Syntax: Difference between revisions

Latest revision as of 17:09, 13 May 2024

Syntax is the linguistic study of how words combine to form sentences.^[1]

There are many broad theories of how syntax arises, rooted in philosophical questions. What is the structure of human language? How do humans acquire language so quickly, and why do they make some kinds of mistakes but not others? And how do these theories apply to conlangs like Toaq, anyway?

Note: You can learn Toaq without ever caring about the stuff in this article, just like you can learn English without being a linguist. But Toaq's author (and its tinkerers) like to play in the space where conlanging and Chomskyan linguistics meet. Okay, on with the show!

Generative grammar and loglangs

One theory of syntax is called generativism. It posits that the human brain has an innate faculty for grammar, which places certain restrictions on the parameters of human languages. We can imagine a universal grammar "generated" by this innate faculty. There have been decades of efforts to describe this universal grammar, and show that human languages all over the world adhere to it in some sense. For example, human languages tend to have noun phrases and verb phrases that, on some level, work the same way everywhere.

This theory is not without criticism: there's little neurological evidence for an innate "language device". However, generativism has also been very successful in explaining natural language syntax and semantics at many levels. If we can describe Toaq's syntax in these same terms, we can be certain that it is a human language, rather than merely a way to "speak out loud" an unnatural mathematical or logical structure.

Hoemaı has been working to describe Toaq's syntax with the same linguistic tools as are used to describe natural langauges, and Toaq is evolving with this goal in mind. This current description of Toaq syntax is influenced by X-bar theory and the Minimalist program — sub-theories of generativism with certain ideas about syntactic structure.^[2]

Toaq being a loglang means that we can unambiguously parse sentences into syntax trees. Kuna is a piece of software that performs this transformation. There is similar software for Lojban called camxes, but while its output is deterministic, the resulting tree is (from a linguist's perspective) quite ad-hoc and not useful for semantic interpretation.

Notes

↑ To linguists, "syntax" is a subset of "grammar", and grammar also includes things like the study of valid word forms.
↑ The forefront of linguistic knowledge has progressed a bit beyond these theories, but they are still very adequate frameworks to serve as points of reference. (As conlangers, we can "cheat" a little and design Toaq so as to not bump into the flaws of the older, more fleshed-out systems.)

[1] To linguists, "syntax" is a subset of "grammar", and grammar also includes things like the study of valid word forms.

[2] The forefront of linguistic knowledge has progressed a bit beyond these theories, but they are still very adequate frameworks to serve as points of reference. (As conlangers, we can "cheat" a little and design Toaq so as to not bump into the flaws of the older, more fleshed-out systems.)

[1]

[2]

@@ Line 12: / Line 12: @@
 [[Hoemaı]] has been working to describe Toaq's syntax with the same linguistic tools as are used to describe natural langauges, and Toaq is evolving with this goal in mind. This current description of Toaq syntax is influenced by '''X-bar theory''' and the '''Minimalist program''' — sub-theories of generativism with certain ideas about syntactic structure.<ref>The forefront of linguistic knowledge has progressed a bit beyond these theories, but they are still very adequate frameworks to serve as points of reference. (As conlangers, we can "cheat" a little and design Toaq so as to not bump into the flaws of the older, more fleshed-out systems.)</ref>
-Toaq being a [[loglang]] means that we can unambiguously parse sentences into syntax trees. [[Zugaı]] is a piece of software that performs this transformation.<ref>It is unfinished, and currently in the process of being replaced by a tool called “Kuna” for parsing Toaq Delta.</ref> There is similar software for [[Lojban]] called ''camxes'', but while its output is deterministic, the resulting tree is (from a linguist's perspective) quite ad-hoc and not useful for semantic interpretation.
+Toaq being a [[loglang]] means that we can unambiguously parse sentences into syntax trees. [[Kuna]] is a piece of software that performs this transformation. There is similar software for [[Lojban]] called ''camxes'', but while its output is deterministic, the resulting tree is (from a linguist's perspective) quite ad-hoc and not useful for semantic interpretation.
-== Movement ==
-Generativist syntacticians say that sentences have a "deep structure" that adheres to universal grammar, but various language-specific constraints transform this into the "surface structure" when the sentence gets actually realized. One important such transformation is '''syntactic movement'''.
-For example, English has something called ''wh-movement'': when we turn a sentence like "Mary wants Bill to dance" into a wh-question, we say "'''Who does''' Mary want () to dance?".
-The generative explanation for this is that the question has a deep structure like "Mary wants '''who''' to dance?", and then for pragmatic reasons, the question word moves to the front of the sentence and gets supported by "does". There is a '''trace''' marked by () in the spot where "who" moved from.
-There is good evidence for wh-traces. English speakers tend to agree that we can't contract the question to "Who does Mary wanna dance?" — we can imagine the wh-trace between "want to" is there, unpronounced, but blocking the contraction.
-Note that the claim is ''not'' that the deep-structure sentence first forms in the speaker's mind, and is then rearranged into surface-structure. The temporal "before and after" perspective on movement is only a useful metaphor for a language's grammar rules.
-=== Movement in Toaq ===
-The tree for a sentence like {{t|Noaq jí kúe nha}} indicates an SVO deep structure: <code>jí <s>noaq</s> kúe</code>. What's going on?
-The generativist "verb phrase" has the verb and the object generated side-by-side. Even in VSO natural languages like Irish, there is evidence for verb-and-object VP structures. Meanwhile, there is also some evidence for verb-and-object structures in Toaq: for example, [[prepositional phrase]]s like {{t|tî kúa}}, or object-incorporating verbs like {{t|po}}.
-A generativist approach for analyzing a VSO language is thus that the verb and object really are side-by-side in the deep structure, and that the verb moves up to the front of the sentence for ''some'' reason.<ref>Hoemaı has suggested that there is "room for fanfic" as to why this happens in Toaq. Perhaps {{t|Noaq jí kúe nha}} is a bit like "Read it, I'll do that book!" — except perhaps over time it got watered down and became normal grammar with no actual focusing function.</ref>
-Toaq could have been designed as SVO from the start, and have a surface structure that's closer to the deep structure. There are aesthetical arguments in favor of VSO. For example, VSO grammar is similar to the logic notation <math>P(x,y,z)</math> for predicates and their arguments.
 == See also ==
+* [[Movement]]
 *  [https://docs.google.com/document/d/1XEi-Cg29gAazLwEdjWmSzBYMKiXhDE2OuSQZL2YyUuo/edit?usp=sharing ''First Steps Towards a Compositional Semantics for Toaq''] and [https://toaqlanguage.wordpress.com/2022/10/08/not-quite-a-forest/ ''Not quite a forest''], both blog posts on Toaq syntax.
 * The refgram sections on [https://toaq.net/refgram/syntax/ Syntax] and [https://toaq.net/refgram/semantics/ Semantics].