Colloquial Sinhala verb fragment

@cite{beavers-zubair-2013} @cite{gair-paolillo-1997} @cite{inman-1993}

A minimal fragment of the Sinhala (Sinhalese) verb system as formalized in @cite{beavers-zubair-2013}. The data here is the verb inventory needed to drive the empirical predictions of B&Z 2013 about anticausativization, the volitive/involitive stem contrast, and the typology of causer sorts.

Volitive/involitive stem alternation

Every Sinhala verb root has a volitive stem and (most have) an involitive stem. The stems are distinguished by a thematic-vowel alternation: front vowel + -e- in the present for involitive, -a- or -i- otherwise. The volitive defaults to a volitional / intentional reading; the involitive defaults to non-volitional / accidental. Crucially, this is not truth-conditional — what the volitive grammaticalizes is sortal (causer ∈ U_E, the event sort); what the involitive does is fail to grammaticalize anything (no sortal restriction).

Causer-sort typology

Verbs are classified by the sort their causer must satisfy (@cite{beavers-zubair-2013} ex. (81), p. 40):

• kadann / kædenn 'break': causer ∈ U (any) — anticausativizes
• minimarann 'murder': causer ∈ U_E (event) — does not anticausativize
• kapann 'cut': causer ∈ U_E (event) — does not anticausativize

Roots without involitive stems (minimarann, kapann) precisely correspond to roots whose causer sort excludes individuals — the predictive engine of B&Z's analysis.

structure Fragments.Sinhala.Verbs.SinhalaVerb : Type

A Colloquial Sinhala verb root.

involitiveForm is none for verbs whose causer sort excludes individuals (e.g., minimarann 'murder' has no involitive form because its U_E causer can never resolve to an individual, which is what the involitive operator would require).

• gloss : String
• volitiveForm : String
• involitiveForm : Option String
• causerSort : Semantics.Causation.CauserSort

def Fragments.Sinhala.Verbs.instReprSinhalaVerb.repr : SinhalaVerb → ℕ → Std.Format

Equations

• One or more equations did not get rendered due to their size.

@[implicit_reducible]

instance Fragments.Sinhala.Verbs.instReprSinhalaVerb : Repr SinhalaVerb

Equations

• Fragments.Sinhala.Verbs.instReprSinhalaVerb = { reprPrec := Fragments.Sinhala.Verbs.instReprSinhalaVerb.repr }

@[implicit_reducible]

instance Fragments.Sinhala.Verbs.instBEqSinhalaVerb : BEq SinhalaVerb

Equations

• Fragments.Sinhala.Verbs.instBEqSinhalaVerb = { beq := Fragments.Sinhala.Verbs.instBEqSinhalaVerb.beq }

def Fragments.Sinhala.Verbs.instBEqSinhalaVerb.beq : SinhalaVerb → SinhalaVerb → Bool

Equations

• One or more equations did not get rendered due to their size.
• Fragments.Sinhala.Verbs.instBEqSinhalaVerb.beq x✝¹ x✝ = false

def Fragments.Sinhala.Verbs.kadann : SinhalaVerb

kadann (vol) / kædenn (invol) 'break'. @cite{beavers-zubair-2013} ex. (76): [[kada-]] = λyλv∈U λe[...].

Equations

• Fragments.Sinhala.Verbs.kadann = { gloss := "break", volitiveForm := "kadann", involitiveForm := some "kædenn", causerSort := Semantics.Causation.CauserSort.any }

def Fragments.Sinhala.Verbs.gilann : SinhalaVerb

gilann (vol) / gilenn (invol) 'drown'. The canonical example: vol+nominative = intentional drowning; invol+postposition atiŋ = accidental drowning; intransitive gilenn = anticausative 'drown' (paper's exx. (2)-(3)).

Equations

• Fragments.Sinhala.Verbs.gilann = { gloss := "drown", volitiveForm := "gilann", involitiveForm := some "gilenn", causerSort := Semantics.Causation.CauserSort.any }

def Fragments.Sinhala.Verbs.marann : SinhalaVerb

marann (vol) / mærenn (invol) 'kill/die'.

