherently
recippredicate  alternating with  , ranging over pairs of rocal predicates as such predicates for which there are
entities in : attested the instances of constructions expressing
mutuality without overt reciprocal marking. This approach is
(‡) based on the following observation:
∀,  ∈ ,  ̸=  :
 ( + ) ↔ (, ) &amp; (, )</p>
        <sec id="sec-5-1-1">
          <title>Criterion. If for a given predicate there is (even a single)</title>
          <p>In other words, Winter limits the requirement of sym- example where mutuality is expressed in the construction
metry only to constructions logically equivalent to spe- that is not overtly marked then the feature of
mutualcific constructions with plural subject and mutual mean- ity must be already contained in the lexical meaning of
ing – the predicate  is not necessarily symmetric in the given predicate (at least as a potential feature) and
all its instances. The proposed criterion serves as an- the predicate thus comes under the class of inherently
other interesting approach addressing the problem of the reciprocal predicates.
discussed inadequacy of the logic-based condition (*). To put it diferently, the criterion abstracts from
prop</p>
          <p>
            Winter applies the condition (‡) to English (and to erties brought about by specific morphosyntactic
cona small set of other languages like Hebrew). However, structions (constructions with overt reciprocal marking);
it is not clear how to adapt such approach to typologi- instead, unlike [
            <xref ref-type="bibr" rid="ref1">1</xref>
            ], it examines lexical properties of
excally diferent languages. For example, Czech attests that amined predicates themselves.
predicates intuitively classified as inherently reciprocal This approach has been employed in the VALLEX
lexipredicates typically form constructions parallel to those con9 [
            <xref ref-type="bibr" rid="ref20">20</xref>
            ], where three types of verbs are distinguished
considered as unary-collective ones, as, e.g., diskutovat with respect to reciprocity:
(s někým) ‘discuss with somebody’, which, however, can
be interpreted distributively as well, compare (17a) and
(17b).
(17) a. . . . během malování portrétu Masaryka diskutovali
malíř a prezident i o Komenském . . . (SYN v10)
‘. . . during the painting of Masaryk’s portrait, the
painter and the president also discussed Comenius
. . . ’
b. Britové a Němci diskutují o veřejné službě. (SYN
v10)
‘British and Germans discuss about public service.’
          </p>
          <p>
            A. inherently reciprocal predicates, i.e., predicates
bearing the feature of mutuality in their lexical
meaning, as defined in the previous paragraph
(e.g., bojovat ‘to fight’, diskutovat ‘to discuss’, and
kombinovat ‘to combine’); this type comprises
predicates that typically fall under (i)-(ii) in the
taxonomy defined in [
            <xref ref-type="bibr" rid="ref1">1</xref>
            ], i.e., predicates
expressing mutuality for at least some of their instances
even without overt reciprocal marking, see
Section 4.1;
          </p>
        </sec>
      </sec>
    </sec>
    <sec id="sec-6">
      <title>6. Conclusion</title>
      <p>While typological studies have revealed that inherently
reciprocal predicates (i.e., predicates bearing the feature
of mutuality in their lexical meaning) appear most likely
in all natural languages, available formal criteria
striving to define such predicates are not satisfactory as they
are either too strict (applied to the whole predicate in
its all instances) or too weak (applied only to
individual instances of a predicate) (as exemplified on Czech
predicates). In this paper, we have thus proposed the
modified version of such criteria. The proposed
criterion abstracts from particular syntactic constructions of
examined predicates (and from features brought about
by these constructions). Thus it allows us to focus on
the features contained already in their lexical meaning.
Based on this criterion we have introduced a more subtle
classification of predicates expressing the relation of
mutuality. This classification has been applied in the valency
lexicon VALLEX.</p>
      <sec id="sec-6-1">
        <title>Acknowledgment</title>
        <sec id="sec-6-1-1">
          <title>The work described herein has been supported by the</title>
          <p>grant Language Understanding: From Syntax to
Discourse (LUSyD) of the Czech Science Foundation (GAČR),
No. GX20-16819X.</p>
          <p>The work on the VALLEX lexicon has been using data
and tools provided by the LINDAT /CLARIAH-CZ Research
Infrastructure (https://lindat.cz), supported by the
Ministry of Education, Youth and Sports of the Czech
Republic (project No. LM2023062).
lemmas
lexical units</p>
          <p>lexemes</p>
        </sec>
      </sec>
    </sec>
  </body>
  <back>
    <ref-list>
      <ref id="ref1">
        <mixed-citation>
          [1]
          <string-name>
            <given-names>L. R.</given-names>
            <surname>Gleitman</surname>
          </string-name>
          ,
          <string-name>
            <given-names>B. H.</given-names>
            <surname>Partee</surname>
          </string-name>
          , Understanding Symmetrical Predicates,
          <year>2022</year>
          .
