The exponents of the heads in (30).
Derives DecidableEq, Repr.
Used by exponentTable, vocabulary, topDown, exponents
Docs / Studies / Studies.ChengCheng2026
Using light verbs as verbalizers
Cheng, H. & Cheng, L. L.-S. (2026). Natural Language & Linguistic Theory, 44(4)
Cheng and Cheng analyze the Mandarin light verb dǎ 'hit' as a rootless exponent of the verbalizer, a head below the causative gěi and Voice bǎ. In dǎ+R (dǎ-sǎo 'clean') a root adjoins to the verbalizer; in dǎ+N (dǎ diànhuà 'make a phone call') the verbalizer takes a noun phrase.
Implementation notes
- Heads are parts of a partition of the
Harley2017.Features. The paper's causative v is Harley's lower Caus, which Harley takes to be productive; the paper's (28) calls it lexical. - One linking hypothesis connects the tests to structure: a form separating dǎ from what follows it is acceptable iff the two do not share a complex head.
Combination.ShareHeadcompares complex heads as trees, which suffices because the heads compared differ in shape.
TODO
- §5.4 attributes the idiosyncrasy of dǎ+N to a First Phase defined by dǎ, but the noun's
root is not local to dǎ. Whether it belongs to idioms below Voice (
ComplexHead.IdiomLocal) or to the noun, which (68) bears on, needs sourced data on diànhuà. - Whether dǎ-yíng (22) admits the potential infix, the mark of a resultative in footnote 20, needs a sourced judgment.
- The independence claims of §3 beyond
subsetPrinciple_v, and the light verbs of §6.3.
References
- Cheng & Cheng 2026, Using light verbs as verbalizers
- Harley 2017, The ‘Bundling’ Hypothesis and the Disparate Functions of Little v formalized
- Sybesma 2021, Voice and Little v and VO–OV Word-Order Variation in Chinese Languages
- Alexiadou & Lohndal 2017, The structural configurations of root categorization formalized
- Bobaljik & Harley 2017, Suppletion is local: Evidence from Hiaki formalized
Three heads
Exponent names as they appear in the data.
- exponentTable = [("dǎ", Exponent.da), ("gěi", Exponent.gei), ("bǎ", Exponent.ba)]
Built from
Used by exponents
The context-free items dǎ for v, gěi for the lower Caus and bǎ for Voice; the last two follow Sybesma 2021.
The definition is too large to print here; read it in the source.
Built from
Used by subsetPrinciple_v
(l r : List (Finset Harley2017.Feature)) :
DistributedMorphology.subsetPrinciple vocabulary { focus := {Harley2017.Feature.v}, leftCtx := l, rightCtx := r } = some Exponent.da
Built from
The exponents of the heads bearing Voice, the lower Caus and v, in that order.
The definition is too large to print here; read it in the source.
Built from
Used by topDown_eq_iff
The exponents a datum lists, in surface order.
- exponents d = List.filterMap (fun (x : String) => List.lookup x exponentTable) (d.features "exponent")
Built from ,
Used by exponents_ex_30, topDown_eq_iff
(d : Datum) :
d ∈ [Examples.ex_30a, Examples.ex_30b] → exponents d = [Exponent.ba, Exponent.gei, Exponent.da]
Built from
A partition of v, the lower Caus and Voice realizes the exponents of (30) iff it is discrete.
{P : Finpartition Harley2017.vCausVoice} {d : Datum} (hd : d ∈ [Examples.ex_30a, Examples.ex_30b]) :
Built from ,
Separability
The two kinds of combination with dǎ.
- root : Combination
dǎ+R, as in dǎ-sǎo 'clean'.
- noun : Combination
dǎ+N, as in dǎ diànhuà 'make a phone call'.
Derives DecidableEq, Repr.
Used by combinationTable, Combination.tree, Combination.daPos, Combination.rootPos, Combination.ShareHead, Combination.instDecidableShareHead, test?, mayCondition'_rootPos_daPos_iff
Combination names as they appear in the data.
- combinationTable = [("R", Combination.root), ("N", Combination.noun)]
Built from
Used by test?
The lexicon of the dǎ+R structure, with trivial words.
- lexicon = { x := DistributedMorphology.Categorizer.v, root := (), head := (), arg := () }
Used by Combination.tree, x0At_tree_lexicon
The dǎ+N structure (53): the verbalizer over an nP.
The definition is too large to print here; read it in the source.
Used by Combination.tree
The structure of a combination; for dǎ+R, the adjunction structure (50).
- Combination.root.tree = DistributedMorphology.RootCategorization.Attachment.tree lexicon DistributedMorphology.RootCategorization.Attachment.modifier
- Combination.noun.tree = nounTree
Built from , ,
Used by Combination.ShareHead, mayCondition'_rootPos_daPos_iff
The position of dǎ.
- Combination.root.daPos = { toList := [0, 1] }
- Combination.noun.daPos = { toList := [0] }
Built from
Used by Combination.ShareHead, mayCondition'_rootPos_daPos_iff
The position of the root after dǎ.
- Combination.root.rootPos = DistributedMorphology.RootCategorization.rootPos
- Combination.noun.rootPos = { toList := [1, 0, 0] }
Built from
Used by Combination.ShareHead, mayCondition'_rootPos_daPos_iff
dǎ and the root after it lie in one complex head.
- c.ShareHead = (DistributedMorphology.x0At c.tree c.daPos = DistributedMorphology.x0At c.tree c.rootPos)
Built from , , ,
Built from ,
Under either attachment, dǎ and the root lie in one complex head.
Built from
The combination a datum tests, and whether its form separates dǎ from what follows it.
- test? d = do let c ← d.parse? "combination" combinationTable let s ← d.parse? "separated" [("yes", true), ("no", false)] pure (c, s)
Built from ,
Used by length_filterMap_test?, diagnostics_iff
Built from
A test form is acceptable iff it keeps dǎ with what follows it or the two lie in different complex heads ((44), (46), (48), (49)).
(d : Datum) :
d ∈ Examples.all → ∀ p ∈ test? d, d.judgment = Judgment.acceptable ↔ p.2 = true → ¬p.1.ShareHead
Built from ,
Locality of the root
dǎ is local to the root after it iff the combination is dǎ+R.
(c : Combination) :
Built from , , ,