Causative templates and argument selection

Why the dictionary keeps this

A translation is accurate when the arguments match: which theta role is the subject, which is the object, which is the oblique, and whether two surface phrases are one fused argument. That is the characterization this note is for. It is not a verb entry, and it is not a frozen theory. Language pages keep their own headwords and their own Case cloth. This note does not collapse them under one English gloss, and it adds no Navajo data.

The deck’s title line is “Causative Templates and Argument Selection,” with the subtitle “Examples from Japanese.” The opening question is what constrains the way a verb selects its arguments. The deck’s answer, following Dowty 1991, is that subject, direct object, and indirect object are chosen by Proto-Agent and Proto-Patient properties. The Proto-Patient property the deck uses is incremental theme. Alsina 1992 supplies two causative templates. Matsumoto 1998 is the source the deck cites for applying those templates to Japanese lexical and productive causatives. The deck itself marks that application as a guess.

Incremental theme working

An incremental theme is a theme whose parts correspond to the parts of an event. When a verb alternates which argument is the direct object, the object is the argument that bears that property. The alternation exists because the object is the incremental theme.

The deck’s English pair is spray/load. The deck’s Japanese pair of the same shape is the wall-painting construction, 壁塗り構文. Lexical 浴びせる shows the same object alternation, with the same shift in which participant is understood as fully affected. Those three pattern together. Productive -sase- is a different subscript, set out below, even where the deck’s sentences also alternate the accusative.

What the object is in the examples

These are the deck’s examples. They are illustrations of argument selection. They are not new elicitations, and they are not Index 2 citations. Glosses name the theta role that the working read assigns.

English spray and load take this lexical pattern. They do not take the productive -sase- subscript.

One accusative, one oblique. The accusative is whichever argument is the incremental theme.

The deck cites Dowty 1991, §9.3.2, for the strangeness of the English gap, and treats the gap as particular to English. That suggestion stays open. The working point is the Japanese object: it is the incremental theme in both sentences.

The surface alternation resembles spray/load. The subscript does not. These sentences are productive -sase-, a second event, Type I below. Each sentence still has one accusative and one oblique.

Type I working

The deck draws two templates and says, in its own words, that this is what they look like. The first panel is the one that holds.

As written on the slide, CAUSE takes an agent and a patient, and the caused event contains a role written θ together with an inner patient in parentheses. A line under the formula says that θ is the highest theta role. A superscript on θ is what marks that highest role. The superscript is too small to identify as a particular letter, so this note does not name the letter. The grammatical-function line places the subject under the agent of CAUSE, an object under the outer patient, and a further object function under the parenthesized inner patient. The paraphrase beside the panel is that the agent brings about a change by acting on the person in control of the event.

The identity that holds is the one in that paraphrase, which is Alsina’s Type I. The patient of CAUSE is the same argument as the highest role of the caused event. Parentheses mark the fusion. There is one object. The slide’s extra object function under the inner patient is the same fused argument, not a second object. Productive -sase- is this template.

CAUSE ⟨ agent, patientᵢ ⟩
        [ event ⟨ (θᵢ) … ⟩ ]

θ     highest role of the caused event
( )   fusion of that role with the patient of CAUSE
ᵢ     one argument, and one object
subject   agent of CAUSE
object    the fused argument

The person in control of the caused event is that fused argument. The causer acts on that person. The deck’s own 浴びさせた sentences agree with the count: one accusative, one oblique, no second object.

The slide draws the parentheses on the inner patient, and it writes the extra object function under that patient. The working line puts the parentheses on θ, because the panel’s paraphrase identifies the patient of CAUSE with the person in control, the highest role. A later reading that places the fusion on a different argument can move this line. It is not locked.

Type II revised

The second panel writes CAUSE with an agent and a patient, and a caused event in which the embedded agent is in parentheses and the inner patient is not. The grammatical-function line places the oblique under that inner patient. The paraphrase beside the panel says the causer acts on the base agent by causing an event that affects an individual.