TODO: B&Z 2013 do not directly classify marann on the U/U_E axis; it is included here by analogy with the other alternating roots. Cross-linguistically kill often patterns with U_V (effector OK, individual subject pragmatically marked) — see @cite{levin-hovav-1995} ch. 3 on kill vs. murder.

Equations

• Fragments.Sinhala.Verbs.marann = { gloss := "kill", volitiveForm := "marann", involitiveForm := some "mærenn", causerSort := Semantics.Causation.CauserSort.any }

def Fragments.Sinhala.Verbs.minimarann : SinhalaVerb

minimarann 'murder' — no involitive form. @cite{beavers-zubair-2013} ex. (65b): [[minimara-]] = λyλv∈U_E λe[...]. The U_E sortal restriction is incompatible with U_I, blocking both involitive inflection and anticausativization.

Equations

• Fragments.Sinhala.Verbs.minimarann = { gloss := "murder", volitiveForm := "minimarann", involitiveForm := none, causerSort := Semantics.Causation.CauserSort.event }

def Fragments.Sinhala.Verbs.kapann : SinhalaVerb

kapann 'cut' — no involitive form. Patterns with minimarann (event-sort causer) because cutting requires a sharp instrument and intentional manipulation.

Equations

• Fragments.Sinhala.Verbs.kapann = { gloss := "cut", volitiveForm := "kapann", involitiveForm := none, causerSort := Semantics.Causation.CauserSort.event }

def Fragments.Sinhala.Verbs.vinaashKarann : SinhalaVerb

vinaash-karann (vol) / vinaash-kerenn (invol) 'destroy'.

Cross-linguistic note (@cite{beavers-zubair-2013} §7.4, p. 40): the destroy-class is the wedge B&Z use to motivate the U_V constructor in their lattice — but the wedge is typological, not Sinhala-internal. B&Z 2013 explicitly note that "[the] equivalent [of destroy] does alternate in Sinhala (as do equivalents in Spanish, French, Hebrew, and Greek)", citing Härtl 2003 for German destroy-class verbs that do not alternate. B&Z's ex. (80) [[destroy]] = λyλv∈U_V λe[…] is the analysis of English / German destroy (which does not anticausativize), not of the Sinhala cognate. Sinhala vinaash-karann gets the unrestricted .any sort, consistent with its observed alternation behavior.

The non-trivial empirical content of the lattice is therefore cross-linguistic: destroy_no_anticausative is verified for English/German destroy but not for Sinhala. See the planned English/German fragment for the contrastive instantiation.

Equations

• Fragments.Sinhala.Verbs.vinaashKarann = { gloss := "destroy", volitiveForm := "vinaash-karann", involitiveForm := some "vinaash-kerenn", causerSort := Semantics.Causation.CauserSort.any }

def Fragments.Sinhala.Verbs.allVerbs : List SinhalaVerb

The canonical inventory used in B&Z 2013's empirical arguments.

Equations

• One or more equations did not get rendered due to their size.

def Fragments.Sinhala.Verbs.hasInvolitive (v : SinhalaVerb) : Bool

Whether a verb has an involitive stem form.

Equations

• Fragments.Sinhala.Verbs.hasInvolitive v = v.involitiveForm.isSome

theorem Fragments.Sinhala.Verbs.fragment_involitive_correlates_with_causer_sort : (allVerbs.all fun (v : SinhalaVerb) => hasInvolitive v == decide (v.causerSort = Semantics.Causation.CauserSort.any ∨ v.causerSort = Semantics.Causation.CauserSort.individual)) = true

Fragment-internal correlation, not a B&Z prediction.

Within this fragment, the verbs lacking an involitive form happen to be exactly those whose causer sort excludes individuals (minimarann, kapann; both U_E). This is a stipulated correlation in the data, not a derivation: B&Z 2013 (p. 38) are explicit that the involitive is the elsewhere form — it does not encode anything semantically and is "forced" only because the volitive (which requires U_E) cannot apply. Counterexamples to a naïve sort↔morphology biconditional include experiencer-subject verbs dænenn/ridenn 'feel'/'ache' (p. 36 around (74)) which take individual subjects but lack volitive forms — i.e., they are involitive-only, not absent-of-involitive.

This theorem is therefore a test of the fragment data, not a test of B&Z's theory.