Category Archives: Generative Approach

Perpetual Epicentral Density Sphere

Dandelion seeds (achenes) can be carried long ...

Image via Wikipedia

The Perpetual Epicentral Density Sphere is a unpublished model of cognition that I began developing in the late ninety’s and early years of 2000 as a result of my training in linguistics and anthropology.  I worked on it before I knew what Cognitive Linguistics actually was and, in fact, it was part of my conversion to CogLing.  With a nod to remix culture and assemblage theory, this model tried to blend together several validated models of discourse and culture in a super feeble attempt to bridge the epistemological gulf between realism and relativism and explain the complete communication picture in a systematic and procedurally elegant way (read: quasi-arbitrary).  This was the early days of my interest in systems theory, but I was really naive about the complexity of complexity.  Basically, the more that I worked on the model the more I came to see that Cognitive Linguistics already had working solutions to a lot of the questions I was addressing.  In fact, when I read back through my notes now, I see that it is actually a model of attention and dynamic construal.  I won’t tell you anything more about the mechanics right now, but I expect that one day I will pull it back off the shelf and show it to the world; in the mean time, here is an analogy from my original manuscript (part of which is in this book) for you to chew on:

“Pick the white puff of seeds on a dandelion clock, pick the stem and hold it in your hand, and meticulously and decidedly, remove every single floating seed, one-by-one until there is only one remaining.  This is what the first thought does when it moves to the second thought.  You have deselected every seed, ignored them all but one, the one you highlighted, the one you lit up – the one you selected.  This, the lone seed on the stem, a sphere at the base, the rod line of the seed body, and the end of the seed a circle of tiny white hairs that extend radially from the stem in many many directions, this seed is a snapshot. You are holding in your hand the most natural visible representation of how your thoughts live and travel.  Now put that last seed up to your mouth and do the wind’s job and send that seed floating.  Where it lands it either develops or dies, just like your thought.”

“If you can take a line of thought and follow it to every dead-end, out every open door and window, to every destination – and catch it resting – you will see simply and plain how lawless and unruly even simple thoughts are in their brief lives.”

You can view images from my portfolio that were inspired through this process at: RyanDewey.org

Advertisements
Tagged , , , , , , , ,

Why I Care About Ontology

Staircase perspective.

Image via Wikipedia

In a broad sense of the term, any particular ontology serves as a framework against which we interpret information; think of it like an organized perspective that we use as a lens to view and understand the world.

When you study a culture you are concerned with figuring out how they understand the world and how they make sense of the world.  This study of sense-making follows the concept that people do what seems logical and rational to their perspective.  I am interested in mapping these sense-making resources and translating them into a format that reveals the internal consistency and relationships between the rationale and the external context and stimuli.

If you look back through the histories of science & philosophy you will see the pendulum swing between two extremes in regard to ontology, one extreme believes that there is a unified and complete ontological structure to the world, the other extreme believes that there are many unified and complete ontological structures.  This is the debate between whether we can know truth objectively, or whether it is known as a perspective.

In my work I explore how we construct global formal ontologies as well as how we generate idiosyncratic folk-ontologies; this is the both/and response – I want to know how we engage in thinking when it is shaped by public beliefs and private beliefs and what it takes to reconcile the disjunctions.

One of the ideas that has kept my attention the longest is that the structure of a thought represents the choices that have been made to arrive at that thought, and that structure also shows us what has been ignored as the thought is being formed.  I want to know how the external world interacts with the internal world.  I want to know how it directs our thinking.  This is the foundational riddle that makes me want to do linguistics.

Tagged , , , , , , , , , , , , ,

On Failure & Resilience in Optimization of Human Systems, Ecological Systems, and Networked Systems of Systems

I was recently watching Eleanor Saitta’s talk called “Your Infrastructure Will Kill You“.  Part of her talk outlined how optimization equals fragility (more or less).  That to the degree that something is cleaner, more elegant, or more efficient, it is fragile, and a break in the system can be potentially catastrophic.