The parentheses are already on the embedded agent. That part of the drawing is the causee. The revision is the grammatical function. The oblique belongs to that parenthesized causee. The embedded patient is fused with the patient of CAUSE, and that patient is the object. That patient is the incremental theme. Placing the oblique under the inner patient swaps the object and the oblique.

CAUSE ⟨ agent, patientᵢ ⟩
        [ event ⟨ (agentⱼ), patientᵢ ⟩ ]

( )      the embedded agent, the causee
subject  agent of CAUSE
object   patientᵢ
         embedded patient, fused with the patient of CAUSE
         incremental theme
oblique  agentⱼ
         the parenthesized causee

Read against lexical 浴びせた: the accusative is the incremental theme, and the で-phrase or the に-phrase is the oblique. English spray/load is this pattern. The object is the wall or the paint, whichever is the incremental theme. With and on are the oblique. Where the caused event has its own agent, that agent is the oblique causee, and the object remains the incremental theme.

The passive on the earlier slide

Before the templates, the deck passivizes the productive causative. The point of those sentences is narrow. Whichever argument is the structural object can become the subject. That is a passive of a causative. It is the deck’s clue that the two causatives do not select arguments in the same way. It does not name either causative. The spellings on that slide, and the spellings on the aspect slide that follows, are not reproduced.

Aspect

Slides after the templates derive aspect from which argument CAUSE targets, and they treat that aspect as the reason one causative allows more than one reading. Telicity is not in the subscript. Those slides are not adopted.

Applicatives working layer, absent from the deck

The deck has no applicatives. The dictionary uses them as a layer under the causative templates. Low Appl_HAVE sits inside one verb domain. On pay_GIVE the arguments of that domain are already these:

• Payer is Spec of Voice.
• Recipient is low Appl_HAVE.
• Amount is the MeasureP complement, ⟨n, U⟩, the incremental theme of e_HAVE.
• Consideration is AGAINST on that same event. It is not a fourth structural Case.

VoiceP
  Payer
  ApplP_LOW
    Recipient    Appl_HAVE
      √PAY
        MeasureP    Amount ⟨n, U⟩
  AGAINST    Consideration

The head of the applicative is Appl_HAVE. The projection is ApplP_LOW. High Appl is an affectee of the event, a separate layer from Recipient. Productive -sase- is neither of those heads. It is derivational morphology, an outer wrap, and a second event: Type I. It is not the lexical causative, and it is not a fourth argument of the inner verb.

The cloth stays on the language page. English pay, Japanese 払う, Spanish pagar, French payer, Italian pagare, Hungarian fizet, Polish płacić, Bulgarian плащам, and Mandarin 付 each keep their own Recipient, Amount, and Consideration. -sase- over Japanese 払う, 売る, 払う of a bill, or 割り当てる is the outer wrap, with the causee distinct from the inner applied argument, and still one accusative in the inner domain. Bulgarian плащам and Polish płacić do not acquire a CAUSE domain from this note.

What stays open

• The English gap with かぶせる. The deck records it and cites Dowty. The reason English lacks the locative is not settled here.
• The letter of the superscript on θ.
• How far the revised Type II causee, as opposed to an instrument or a location, is the right oblique for every lexical causative and for every spray/load oblique. The object in those alternations is the incremental theme. That part is the working claim. The causee label is the revision of the template, and it can move.
• Matsumoto’s mapping of the two Alsina types onto Japanese. The deck offers it as a guess. Type I is what this note uses. Type II is the revised line above.

On the verb pages

The argument selection of a particular verb stays on that verb’s page. This note is the shared subscript those pages point at.

• pay_GIVE and the language pages linked above — low Appl_HAVE.
• sell and 売る — incremental theme of the goods, and Japanese -sase- as an outer wrap.
• pay_BILL · Japanese, allocate_GOAL · Japanese, charge_ACCUSE — -sase- outside the inner object.
• Incremental theme baseline — Polish jeść / zjeść, a measure of consumption. A different use of the same name.