Summarine

Computer simulations of language change notes

This website collects my personal notes on Computer simulations of language change. These notes are provided to bring full transparency to my research process. Of course, since they are only notes, they do not reflect my final thoughts on a topic, and should not be interpreted as such. To read finished papers, please consult my website. Do not use these notes as a basis for your own scientific research. Start from high-quality, peer-reviewed scientific literature instead.

Frequencies, probabilities, and association measures in usage-exemplar-based linguistics

Introduction

two large fields

  • generative grammar
  • cognitive/functional linguistics

While corpus linguistic approaches have been around for quite a while, the methodological evolution of corpus linguistics is still a relatively young development and many corpus-based studies are lacking the methodological sophistication of much of the experimental literature. This situation poses a bit of a challenge because, while a usage-based approach to language — an approach stipulating that the use of language affects the representation and processing language — does not require usage data, the two are of course highly compatible. This makes the development of an appropriate corpus-linguistic toolbox an important goal for usage-based linguistics.

this paper

  • collostructional analysis
  • developed to apply recent developments in corpus linguistics to issues and questions in cognitive/usage-based linguistics

Collostructional analysis: A brief overview

Perspective 1: CA and its goals

All of corpus linguistics is by definition based on frequencies — either on the question of whether something occurs (i.e., is a frequency n>0?) or not (i.e., is n=0?) or on the question of how often something occurs (how large is n?) — which makes it a distributional discipline. Since linguists are usually not that much interested in frequencies per se but rather structure, semantics/meaning, pragmatics/ function, etc., corpus-linguistic work has to make one very fundamental assumption, namely that distributional characteristics of an element reveal many if not most of its structural, semantic, and pragmatic characteristics

(p. 479)

collocations

  • the elements with which words in question co-occur

colligations

  • the grammatical patterns with which words co-occur

association measures

  • measures that downgrade/penalize words whose high frequency around a word of interest 𝑤 may reflect more their overall high frequency than their revealing association with 𝑤

Workflow of working with association measures

  1. retrieves all instances of a word 𝑤;
  2. computes an AM score for every collocate of 𝑤 (cf. Wiechmann 2008 or Pecina 2009 for overviews);
  3. ranks the collocates of 𝑤 by that score;
  4. explores the top 𝑡 collocates for functional patterns (where functional encompasses ‘semantic’, ‘pragmatic’, ‘information-structural’, …)

↳ the purpose of ranking words on the basis of such AMs

  • produce a ranking that will place words at the top of the list that
    1. have a relatively high frequency around 𝑤 while
    2. not being too frequent/promiscuous around other words

Perspective 2: CA and its mathematics/computation

collocation analysis

  • the extension of AMs
  • from lexical co-occurrence — a word M and its lexical collocates — to lexico-syntactic co-occurrence: a construction 𝑐 and the 𝑥 words 𝑤1, 𝑤2, …, 𝑤𝑥 in a particular slot of 𝑐

Bybee’s points of critique

Perspective 1: CA and its goals

most frequent implementation of CA

  • two features
    1. downgrades the influence of words that are frequent everywhere
    2. weighs more highly observed relative frequencies of co-occurrence that are based on high absolute frequencies of co-occurrence

Bybee’s criticism

  • “lexemes do not occur in corpora by pure chance” → “it is entirely possible that the factors that make a lexeme high frequency in a corpus are precisely the factors that make it a central and defining member of the category of lexemes that occurs in a slot in a construction”

Using the Spanish adjective solo ‘alone’ as an example, she goes on to say that, for solo, “Collostructional Analysis may give the wrong results [my emphasis, STG], because a high overall frequency will give the word solo a lower degree of attraction to the construction according to this formula” (2010: 98).

Perspective 2: CA and its mathematics/computation

bottom right cell

  • the number of constructions in the corpus

Bybee’s criticism

  • “there is no known way to count the number of constructions in a corpus because a given clause may instantiate multiple constructions”

Perspective 3: CA and its results, interpretation, and motivation

The perceived lack of semantics

collocation analysis

  • has a “lack of semantics”

Bybee’s criticism

  • “[p]roponents of Collostructional Analysis hope to arrive at a semantic analysis but do not include any semantic factors in their method. Since no semantic considerations go into the analysis, it seems plausible that no semantic analysis can emerge from it” (Bybee 2010: 98)

The perceived lacks of semantics and discriminatory power

The absence of cognitive mechanisms underlying CA

Gries and colleagues argue for their statistical method but do not propose a cognitive mechanism that corresponds to their analysis. By what cognitive mechanism does a language user devalue a lexeme in a construction if it is of high frequency generally? This is the question Collostructional Analysis must address. (2010: 100f.)

