The Subclause Reform is a proposal that changes the grammar of subclauses to fix some ambiguities and make them more naturalistic. It is inspired by the mö proposal and builds on insights from the article CPs: Copies and Compositionality.[1]
Motivation
In official Toaq, content clauses have the exact same grammar as the main clause of a sentence, and can appear wherever a determiner phrase might appear. But this grammar actually turns out to be ambiguous! Consider the following sentence:
Feq jí ꝡä za ruqshua râo ní nuaq.
Which verb does the adjunct râo ní nuaq modify? It's not clear where the subclause ends. This example could mean either "I remember that tonight, it's going to rain" or "Tonight, I remember that it's going to rain".
Another tricky thing is that in official Toaq, you can underfill verbs, for example by using a transitive verb intransitively:
Leo nháo da.
Now how would we say "It delights me that they tried"? If we allow verbs in subclauses to be underfilled, this also creates ambiguity:
Jaıca ꝡä leo nháo jí.
Again, it's not clear where the subclause ends, because jí could belong to either the inner or outer clause, depending on which verb is underfilled. Clearly, we need a better solution.
Content clauses
To avoid ambiguity when a content clause has a trailing adverb, we say that subclauses starting with ꝡä/lä/etc. can only appear at the very end of a clause, coming after the outer clause's trailing adverbs:[2]
Zaı jí, ꝡä jıa tao nháo hóq nhûq súq.
I hope that they will do it for your sake.
Zaı jí nhûq súq, ꝡä jıa tao nháo hóq.
I hope for your sake that they will do it.
Normally, such a clause will fill in the final slot of the verb. But if you want to use a content clause as the subject of a transitive verb, for example, then you can use the word có, which is a lot like the 'it' in English "It delights me that they tried".[3]
Gı có, ꝡä tı súq ní.
It's good that you're here.
Jaıca có jí, ꝡä leo nháo.
It delights me that they tried.
Strict clauses
The other central idea of this proposal is to restrict the grammar of certain subclauses. A strict clause has the following properties:
- The verb may not be underfilled. This means that chuq póq is not a valid strict clause, and instead we must use prefixes like hao- to explicitly reduce the verb's arity: hạochuq póq.
- Adverbs may only appear at the start of the post-field, not at the end. So to turn geq jí páı tî hú seaırıaq into a strict clause, you would need to reword it as either geq tî hú seaırıaq jí páı or tî hú seaırıaq nä geq jí páı.
- All subclauses of a strict clause must also be strict.
The point of these restrictions is that, when taken together, you can put a strict clause nearly anywhere in a sentence and still tell unambiguously where the clause ends.
Relative clauses
Relative clauses will be our first example of clauses that benefit from being strict. In the following examples, strictness makes it possible to know for sure which constituents belong to which clause:
Pıe jí cháı, ꝡë baı tâocıa súq hóa.
I drink the tea that you unintentionally made.
Pıe jí cháı, ꝡë baı súq hóa, tâocıa.
I unintentionally drink the tea that you made.
Chum kaqsı póq, ꝡë hạobuı hóa, jí.
The person who is outside is watching me.
But the grammar isn't only sensitive to where relative clauses end, we also have to care about the nouns that they attach to; their antecedents.
Jéa lô chao, ꝡë cho súq hóa
Jéa po cháo, ꝡë cho súq hóa
Jéa, ꝡë nua hóa cháo, ꝡë cho súq hóa
These examples are still ambiguous, because in each of them the relative clause ꝡë cho súq hóa could have either jea or chao as its antecedent. To solve this, we say that each relative clause must be matched by a sú in the antecedent, which acts as a placeholder to be filled in the with the clause's content.
Jéa sú [lô chao], ꝡë cho súq hóa
The buyer [of the vehicle] that you liked
Jéa lô [chao sú, ꝡë cho súq hóa]
The buyer of the [vehicle that you liked]
By marking the antecedent explicitly, sú makes it clear which noun the relative clause attaches to. Of course, this makes relative clauses require some forethought and also become a bit more verbose, neither of which are desirable. So we actually allow sú to be covert (i.e. unspoken) most of the time, only becoming overt in strict clauses, predicatizers, and object-incorporated phrases.