In thinking about her comments I thought of a few examples where I have observed optimization creating a state of fragility, here are a few of my thoughts:

  • This has interesting considerations for general principles of design, specifically the form/function aspect of design. Probably the point at which form begins to extend beyond the needs of function the focus on form becomes gratuitous and potentially even hazardous (depending on the type of system).  Ironically, optimization in this case is not absolute optimization, but only optimization considering a specific set of requirements: when things are running smoothly then the system is optimized.
  • Another place for failure is when relationships are optimized; when the dispersal of information through a system relies on optimized relationships it only takes the breaking of one of the nodes in a network to create a chain reaction of subsequent nodes being uninformed.We think about how a well-connected network effectively distributes information, specifically in recent thought this informed an analysis of William Dawes vs. Paul Revere – showing how Revere’s relationship of network brokers enabled him to broadcast more extensively than Dawes’ impoverished closed network.  This is good thinking, except that it misses the point of threat: Revere was a weak link in that optimized chain of information; had he been eliminated his message would have been eliminated.  Revere and his network, although connected and optimized, were fragile.
  • In generative linguistics there is an optimization of the lexicon. Economy in space is valued above economy in processing; if this is opaque to you, I mean that generative linguistics tries to minimize the amount of information that it stores as unique units.  It is called “generative” because it generates complex utterances from values stored in the lexicon through recursion, instead of storing those values as wholes.  But there is a weakness; in optimizing the lexicon the generative power of the spell-out rules of Universal Grammar are fragile when it comes to dealing with actual language usage (which is the test of a linguistic theory, is it not?), and the rules fail to account for some foundational constructions of language (like idioms for example).

Ok, so there are lots of places where optimization leads to failure, but what are some ways in which optimization leads to resilience?  What are some solutions to these problems?

Redundancy is a great solution, but it is bulky.

  • When function is optimized it allows you to work backwards in the process of making things have better forms. This is actually how a lot of design progresses.  Think about how every piece of electronic technology that we have today had a larger predecessor.  Think about those clunky mobile phones from the eighties with the handset, base and cord in a leather bag, now look at the mobile phone you carry in your pocket.  Functionality was concept proven in the clunky design, and the form was optimized to enhance the function.
  • It seems to me that the optimization weakness in the Revere incident was that Revere was the weak link.  Instead of depending on the optimization of Revere’s relationships, perhaps the message itself needed optimization. One possible way to optimize the information load of a message is to abstract it (as was done with the signal of the hanging lanterns to indicate the route of attack), and another way is to reduce dependence upon a single messenger exploiting a network (In 1775 the sexton who hung the lanterns was a single messenger, Revere was a single messenger, et cetera).  Flooding a network with messengers bearing an abstracted signal would have been less fragile (put aside for the moment the need for secrecy in the 1775 incident).  In situations where secrecy is not vital, consider how this kind of network flooding would communicate the coherence of the message; when you hear the same thing from five people you at least start giving some credence to the constancy of the message.  In such cases what may have been unknown or even background information becomes salient and foregrounded through repeated exposure. Also, consider other types of signals that can be exploited to prompt a response of crowd mobilization, noises work particularly well.  Sirens and loud noises alert and orient people’s attentional systems toward the source of the signal, and that source becomes figural in the contextual noise of that signal.
  • Optimization can lead to resilience in online processing, like in a maximized lexicon that places the task of optimization on the processing skills required in the selection and extraction of form-meaning elements from the inventory.  Cognitive approaches to the lexicon seek to preserve economy in processing; instead of having a minimal lexicon with lots of processing rules, the cognitive approach has an ordered inventory of form-meaning pairings (including monomorphemic elements, constructions, and phrases that are learned whole), with an optimized processing system of constraints, schemas, and other elements of cognitive processing (see this week’s post on emergentist vs. universalist view for understanding the contrast in general cognitive processing vs. modular mind).
Tagged , , , , , , , , , , , , , , , , ,

An Emergentist vs. Universalist view of Language and Cognition

Distributed cognition

Image by Lisa Brewster via Flickr

I wanted to present a list that outlines some of the main differences in thought about language between Emergentist and Universalist perspectives.  This is important I think because it shows how only certain kinds of programmers and mathematicians can work successfully within a Cognitive framework.

Consider these characteristics of an Emergentist (Cognitive) view:

  1. Singular Mind (General Cognitive Abilities)
  2. Distributed Cognition
  3. Neo-Empiricist
  4. The Complex System IS the primitive
  5. Prototypes
  6. Online and Dynamic Processing
  7. Usage Based View of Language
  8. Falsifiable
  9. The Appropriate Level of Granularity is the Form-Meaning Pair (i.e., constructions)

Now, compare that list with this Universalist (usually Generative) view on the same issues:

  1. Modular Mind
  2. Localization in Neuroscience
  3. Innate
  4. Atomistic, Reduce!
  5. Feature based categories & Atomistic Set Theory
  6. Stable Structures and “Switches” that enable cognition
  7. Competence Based View of Language
  8. Language is the de facto expected product of the mind
  9. Reductionism refines phenomena out of existence

Can some middle perspective be taken that combines both extremes?  What are your thoughts?