Clarifications, repudiations, and responses

Perspective 1: CA and its goals

  1. There are many flavours of AM.

Bybee’s logic would force us to say that the as-predicative, exemplified in (4) and discussed by Gries, Hampe & Schönefeld (2005), is most importantly characterized not by regard (the verb with the highest collostruction strength), but by see and describe, which occur more often in the as-predicative than regard (and maybe by know, which occurs nearly as often in the as-predicative as regard). Given the semantics of the as-predicative and the constructional promiscuity and semantic flexibility of especially see and know, this is an unintuitive result;

  1. V NP~Direct Object~ as complement constituent
  2. I never saw myself as a costume designer
  3. Politicians regard themselves as being closer to actors
  1. Relative frequencies are simply important, and informative

Perspective 2: CA and its mathematics/computation

The issue of the corpus size

“exact number of constructions”

  • cannot be easily estimated
  1. “a given clause may instantiate multiple constructions” (Bybee 2010: 98);
  2. researchers will disagree on the number of constructions a given clause instantiates;
  3. in a framework that does away with a separation of syntax and lexis, researchers will even disagree on the number of constructions a given word instantiates

obvious remedy

  • choose a level of granularity close to the one of the studied phenomenon

The distribution of pFYE

Many effects in learning, memory, and cognition are not linear

  • the power law of learning (cf. Anderson 1982, cited by Bybee herself);
  • word frequency effects are logarithmic (cf. Tryk 1986);
  • forgetting curves are logarithmic (as in priming effects; cf. Gries 2005, Szmrecsanyi 2006), …

Perspective 3: CA and its results, interpretation, and motivation

The perceived lacks of semantics

Problems

  1. her claim appears to contradict the exemplar-model perspective that permeates both her whole book and much of my own work; “[s]ince no semantic considerations go into the analysis, it seems plausible that no semantic analysis can emerge from it”

↕

There is a whole body of work in, e.g., computational (psycho)linguistics where purely frequency-based distributional analyses reveal functionally interpretable clusters. Two classics are Redington, Chater & Finch (1998) and Mintz, Newport & Bever (2002). Both discuss how multidimensional distributional analyses of co-occurrence frequencies reveal clusters that resemble something that, in cognitive linguistics, is considered to have semantic import, namely parts of speech. And even if one did not postulate a relation between parts of speech and semantics, both reveal that something can emerge from a statistical analysis (parts of speech) that did not enter into the analysis.

(p. 493)

  1. it does not engage fully with the literature

In sum, the statement that “[s]ince no semantic considerations go into the analysis, it seems plausible that no semantic analysis can emerge from it” can only be upheld by ignoring both the distributional linguistics literature that Bybee is otherwise sympathetic towards and the specific collostructional literature that she means to criticize and that shows the opposite.

  1. it is based on a partial representation of ca, and so it is really arguing against a straw man

The perceived lacks of semantics and discriminatory power

Bybee systematically chooses to not mention results of even a single study with experimental and/or corpus-based data running counter to her claims, but even a cursory glance at the literature shows that the picture is the opposite of the one she painted or, at least, much more complicated.

(p. 496)

The absence of cognitive mechanisms underlying CA

First, given the strong (experimental and otherwise) support of collostruction strength in many studies that all adopt a cognitive-linguistic/usage-based framework, it is surprising there should be a special need for a cognitive underpinning in addition to what all these studies are based on anyway. Second, the earliest studies make it very clear what their cognitive underpinning is. In Section 2.3 above, I already provided several quotes (from the studies Bybee refers to) to illustrate the CA position: Ultimately, collostruction strengths are based on (i) the conditional probabilities p(word|construction) and p(construction|word), which are related to notions of cue validity, cue reliability (cf. Goldberg 2006: Ch. 5–6 and Stefanowitsch to appear), associative learning measures such as ΔP, and prototype formation, and (ii) the frequencies that give rise to the probabilities, which are correlated with entrenchment. Put yet another way: “it is assumed […] that the statistical associations found in the data are reflected in psychological associations in the mind of the language user” (Stefanowitsch 2006:258).

(p. 496)

Towards a new empirical perspective and its theoretical implications

A cline of co-occurrence complexity and its motivations/implications

Approach 1: Raw frequencies/percentages

Approach 2: Association measures

Approach 3: Full cross-tabulation

Approach 4: Dispersion of (co-)occurrence

Why CA works at all and a brief excursus on Zipf

Towards a refined usage- / exemplar-based definition of construction

Computer simulations of language change notes

This website collects my personal notes on Computer simulations of language change. These notes are provided to bring full transparency to my research process. Of course, since they are only notes, they do not reflect my final thoughts on a topic, and should not be interpreted as such. To read finished papers, please consult my website. Do not use these notes as a basis for your own scientific research. Start from high-quality, peer-reviewed scientific literature instead.