          <article-title>(Talk presented at the Fred Jelinek Seminar Series</article-title>
          , Institute of Formal and applied Linguistics).
        </mixed-citation>
      </ref>
      <ref id="ref2">
        <mixed-citation>
          [2]
          <string-name>
            <given-names>D. T.</given-names>
            <surname>Langendoen</surname>
          </string-name>
          ,
          <article-title>The logic of reciprocity, Linguistic Inquiry 9 (</article-title>
          <year>1978</year>
          )
          <fpage>177</fpage>
          -
          <lpage>197</lpage>
          .
        </mixed-citation>
      </ref>
      <ref id="ref3">
        <mixed-citation>
          [3]
          <string-name>
            <given-names>M.</given-names>
            <surname>Dalrymple</surname>
          </string-name>
          ,
          <string-name>
            <given-names>M.</given-names>
            <surname>Kanazawa</surname>
          </string-name>
          ,
          <string-name>
            <given-names>Y.</given-names>
            <surname>Kim</surname>
          </string-name>
          ,
          <string-name>
            <given-names>S.</given-names>
            <surname>McHombo</surname>
          </string-name>
          ,
          <string-name>
            <given-names>S.</given-names>
            <surname>Peters</surname>
          </string-name>
          ,
          <article-title>Reciprocal expressions and the concept of reciprocity</article-title>
          ,
          <source>Linguistics and Philosophy</source>
          <volume>21</volume>
          (
          <year>1998</year>
          )
          <fpage>159</fpage>
          -
          <lpage>210</lpage>
          . doi:
          <volume>10</volume>
          .1023/a:
          <fpage>1005330227480</fpage>
          .
        </mixed-citation>
      </ref>
      <ref id="ref4">
        <mixed-citation>
          [4]
          <string-name>
            <given-names>E.</given-names>
            <surname>König</surname>
          </string-name>
          ,
          <string-name>
            <given-names>S.</given-names>
            <surname>Kokutani</surname>
          </string-name>
          ,
          <article-title>Towards a typology of reciprocal constructions: focus on German and Japanese</article-title>
          ,
          <source>Linguistics</source>
          <volume>44</volume>
          (
          <year>2006</year>
          )
          <fpage>271</fpage>
          -
          <lpage>302</lpage>
          . doi:
          <volume>10</volume>
          .1515/LING.
          <year>2006</year>
          .
          <volume>010</volume>
          .
        </mixed-citation>
      </ref>
      <ref id="ref5">
        <mixed-citation>
          [5]
          <string-name>
            <given-names>E.</given-names>
            <surname>König</surname>
          </string-name>
          ,
          <string-name>
            <given-names>V.</given-names>
            <surname>Gast</surname>
          </string-name>
          ,
          <article-title>Reciprocity and reflexivity - description, typology and theory</article-title>
          , in: E.
          <string-name>
            <surname>König</surname>
            , V. Gast (Eds.), Reciprocals and Reflexives: Theoretical and
            <given-names>Typological</given-names>
          </string-name>
          <string-name>
            <surname>Explorations</surname>
          </string-name>
          , Mouton de Gruyter, Berlin - New York,
          <year>2008</year>
          , pp.
          <fpage>1</fpage>
          -
          <lpage>32</lpage>
          .
        </mixed-citation>
      </ref>
      <ref id="ref6">
        <mixed-citation>
          [6]
          <string-name>
            <given-names>V. P.</given-names>
            <surname>Nedjalkov</surname>
          </string-name>
          (Ed.),
          <string-name>
            <surname>Reciprocal</surname>
            <given-names>Constructions</given-names>
          </string-name>
          , John Benjamins Publishing Company, Amsterdam - Philadelphia,
          <year>2007</year>
          .
        </mixed-citation>
      </ref>
      <ref id="ref7">
        <mixed-citation>
          [7]
          <string-name>
            <given-names>E.</given-names>
            <surname>König</surname>
          </string-name>
          , V. Gast (Eds.), Reciprocals and Reflexives: Theoretical and
          <string-name>
            <given-names>Typological</given-names>
            <surname>Explorations</surname>
          </string-name>
          , Mouton de Gruyter, Berlin - New York,
          <year>2008</year>
          .