Tagged , , , , , , ,

Token Frequency and a Usage-Based Grammar

Token Frequency motivates learning through acknowledgement of repetition and familiarity.  This frequency reflects the type as it is instantiated through the various token usages.  In the Usage-Based perspective the token is the instance that is repeated and subsequently learned, giving rise to the type, and by extension, equating the type with its function.  Type and Token relate functionally and enable an emergence of grammar (without depending on the a priori insistence on Universal Grammar).

A Generative account does not need the idea of repetition to ground the learning of constructions since Universal Grammar already contains the abstraction of constructions.  Generative thought does not arrive at the Type by repetition, as the Type exists in UG, and is made manifest as Tokens in language use.  Token frequency is irrelevant in Generative models, except as it correlates to speaker competence and performance, and as it is paradigmatically selected in the UG compliant syntax.  In this way, Token and Type relate formally to one another and enable rules like PSRs to maximize multiple reinvestment of language.  The Generative understanding of Type:Token is thus driven in a top-down direction, not permitting token frequency to alter the grammar from the ground up since UG dominates from the top of the hierarchy.

This is why I think that the Usage-Based approach more adequately addresses the emergence of grammar through habituation and entrenchment.

Tagged , , , , , , , , , , ,

Sample Sentences Using Spradley’s Nine Semantic Relations from The Ethnographic Interview

Cover of "The Ethnographic Interview"

Cover of The Ethnographic Interview

I love James Spradley’s work on ethnographic interviews, componential analysis, taxonomic analysis, and participant observation, but Spradley’s work on semantic analysis has been the most thought-provoking for me theoretically.  Here I list out his nine semantic relationships and give some sample descriptive sentences to show you how the semantic relation describes the two elements in the relationship.  I have to say, however, that none of these sentences are very natural in a natural language kind of way.  In fact, the one concern that I have with Spradley’s view of semantics (from my usage-based cognitive view of language) is that it does not adequately lend itself to a straightforward modeling of the semantics of a natural language sentence.  Instead, if you want to use this for natural language, it has to be on a propositional level.

These semantics are best for modeling culture and the dynamics of a culture.  After all, they were drawn up in a methodology for ethnography.  In the sentences I present below you will find that they have a rigid and non-human sound to them; in fact, I think (and this is my opinion), that if you want to use Spradley’s semantics for anything other than modeling culture, that they are best used in formal system modeling, such as an expert system. Continue reading

Tagged , , , , , , , , , ,

What does Soulwax’s website, DJing, & Construction Grammar have in common?

Soulwax’s website extends an invitation for viewers to participate in DJing as they explore the website.  From my first exposure this has been an amazing experience.  The intuitive guided navigation doubles as a loading of the clips so that your browser cache holds the clip for later manipulation in the mixing.  If you patiently experience each of the clips instead of navigating away from the site, you will get the chance to mix the video loops and beats by clicking your mouse. Continue reading

Tagged , , , , , , , , , , , , , , , , , , , , ,

My Top 10 All-Time-Favorite Formal Linguistics Books

Generated using http://www.ironcreek.net/phpsy...

Image via Wikipedia

Occasionally I read a good Generative book to balance my thinking in linguistic theory…I am drawn to the ideas of formalism, usually when I am thinking about how to develop anthropological linguistic approaches for Cognitive Linguistics field work in Non-Indo-European languages, but also because I like the feel of generative thought a little.

Here are some of the generative books that I love the best:

  1. Lexical Categories – Baker
  2. Optimality Theory – Kager
  3. Phonology in Generative Grammar – Kenstowicz
  4. Tools for Analyzing the World’s Languages – Bickford
  5. Introduction to Typology – Whaley
  6. Language Form and Language Function – Newmeyer

And then there are a class of books that are in accord with generative thought (i.e., modular mind) that I find to be classic inspiration for the study of conceptualization:

  1. The Society of Mind – Minsky
  2. What Is Thought – Baum
  3. Consciousness and the Computational Mind – Jackendoff
  4. A Prosodic Model of Sign Language Phonology – Brentari

If you are a Cognitive Linguist and you are wondering for some good places to start studying Generative/Transformational/Formal Linguistics this list should be a good primer for you.  It covers all of the modules: the lexicon, phonology, syntax, morphology, meaning/semantics; as well as touching on topics like: theory of mind, compositionality, categorization, typology, rewrite rules, distribution and word order, signed languages, development, and gesture.  Enjoy!

If you are a Generative Linguist stumbling across this post and wonder where to begin reading about Cognitive Linguistics, please go to the menu of this page and select “CogLing 100 Book List” which will get you going in the right direction.  Bon Chance!

Tagged , , , , , , , , , , , , , , , ,

CogDef #1 (cognitive definitions)

I have several friends who are strict Generativists/Formalists and I want to start a little series that summarizes information about topics that sometimes get muddled up in cross-theoretical discussions.  This is the first in the series.

