Talk:Formalism (linguistics)

Poylysemy

User:OpenNotes1, thanks for your edits - you've been working hard! So, this is a new article and you start it by citing McElvenny. The source actually states: "'Form' and 'formalism' are a pair of highly productive and polysemous terms". Indeed they are. But how should one take the ambiguity into account? I think the four main meanings of 'formalism' in linguistics are as follows:

1. 'a formalism' = a syntactic model. This is a dictionary entry.
2. formalism: borrowed from Russian literary theory. Jakobson is an adherent.
3. Formalism: borrowed from biology. Chomsky, Newmeyer, Fitch and Gould are adherents. Similar to structuralism (biology).
4. Formal linguistics as a branch of formal sciences (i.e. mathematics, computation)

Furthermore, Hjelmslev earned the nickname "formalist" from some (classical) functionalists because he used the word 'function' in the meaning 'dependency' (as is explained by Daneš). However, Hjelmslev is not a 'formalist' in the sense that semantics and pragmatics are parts of his theory of language. Thus, his idea is quite the opposite from Chomsky's. These should not be confused despite ambiguity in the terminology. User:OpenNotes1, you're trying to make perfect sense of the terminology. Unfortunately it will be impossible because there is no perfect sense. To avoid total confusion, encyclopedians will have to disambiguate and explain the differences. For example, it is not possible to find a useful single definition for the two states called Georgia. Weidorje (talk) 10:31, 26 January 2021 (UTC)

OpenNotes1, are you aware that there's already a WP article on formal linguistics which discusses the formalizing and computational aspects of linguistics?

Formalism does not require autonomy of syntax

In the section "Ideas", it is asserted that there is no formal linguistics without an assumption of autonomy of syntax. But the vast majority of formal approaches (in phonology, in semantics, and even in syntax), don't rely on--or bear on--such an assumption. Mundart (talk) 09:51, 12 May 2021 (UTC)

Hjelmslev's model is a semiotic one, thus it is anti-autonomy in this sense. People confuse this issue with the one concerning the status of linguistics as an autonomous science (which does not relate to formalism). The article says that formalism was inspired by Saussure, but this is only because Hjelmslev combined Carnap's mathematical grammar with Saussurean semiotics (aka semiology). Saussure himself has nothing to do with it. As regards the foundations in mathematics, Hilbert and Carnap had worked on a "formalism" which is a strictly axiomatic approach to syntax (of arithmetic), hence the name. Carnap, however, later started working on semiotics, too.

It is also true that generative syntax is not really autonomous from semantics. Contrary to what Chomsky's predecessors like Bloomfield claimed, their analysis of syntax is thoroughly based on meaning: (i) lexical classes are based on meaning, and these form one basis for the analysis; (ii) dependencies are also based on meaning-relations. There was never a true distributionalism because how could you get one without considering meaning? You could only do something like word frequency lists. Bloomfield's enterprise was fundamentally nonsensical. As regards Chomsky, autonomy of syntax (from meaning) is understood on the basis of his claim that syntactic structures are a genetic mutation, and that people associate (arbitrary) meaning with the innate structures that stem from their brains.

Chomsky's approach is essentially not an axiomatic one. From one angle, you could describe it as anti-formalism because his core point is that natural languages are not axiomatic and not logical, but natural. He uses a mathematical model for analysis, but mathematics itself has no explanatory value in generative grammar, only descriptive. This is in stark opposition to the core point made by the formalists who precisely argued against the kind of psychologism/species-relativism that Chomsky represents.

The problem of this article is that the word 'formalism' has been used in many different senses, but it is trying to make them all work together. So, instead of looking at what ideas are closely linked, it asks which disconnected, even antagonistic ideas work together to create a confusion under a common banner. It is like merging Georgia with Georgia. Weidorje (talk) 08:22, 10 May 2021 (UTC)

Agreed. I also agree it's important to distinguish formal approaches from non-formal ones. It's just that these distinctions in approaches (and practices) are completely orthogonal to the autonomy of syntax. I'd love to disentangle these here, and set aside discussion of autonomy to its own page. Mundart (talk) 09:51, 12 May 2021 (UTC)

Okay, so you're talking about the page Autonomy of syntax. A good question: what kind of linguistics is not formal(ist)? That depends on the definition. Some people are going to say that "functionalist" approaches are not formalist. And when you ask them what functionalism is, they'll tell you it is that which is not formalism. But what kind of definition is that? I can find a paper by Croft and another one by Hengeveld stating the obvious that formalism is not the "opposite" of functionalism if formalism means the use of mathematical grammars for linguistic analysis. For Croft, it is the difference between biological innatism and biological non-innatism. That has little to do with the original ideas of formalism and functionalism. Weidorje (talk) 20:50, 12 May 2021 (UTC)