Jéa [lô chao], ꝡë cho súq hóa
The buyer [of the vehicle] that you liked
Jéa lô [chao sú, ꝡë cho súq hóa]
The buyer of the [vehicle that you liked]
Tú poq, [ꝡë zao hóa ní suaqche], ꝡë ao pua tı hóa gíaqtue po súaqche
Everybody [who knows this singer] that would enjoy going to their concerts
Tú poq, ꝡë zao hóa [ní suaqche sú, ꝡë he fıeq hóa báq de]
Everybody who knows [this singer that writes beautiful songs]
Another way to think of this behavior is that ꝡë prefers to avoid center embedding, and will only attach to nouns in a way that could produce center embedding if you force it to with sú. Tightly grouping ꝡë with the tonal morpheme 
to produce ꝡé will also force this attachment, despite using a covert sú.
Tú poq, ꝡë zao hóa [ꝡé he fıeq hóa báq gıaq de]
Everybody who knows [the one that writes beautiful songs]
Event accessor phrases
In official Toaq, the event accessor ë is said to turn a subclause into a verb. But this grammar creates multiple problems:
- The semantic type of event accessor phrases is different from that of verbs; it lacks an event place.
- It can be ambiguous where the subclause ends, especially if an event accessor verb were followed by adjectives.
Therefore this proposal changes event accessor phrases to have the grammar of relative clauses rather than verbs. That is, ë opens strict clauses, and may only appear after a nominal antecedent marked with sú. Just as with relative clauses, this sú may be covert.
Ké paı, ꝡë luı gaı hóa tú sú ë hạobuaq jí
Those friends that have seen every time I've failed
Jôı tú mabala ë dea nháo láqbıo nä jeaq rueqmoe jí.
With each horrible striking of the bell, I grew more tense.
Kaq jí é marao súq.
I saw you dance.
Relative content clauses
To turn a content clause into a nominalized construct, like English "the fact that" or "the plan to", we introduce a new piece of grammar: a relative content clause. The word ꝡä attaches to a determiner phrase as if it was a relative clause, and says that the phrase's propositional content is given by the content clause that follows.
Cho jí sío sú, ꝡä mala tı sía seq úmo ní rıaq.
I like the thought that no one apart from us has ever been here before.
Táosıo sú, lä seraq nhâna kú râo núaq já, bï chı duı hıam jí hóq.
I think the plan to attack them at night is too dangerous.
Just like normal relative clauses, these clauses are strict and must be matched by a sú in the antecedent. But the conditions under which sú may be covert are a little different: ꝡä and its friends much prefer to act as the complement of a verb, so sú will normally need to be overt to force these words to attach as a relative content clause. Sú may only be covert when ꝡä is tightly bound to the tonal morpheme as ꝡá:
Hạle, ꝡá sho suhu hóe, ꝡá dana súq jí.
It is more likely that the sun turns into a pig than that you beat me.
Má tı ríaq Éoropa, bï bu moaq jí hụ́ma.
As for whether that place is in Europe, I do not remember that.
Notice that nominalizing a content clause with ꝡá allows it to appear directly in topic or subject position, whereas a ꝡä clause is much more limited in its positioning.
Nullary verbs
As a final note, we need to change nullary verbs to take a dummy argument, or else this kind of sentence would still be ambiguous:
Feq jí ꝡá za ruqshua râo ní nuaq.
This sentence now becomes:
Feq jí, ꝡá za ruqshua râo ní nuaq ía.
I sense that it's going to rain tonight.
This dummy argument ía corresponds to the 'it' in the English translation; it doesn't refer to anything, but is necessary for the sentence to be grammatical.
- ↑ See also Arguments Against CPs as Arguments, which presents the article in a more compelling lecture format and provides additional discussion.
- ↑ This word order is created by a remnant movement pattern: the CP starts in verbal argument position, then it moves up above TP, and finally the "remnant" of TP moves up above CP.
- ↑ Syntactically, we understand có as a trace of type that the CP leaves behind when it moves. This trace may be covert if it occurs in the verb's final slot.