In the Cognitive Enterprise semantics drives the modeling in grammar (which is why it is a functional model). Grammar does not strictly mean syntax, as formal theories assume, but entails the entire package of language. Here is how we break down the approaches to semantics and syntax:

  1. Cognitive Semantics is used as a lens to view the conceptual structure and conceptualization
  2. Cognitive Grammar is used to view the cognitive principles that give rise to linguistic orientation
  3. Construction Grammar is used to view the units that emerge from cognitive semantics & comprise the grammar.

Going back through Evans & Green I found that I had written these definitions in the margin of page 49, hopefully they are clear.

Bibliography

Evans. V. & Green. M. (2006). Cognitive Linguistics, an introduction, LEA

Tagged , , , , , , , , , , , , , ,

Mashing Ray Jackendoff’s Ontological Categories with James Spradley’s Cultural Dimensions & Wh- Question Words

While I don’t necessarily agree with Jackendoff’s use of the semantic primitives, and while I recognize that this comes from a book that is from the early 80’s, I still want to present these ontological categories because they represent one way that the world is sliced, and they have interesting correlations to wh– question words.

Frankly, they are a little too close to Aristotalian categories for me to have much use for them in the stuff I am doing, but still, this represents an interesting perspective that ties in with ethnographic inquiry [specifically James Spradley’s work in Participant Observation].

Jackendoff proposes categories like THING, PLACE, DIRECTION, ACTION, EVENT, MANNER, AMOUNT and proposes a correlate question word which would access this category.

Spradley talks about 9 cultural dimensions: SPACE, OBJECT, ACT, ACTIVITY, EVENT, TIME, ACTOR, GOAL, FEELING, and looking at these nine dimensions in a matrix provides an ethnographer with 81 questions to keep their observations focused. But Jackendoff is more interested in seeing how these types of categories are manifest in linguistic structure, and less interested in modeling culture. Continue reading

Tagged , , , ,

Cognitive Mindfulness #14

This comes from the introduction to a book edited by Dirk Geeraerts:

If conceptual perspectivization is the central function of a grammar, the typical formal categories of grammatical description (like word classes or inflection) will have to be reinterpreted from a semantic point of view. [7 Geeraerts, 2006]

It made me think twice…especially since syntax has been the normal focus of grammar in the last 50-70 years; a meaning based grammar makes a lot more sense to me than a rewrite rule based grammar.

Bibliography:

Geeraerts, Dirk, editor, (2006). Cognitive Linguistics: Basic Readings, Mouton de Gruyter

Tagged , , , , , , , , , , , , ,

Lady Gaga Is Her Own Black Box

Sorry about the obtuse title, I don’t know how to name this odd post that integrates multiple concepts that tangentially relate.  Hopefully the ambiguity of description entices you to read.

It seems to me, that because Cognitive Semantics is constructional rather than compositional, it is suited for integration with Object Oriented Philosophy.

This statement might get some negative reactions, but I am learning, so, bring on the criticism.

Here is why I think Constructions are better suited for OOO than a strict Compositional approach: at the level of a relation between two hierarchies, concepts which construct do not need to regard the hierarchical levels as boundaries which must be maintained, in other words, a hierarchical element on level 5 in hierarchy A does not have to match across hierarchies to another hierarchical element on level 5 in hierarchy B. Continue reading

Tagged , , , , , , , , , , , , , , , , , , , , ,

Cognitive Science & Engineering by Deductive Reasoning

Flywheel from old factory

Image via Wikipedia

Here is a brief passage from my book The Art of War Against Boredom.  I wrote this passage around 8 years ago, and while it is influenced by my background in descriptive linguistics as opposed to strict cognitive linguistics, I still feel that it has something to say about the cognitive enterprise.  For example, in cognitive linguistics the actual language in use reflects the mental processing which produced that actual language.  In this passage below, the designed object reflects the mathematical processes which drove the production of the designed object.  I recognize that this can be interpreted through a generative lens too, but the passage isn’t meant to illustrate linguistic theory, it is a folk-methodology for problem solving.  Anyway, the designed object reflects the process of production. Continue reading

Tagged , , , , , , , , , , , , , , , , , , , , , , , ,

Faint Hints of Functionalism in a Formalist Approach? That’s ok… right?

“It is important to bear in mind that the stratified hierarchies which arise in the course of progress of the algorithm are hypothetical. At each moment during the learning process, stratified hierarchies represent the current knowledge which the learner at that moment has accumulated about constraint interactions underlying a given output form. This knowledge is dynamic, as it continuously changes while the learning algorithm processes information from output forms.” [Kager, 299]

Tagged , , , , , , ,