        </mixed-citation>
      </ref>
      <ref id="ref8">
        <mixed-citation>
          [8]
          <string-name>
            <given-names>M.</given-names>
            <surname>Haspelmath</surname>
          </string-name>
          ,
          <article-title>Further remarks on reciprocal constructions</article-title>
          , in: V. P. Nedjalkov (Ed.),
          <string-name>
            <surname>Reciprocal</surname>
            <given-names>Constructions</given-names>
          </string-name>
          , John Benjamins, Amsterdam - Philadelphia,
          <year>2007</year>
          , pp.
          <fpage>2087</fpage>
          -
          <lpage>2115</lpage>
          .
        </mixed-citation>
      </ref>
      <ref id="ref9">
        <mixed-citation>
          [9]
          <string-name>
            <given-names>L. R.</given-names>
            <surname>Gleitman</surname>
          </string-name>
          ,
          <string-name>
            <given-names>B. H.</given-names>
            <surname>Partee</surname>
          </string-name>
          , Symmetry and
          <string-name>
            <given-names>Symmetrical</given-names>
            <surname>Predicates</surname>
          </string-name>
          ,
          <year>2020</year>
          .
          <article-title>(UMass Blog, updated handout</article-title>
          ).
        </mixed-citation>
      </ref>
      <ref id="ref10">
        <mixed-citation>
          [10]
          <string-name>
            <given-names>J. P.</given-names>
            <surname>Knjazev</surname>
          </string-name>
          ,
          <article-title>Lexical reciprocals as a means of expressing reciprocal situations</article-title>
          , in: V. P. Nedjalkov (Ed.),
          <string-name>
            <surname>Reciprocal</surname>
            <given-names>Constructions</given-names>
          </string-name>
          , John Benjamins, Amsterdam - Philadelphia,
          <year>2007</year>
          , pp.
          <fpage>115</fpage>
          -
          <lpage>146</lpage>
          . doi:
          <volume>10</volume>
          .1075/tsl.71.
          <year>07knj</year>
          .
        </mixed-citation>
      </ref>
      <ref id="ref11">
        <mixed-citation>
          [11]
          <string-name>
            <given-names>J.</given-names>
            <surname>Panevová</surname>
          </string-name>
          , Znovu o reciprocitě,
          <source>Slovo a slovesnost 68</source>
          (
          <year>2007</year>
          )
          <fpage>91</fpage>
          -
          <lpage>100</lpage>
          .
        </mixed-citation>
      </ref>
      <ref id="ref12">
        <mixed-citation>
          [12]
          <string-name>
            <surname>G. Rákosi,</surname>
          </string-name>
          <article-title>The inherently reflexive and the inherently reciprocal predicate in Hungarian: Each to their own argument structure</article-title>
          , in: E.
          <string-name>
            <surname>König</surname>
            , V. Gast (Eds.), Reciprocals and Reflexives: Theoretical and
            <given-names>Typological</given-names>
          </string-name>
          <string-name>
            <surname>Explorations</surname>
          </string-name>
          , Mouton de Gruyter, Berlin - New York,
          <year>2008</year>
          , pp.
          <fpage>411</fpage>
          -
          <lpage>450</lpage>
          . doi:
          <volume>10</volume>
          .1515/9783110199147.411.
        </mixed-citation>
      </ref>
      <ref id="ref13">
        <mixed-citation>
          [13]
          <string-name>
            <given-names>S.</given-names>
            <surname>Kemmer</surname>
          </string-name>
          , The Middle Voice, John Benjamins, Amsterdam/Philadelphia,
          <year>1993</year>
          .
        </mixed-citation>
      </ref>
      <ref id="ref14">
        <mixed-citation>
          [14]
          <string-name>
            <given-names>V.</given-names>
            <surname>Kettnerová</surname>
          </string-name>
          ,
          <string-name>
            <given-names>M.</given-names>
            <surname>Lopatková</surname>
          </string-name>
          ,
          <article-title>Ke způsobům vyjádření vzájemnosti v češtině</article-title>
          ,
          <source>Slovo a slovesnost 81</source>
          (
          <year>2020</year>
          )
          <fpage>243</fpage>
          -
          <lpage>268</lpage>
          .
        </mixed-citation>
      </ref>
      <ref id="ref15">
        <mixed-citation>
          [15]
          <string-name>
            <given-names>V.</given-names>
            <surname>Kettnerová</surname>
          </string-name>
          ,
          <string-name>
            <given-names>M.</given-names>
            <surname>Lopatková</surname>
          </string-name>
          ,
          <article-title>Proč má ježek bodliny přilepené k tělu, ale nemá tělo přilepené k bodlinám? K charakteristice inherentně recipročních predikátů (Towards characteristics of inherently reciprocal predicates), in: Vzťahy v jazyku - jazyk vo vzťahoch</article-title>
          , submitted, pp.
