Natural language without semiosis

Omer Preminger

University of Maryland

Friday, October 30, 2020

Zoom (email for link)

This talk aims to show that the atoms of linguistic composition are not Saussurean signs (viz. arbitrary pairings of form and meaning; Saussure 1916, Hjelmslev 1943).

Setting aside ideophones and cases of onomatopoeia, most modern approaches to linguistic theory take it as a given that the atoms of morphosyntactic composition – be they ‘words’ or morphemes – are form- meaning pairings (which can be associated with additional, sui generis syntactic features). I will argue that this is incorrect: architecturally speaking, natural-language expressions are entirely devoid of Saussurean signs (with the possible exception of monomorphemic utterances like “wow!”, “ugh”, and the like).

I will argue in favor of a grammatical architecture where atoms of linguistic composition are entirely abstract, and are not directly associated with form or with meaning. Instead, these atoms, once syntactically arranged, constitute the input to a set of mapping rules to form, and to a separate set of mapping rules to meaning. These mapping rules are many-to-one rules and, importantly, nothing forces the set of atoms that map onto a particular element of form to also map, as a set, onto a particular element (or elements) of meaning. In fact, the input sets to form and to meaning can stand in all manner of misalignment, including what I term proper partial overlap, an illustration of which is given in (1), and an example of which is given in (2):

  1. abstract demonstration of proper partial overlap:
    1. SYNTAX: [x, [y, z]]
      1. {x} → A
      2. {y, z} → B (descriptively, we are used to calling B an “idiom”)
      1. {x, y} → R (descriptively, we are used to calling R a “suppletive fusional exponent”)
      2. {z} → S
  2. concrete example of proper partial overlap:
    1. SYNTAX: [PAST, [GO, OFF]]
      1. {PAST} → “before now”
      2. {GO, OFF} → “explode”
      1. {PAST, GO} → /wɛnt/
      2. {OFF} → /ɑf/

The expression in (2) is composed of smaller parts, both in terms of its semantics (“before now”, “explode”), and in terms of its morpho-phonology (/wɛnt/, /ɑf/). It would therefore be incorrect to claim that (2), as a whole, constitutes an ‘arbitrary’ pairing of form & meaning. At the same time, there is nothing else in (2) that constitutes a pairing of form & meaning, either – only pairings of abstract syntactic nodes with meaning (2.b.i‑ii), and separate, incommensurate pairings of abstract syntactic nodes with form (2.c.i‑ii). Thus, (2) involves no Saussurean signs whatsoever.

I will show that empirically, cases of proper partial overlap abound, as do other types of cases predicted by the proposed architecture. Lastly, I will argue that even those contemporary linguistic frameworks that distance themselves from outright Saussureanism, such as Distributed Morphology (Halle & Marantz 1993, 1994) and Nanosyntax (Starke 2009, Caha 2009, 2019), retain certain Saussurean vestiges that render them less explanatory than the current proposal.

The event poster.
View the PDF