[Lingtyp] Analyzability and compositionality

William Croft wcroft at unm.edu
Tue Dec 31 15:35:42 UTC 2019

I think it is too strong to say that compositionality is “exclusively” semantic. The definition of compositionality, as Juergen states, is that “an expression is semantically compositional if its meaning is fully predictable based on the meanings of its constituents and the construction that combines them”. The constituents are formal units, so compositionality is a semantic property of (combinations of) forms, or more precisely, symbolic units. Most semanticists also refer only to “the rules of combination”, assumig high general semantic interpretation of the combinations of forms (e.g. in set theoretic terms). Juergen takes a constructional view (as do I), and explicitly mentions the formal struxture that is made up of the constituent forms. In construction grammar, both the constituents and the combining construction are forms paired with meanings.

To me, the difference between compositionality and analyzability is that in the former, the meaning of the whole is claimed to be fully predictable from the meanings of the constituents plus the rules of combination (or the meaning of the combining construction); in the latter, the meaning of the whole includes the meanings of the parts but is not fully predictable from a  semantic rule of combination associated with the construction. Hence an expression may be analyzable but not compositional. If we assume tabletop is an instance of a whole-part complex nominal construction, with a whole-part relation being the semantics of the construction, then tabletop is compositional and analyzable; laptop is arguably not compositional but analyzable; and something like red herring is neither compositional nor analyzable.

This distinction is found in cognitive linguistics, and I believe captures Langacker’s notion of analyzability. The idea is that many linguistic expressions are analyzable even if they are not compositional in the Fregean sense. An expression that is analyzable but not compositional is an encoding idiom in the terms of Fillmore et al. 1988; an expression that is neither is a decoding idiom.

However, as Juergen hints, compositionality/analyzability is more like a continuum. It all depends on how general the rule of combination of the construction is, which is a function of how general the construction is; and how vague a rule of combination you are willing to allow for without concluding that “compositionality” is vacuous. In a more usage-based constructional approach, constructions are more specific and their semantics is also more specific, which makes them compositional, but not in the way that Frege intended I suspect. The logical conclusion is the analysis of phrasal idioms in Nunberg et al. 1994, where even spill the beans is compositional for an idiom-specific rule of combination (spill can mean ‘reveal’ when combined with beans, and beans can mean ‘information’ when combined with spill). I argued in Crofgt (2001:179-85) that Nunberg et al. show that one can dissociate compositionality and conventionality (roughly, the specificity of the rule of combination),

Bill Croft

Croft, William. 2001. Radical Construction Grammar: syntactic theory in typological perspective. Oxford: Oxford University Press.

Fillmore, Charles J., Paul Kay and Mary Catherine O’Connor. 1988. Regularity and idiomaticity in grammatical constructions: the case of let alone. Language 64.501-538.

Nunberg, Geoffrey, Ivan A. Sag & Thomas Wasow. 1994. Idioms. Language 70:491-538.

On Dec 31, 2019, at 12:07 AM, Nigel Vincent <nigel.vincent at manchester.ac.uk<mailto:nigel.vincent at manchester.ac.uk>> wrote:

 UNM-IT Warning: This message was sent from outside of the LoboMail system. Do not click on links or open attachments unless you are sure the content is safe. (2.3)

I have a discussion of the relation between these two terms in my chapter 'Compositionality and change' in Claire Bowern and Bethwyn Evans (eds) 'The Routledge Handbook of Historical Linguistics', 2015, pp.103-123. It is important to distinguish the two since, as Juergen says, compositionality is exclusively a semantic term while analyzability as defined by Langacker is commonly used to refer to linguistic forms.

Professor Nigel Vincent, FBA MAE
Professor Emeritus of General & Romance Linguistics
The University of Manchester

Linguistics & English Language
School of Arts, Languages and Cultures
The University of Manchester


From: Lingtyp [lingtyp-bounces at listserv.linguistlist.org<mailto:lingtyp-bounces at listserv.linguistlist.org>] on behalf of Bohnemeyer, Juergen [jb77 at buffalo.edu<mailto:jb77 at buffalo.edu>]
Sent: Monday, December 30, 2019 11:33 PM
To: joo at shh.mpg.de<mailto:joo at shh.mpg.de>
Subject: Re: [Lingtyp] Analyzability and compositionality

Dear Ian — I’m familiar with ‘compositionality’ solely as a semantic term, in the sense of the Principle of Compositionality, also known as the Fregean Principle. In this sense, an expression is semantically compositional if its meaning is fully predictable based on the meanings of its constituents and the construction that combines them. In this sense, _tabletop_ is *not* fully compositional if one assumes that _tabletop_ and _laptop_ have the same structure, since _tabletop_ has a possessive semantics and _laptop_ does not (or rather, it does etymologically, but in addition applies metonymy, since it designates an object characterized by its being designed to be used on top of a person’s lap - but again, that’s really just etymology). There are other ways of analyzing this: one could argue that the two items are licensed by distinct nominal compound constructions (implausible, since the semantic derivation of _laptop_ seems very ad hoc), or that they are licensed by a single construction whose meaning is heavily underspecified - but then one might want to argue that an expression being integrated by a semantically vague construction is just a special case of reduced compositionality.

To return to your question, ‘analyzability’ is not used as a term in the semantic literature, and so to that extent, the two terms are not interchangeable. But there may be other uses of ‘compositionality’ that I’m not aware of and that are interchangeable with ‘analyzability’.

HTH! — Juergen

On Dec 30, 2019, at 5:59 PM, joo at shh.mpg.de<mailto:joo at shh.mpg.de> wrote:

Dear all,

I would like to know if the terms analyzability and compositionality are used interchangeably, and if so to what extent.
In other words, I would like to know if there is any consensus on how to use these terms differently, or are they just synonyms.
Example: The word “tabletop” is fully compositional/analyzable, and the word “laptop” less so (Langacker 2008).
My impression is that analyzability is used mostly for lexemes whereas compositionality is used mostly for phrases, but there seems to be no clear boundary.
I would appreciate any opinion on this.

From Jena,
Lingtyp mailing list
Lingtyp at listserv.linguistlist.org<mailto:Lingtyp at listserv.linguistlist.org>

Juergen Bohnemeyer (He/Him)
Professor and Director of Graduate Studies
Department of Linguistics and Center for Cognitive Science
University at Buffalo

Office: 642 Baldy Hall, UB North Campus * Mailing address: 609 Baldy Hall, Buffalo, NY 14260
Phone: (716) 645 0127
Fax: (716) 645 3825 * Email: jb77 at buffalo.edu<mailto:jb77 at buffalo.edu> * Web: http://www.acsu.buffalo.edu/~jb77/

Office hours Tu/Th 3:30-4:30pm

There’s A Crack In Everything - That’s How The Light Gets In (Leonard Cohen)

Lingtyp mailing list
Lingtyp at listserv.linguistlist.org<mailto:Lingtyp at listserv.linguistlist.org>
Lingtyp mailing list
Lingtyp at listserv.linguistlist.org<mailto:Lingtyp at listserv.linguistlist.org>

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://listserv.linguistlist.org/pipermail/lingtyp/attachments/20191231/dd4cf45a/attachment.htm>

More information about the Lingtyp mailing list