          <fpage>1</fpage>
          -
          <lpage>11</lpage>
          .
        </mixed-citation>
      </ref>
      <ref id="ref16">
        <mixed-citation>
          [16]
          <string-name>
            <given-names>J.</given-names>
            <surname>Panevová</surname>
          </string-name>
          ,
          <article-title>Česká reciproční zájmena a slovesná valence</article-title>
          ,
          <source>Slovo a slovesnost 60</source>
          (
          <year>1999</year>
          )
          <fpage>269</fpage>
          -
          <lpage>275</lpage>
          .
        </mixed-citation>
      </ref>
      <ref id="ref17">
        <mixed-citation>
          [17]
          <string-name>
            <given-names>L. R.</given-names>
            <surname>Gleitman</surname>
          </string-name>
          ,
          <string-name>
            <given-names>H.</given-names>
            <surname>Gleitman</surname>
          </string-name>
          ,
          <string-name>
            <given-names>C.</given-names>
            <surname>Miller</surname>
          </string-name>
          ,
          <string-name>
            <given-names>R.</given-names>
            <surname>Ostrin</surname>
          </string-name>
          , Similar, and similar concepts,
          <source>Cognition</source>
          <volume>58</volume>
          (
          <year>1996</year>
          )
          <fpage>321</fpage>
          -
          <lpage>376</lpage>
          . doi:
          <volume>10</volume>
          .1016/
          <fpage>0010</fpage>
          -
          <lpage>0277</lpage>
          (
          <issue>95</issue>
          )
          <fpage>00686</fpage>
          -
          <lpage>9</lpage>
          .
        </mixed-citation>
      </ref>
      <ref id="ref18">
        <mixed-citation>
          [18]
          <string-name>
            <given-names>B. H.</given-names>
            <surname>Partee</surname>
          </string-name>
          ,
          <article-title>Symmetry and symmetrical predicates</article-title>
          , in: A. E.
          <string-name>
            <surname>Kibrik</surname>
            ,
            <given-names>V. I.</given-names>
          </string-name>
          <string-name>
            <surname>Belikov</surname>
            ,
            <given-names>B. V.</given-names>
          </string-name>
          <string-name>
            <surname>Dobrov</surname>
            ,
            <given-names>B. V.</given-names>
          </string-name>
          <string-name>
            <surname>Dobrovolsky</surname>
          </string-name>
          (Eds.),
          <article-title>Computational linguistics and intellectual technologies: Papers from the international conference</article-title>
          “Dialogue”,
          <source>Institut Problem Informatiki</source>
          , Moscow,
          <year>2008</year>
          , p.
          <fpage>606</fpage>
          -
          <lpage>611</lpage>
          .
        </mixed-citation>
      </ref>
      <ref id="ref19">
        <mixed-citation>
          [19]
          <string-name>
            <given-names>Y.</given-names>
            <surname>Winter</surname>
          </string-name>
          ,
          <article-title>Symmetric predicates and the semantics of reciprocal alternations</article-title>
          ,
          <source>Semantics and Pragmatics</source>
          <volume>11</volume>
          (
          <year>2018</year>
          )
          <fpage>1</fpage>
          -
          <lpage>47</lpage>
          (early access).
          <source>doi:10.3765/sp. 11</source>
          .1.
        </mixed-citation>
      </ref>
      <ref id="ref20">
        <mixed-citation>
          [20]
          <string-name>
            <given-names>M.</given-names>
            <surname>Lopatková</surname>
          </string-name>
          ,
          <string-name>
            <given-names>V.</given-names>
            <surname>Kettnerová</surname>
          </string-name>
          ,
          <string-name>
            <given-names>E.</given-names>
            <surname>Bejček</surname>
          </string-name>
          ,
          <string-name>
            <given-names>A.</given-names>
            <surname>Vernerová</surname>
          </string-name>
          ,
          <string-name>
            <given-names>Z.</given-names>
            <surname>Žabokrtský</surname>
          </string-name>
          ,
          <article-title>Valenční slovník českých sloves VALLEX</article-title>
          ,
          <string-name>
            <surname>Nakladatelství</surname>
            <given-names>Karolinum</given-names>
          </string-name>
          , Praha,
          <year>2016</year>
          .
        </mixed-citation>
      </ref>
    </ref-list>
  </back>
</article>