Siloni (2012): Reciprocal verbs and symmetry #

Beyond periphrastic reciprocals (each other), languages have reciprocal verbs — predicates that express reciprocity without an anaphoric object, carried as Verb.Reciprocal entries. [Sil12] (building on [Sil08]) splits them by where reciprocalization applies — the Formation parameter of Syntax/Reciprocal.lean, an instance of [RS05]'s lex-syn parameter — and derives nine clustering properties from that split alone: "the distinctions all follow from the fact that the two types of verbs are formed in different components of the grammar."

Reciprocalization acts on argument structure: reciprocalize takes a transitive RoleList and returns an intransitive one whose subject carries both base profiles' entailments — as [RS05]'s bundled complex [θᵢ · θⱼ] role in the lexicon ((35)–(36)), or as two separately assigned roles via last-resort parasitic assignment in the syntax ((43), the Merger ledger, which also derives the raising/passive ban and the ECM derivation (63)). Event-semantically the two loci share one mechanism, [Lan00b]'s group operators (Plurality.GroupStructure) in both domains: a symmetric verb is a set of atomic events whose complex role assigns the group atom over an unordered pair, related to its base by fn. 17's meaning postulate (Verb.SymmetricDenotation, in Syntax/Category/Verb/Symmetric.lean), from which (41) — the underlying directional events — is derived (underlying_of_symmetric); the syntactic reciprocal has the accumulation reading ((46b), AccumulationReading) and the up-packed singular reading ((47), GroupEventReading), whose content provably coincides with the symmetric verb's (GroupEventReading.underlyingReading). Since the "I" reading of embedded reciprocals ([Hig80]; the LF-decomposed periphrastic arm is Studies/HeimLasnikMay1991.lean) requires a sub-event reading plus a sole-role subject, both verb types lack it (I_reading_iff_periphrastic).

The nine surface properties derive from four affordances of the locus of application (pluralEventsAvailable per generalization (29), storesOutputs, spansPredicates, reducesAccusative): the paper's concluding table is a theorem (predictedProperties_lexical), as is the perfect complementarity of the two clusters. The paper's eleven-language sample is recorded per-language (hebrew … bulgarian).

The §3 case against an object-clitic analysis of se is formalized as three diagnostics — functions from the verb's grid to an observable — on which the rival transitive grid and the reciprocalized grid diverge and the data side with reciprocalization: Czech depictive case agreement (depictive_diagnostic), comparative-ellipsis remnants (ellipsis_diagnostic), and [Kay75]'s faire-causative causee marking (causative_diagnostic).

Three further pieces of the paper's argument are worked out against the substrate: the periphrastic sub-event reading is derived from [HLM91]'s each∘other LF rather than stipulated (accumulation_of_eachOtherLF); the (15)/(16) count-adverbial contrast is modeled — five atomic mutual events with inaccessible parts versus five accumulation events with ten accessible directional parts (card_mutualEvent, mutualEvent_not_subEventReading, accumulated_subEventReading); and the discontinuous construction is the dyadic realization of a symmetric entry (§7.5), commutative in its two arguments and entailing the crossed directional events (dyadicReading_comm, DyadicReading.underlying).

Siloni (2012): Reciprocal verbs and symmetry #

Three-way reciprocal classification (§2.4) #

The lex-syn architecture (§3.5, §§5–6) #

Symmetric verbs and reciprocal readings (§2.2, §4.1–4.2) #

Reciprocalization on argument structure (§4.1–4.2) #

Against an object-clitic analysis of se (§3) #

Parasitic assignment (§4.2, (43); the ECM derivation (63)) #

The nine-property cluster (§§2–7) #

Symmetric verbs (§2.3, §7.5) #

The "I" reading (§2.1, §4.3) #

The periphrastic source of sub-events (§2.1) #

Count adverbials: five mutual versus ten directional events (§2.2) #

The discontinuous construction: dyadic symmetric verbs (§7.4–7.5) #

Cross-linguistic sample (§2.4, §5, §7) #