Kalin (2018) — Licensing and Differential Object Marking #
[Kal18] derives differential object marking (DOM) from nominal licensing rather than object visibility, raising, or differentiation. Two parameters interact: (i) which nominals require licensing (in Senaya, only specific ones), and (ii) where the licensers are — every clause has one obligatory primary licenser (always merged, licensing the closest nominal) plus secondary licensers that merge only as a last resort, when their absence would leave some needy nominal unlicensed. DOM is the visible signature of a secondary licenser activating.
The motivating data are from the Neo-Aramaic language Senaya, where DOM
surfaces as differential verbal agreement (an L-suffix), not case — and
[Kal18] argues case and agreement are two reflexes of one licensing
process, so we model the agreement marking abstractly through the licensing
substrate's outcome (Syntax/Case/Licensing.lean). The Senaya facts (paper
examples around the object-agreement and aspect-split data): a specific
object triggers agreement, a nonspecific one does not, and — the crux — in
the perfective base the object position is unlicensed, so a specific
object (which needs licensing) is banned there entirely.
That perfective ban is [Kal18]'s argument against a no-licensing view
of case ([Mar91], [Pre14]): if nominals never needed abstract
licensing, the ban would be unexplained. The flagship theorem below states
this divergence formally, via the shared Assigner harness
(Syntax/Case/Assigner.lean): on the perfective object, a Marantz-style
total configurational account assigns a case, while Kalin licensing crashes —
the two accounts disagree precisely where licensing is unavailable.
Senaya clause configurations #
The aspect split as licenser availability: the imperfective base offers a
secondary licenser for a specific object (yielding DOM agreement); the
perfective base offers none (so a specific object cannot be licensed). The
assignedCase fields abstract the agreement marking — nom for the primary
(subject) relation, acc for the object relation.
Imperfective base: a secondary licenser is available for a specific object.
Equations
- One or more equations did not get rendered due to their size.
Instances For
Perfective base: no secondary licenser — the object position is unlicensed ([Kal18]'s central Senaya claim).
Equations
- One or more equations did not get rendered due to their size.
Instances For
A transitive clause with a specific object: both nominals carry the licensing requirement.
Equations
- Kalin2018.specificObjectClause = [{ label := "subj", lexicalCase := none, needsLicensing := true }, { label := "obj", lexicalCase := none, needsLicensing := true }]
Instances For
A transitive clause with a nonspecific object: the object lacks the licensing requirement, so it is interpretable in situ.
Equations
- Kalin2018.nonspecificObjectClause = [{ label := "subj", lexicalCase := none, needsLicensing := true }, { label := "obj", lexicalCase := none, needsLicensing := false }]
Instances For
The Senaya DOM pattern #
Imperfective: a specific object is licensed by the secondary licenser, surfacing as DOM agreement.
Perfective: a specific object cannot be licensed (no secondary licenser available) and crashes — Senaya's ban on specific objects in the perfective base.
A nonspecific object does not need licensing, so it is fine in the perfective (licensed trivially by the primary, no DOM marking).
Subjects are not differential: the closest nominal to the obligatory primary licenser is always licensed, in either aspect.
The flagship divergence: licensing vs. no-licensing #
On the perfective object, the two accounts assign incompatible
verdicts: a Marantz-style total configurational account gives it a
structural accusative, while Kalin licensing crashes it (uncased,
caseless). This is the witness behind the divergence.
Licensing diverges from total configurational case assignment ([Kal18] vs [Mar91]). The two accounts disagree on the surface case of the perfective object: the dependent-case account assigns it accusative (case assignment is total — it never crashes), whereas Kalin licensing leaves it unlicensed. [Kal18]'s point: under a no-licensing view the perfective ban on specific objects is unexplained; under licensing it follows, because the perfective object position offers no licenser.