Talk:Formal semantics (natural language)/Archives/2025/June

Diverse ways to construct models

I removed the paragraph that starts like this: "Within formal semantics, there are diverse ways how to construct models and relate linguistic expressions to them" because the subsequent text doesn't really describe current approaches or methodological splits. But I do think a paragraph of this sort would be good to have, perhaps discussing things like direct compositionality versus the logical form-approach. I can write this up but it won't be soon. Botterweg (talk) 23:04, 3 June 2025 (UTC)

My impression was that the paragraph already discussed this problem: the contrast between grammatical and logical form was explained in the first part while direct compositionality (via the rule-to-rule hypothesis) was examined in the second part. I'm open to reformulating it to shift the focus on "direct compositionality" instead of "rule-to-rule hypothesis", which is probably the wider term. The book you linked (Jacobson 2014) discusses the problem under the term "direct compositionality" on pp. 8–9: To elucidate, a fairly uncontroversial claim is that the grammar of any natural language is a system of rules (or principles, if one prefers) that define the set of well-formed expressions of the language (i.e., the syntax) and a set of rules (or principles) pairing these with meanings (i.e., the semantics). The hypothesis of Direct Compositionality is a simple one: the two systems work in tandem. Each expression that is proven well-formed in the syntax is assigned a meaning by the semantics, and the syntactic rules or principles which prove an expression as well-formed are paired with the semantics which assign the expression a meaning. Do you know if Jacobson 2014 discusses the contrast between the two approaches somewhere in more detail? Phlsph7 (talk) 09:47, 4 June 2025 (UTC)

If I'm understanding Stokhof correctly, he's referring to the philosophical concept of logical form, i.e. the unambiguous form of a meaning that exists independently of the linguistic vehicle used to express it. That's distinct from the concept I was referring to, though they (annoyingly) share the same name.

The fundamental idea behind the LF approach is that sentences do not have a unique grammatical form, but rather a surface grammatical form which gets pronounced and a separate logical grammatical form (sic) that undergoes semantic interpretation. In these approaches, apparent mismatches between form and meaning are explained using syntactic mechanisms such as movement. This is in contrast with Direct Compositionality, whose fundamental idea is that the surface grammatical form is the only grammatical form and that mismatches should be explained using semantic mechanisms such as type shifting.

There has also been debate about whether semantic rules should be stated in terms of syntactic categories or semantic types. This is related to the LF/DC debate, but the relationship isn't straightforward. For instance, the idea of type-driven interpretation was proposed within a direct compositional system, and the Jacobsen textbook explores type-driven implementations in a few places. There might be something to say here within a broader discussion of categorial versus generative models of the syntax-semantics interface, but this isn't an active debate on its own the way LF versus DC is. Botterweg (talk) 02:28, 8 June 2025 (UTC)

I don't think that the earlier passage in the paragraph implied that logical form "exists independently of the linguistic vehicle used to express it". As I understand it, both Stokhof 2007 and Stokhof 2013 talk about the methodology of formal semantics. We could add additional sources to back up the passage. Would the following source be acceptable to you? It says: The logical form of a sentence (or utterance) is a formal representation of its logical structure—that is, of the structure that is relevant to specifying its logical role and properties.

I'm open to your idea of mentioning a methodological split following an "LF/DC debate", but this needs to be based on high-quality sources that explicitly make this point. Do you know of any? Trying to build an argument for the existence of this split based on other discussions, like the "debate about whether semantic rules should be stated in terms of syntactic categories or semantic types", would probably be WP:OR/WP:SYNTH. Phlsph7 (talk) 09:01, 8 June 2025 (UTC)

The issue of LF/DC is probably the main debate regarding the syntax-semantic interface, so sources abound. Not all of them use the term "direct compositionality" so you could frame the issue as "whether LF exists" or something along those lines if you think there's risk of SYNTH. For the LF view, Chapter 7 of Heim & Kratzer is the locus classicus particularly for empirical arguments based on ACD and scope islands. For the opposing view (whatever you would want to call it) Chris Barker and Polly Jacobson have various handbook articles and textbooks and course notes and so forth. But once again, this is not quite the same issue as rule-by-rule versus semantic type-driven interpretation. Botterweg (talk) 23:30, 9 June 2025 (UTC)

Looks like Coppock & Champollion have a nice discussion that does use the term "Direct Compositionality". Botterweg (talk) 23:37, 9 June 2025 (UTC)

I found the books by Pauline Jacobson quite helpful. I restored a modified version of the paragraph, working in the logical form definition cited above and adding an explanation of direct compositionality and its contrast to approaches focusing on logical form. Phlsph7 (talk) 12:30, 10 June 2025 (UTC)

Not yet a GA

Hey Phlsph7, I really appreciate so so much that work that you've done on this article. So much of it is really excellent and wonderful. But even so I don't think this is ready for GA status yet. My main concern is that there are still substantial factual errors, including the conflation between the syntactic level known as logical form and the historically related but ultimately distinct philosophical concept. But there's also significant gaps (e.g. no mention of crosslinguistic work) and expository inconsistencies (e.g. switching between direct and indirect interpretation without explanation). I understand it's really friggin hard to stitch together different sources that have different terminologies, approaches, and so forth, and you've done a better job that I've done in some of my editing on other topics. But these are important issues that would need to be fixed before this can be called a good article. I can help, though not very much in the immediate future. Botterweg (talk) 00:55, 20 June 2025 (UTC)

Thanks for the feedback! I added a short passages on cross-linguistic work. I don't think that this topic should get much more attention since many of the main sources, like Kearns 2011, Portner 2005, and Winter 2016, give little attention to it and do not discuss it as one of the main topics. I'm also open to more improvement ideas, but it's difficult to assess criticisms without information on the affected passages. If you could quote the passages that you think are problematic, I would be happy to compare them to what the cited sources say. Phlsph7 (talk) 12:20, 20 June 2025 (UTC)

By the way, would you like to join the GA nomination as a co-nominator? Phlsph7 (talk) 10:26, 23 June 2025 (UTC)