[Top] [All Lists]

Re: [ontolog-forum] Practical Semantic Primitives

To: "[ontolog-forum]" <ontolog-forum@xxxxxxxxxxxxxxxx>
From: Gary Berg-Cross <gbergcross@xxxxxxxxx>
Date: Mon, 12 Aug 2013 10:31:30 -0400
Message-id: <CAMhe4f1PeUpK6=7QFO8TM4gd5ziuHikEsmu0ZmVKFC_M9vw2Mg@xxxxxxxxxxxxxx>

Some of this discussion has involved the idea of composition/construction/building of primitive "concepts"  into more complex concepts.

Bruce, for exampe said

 > As regards the ???atoms/molecules??? analogy ??? for me, the right approach

>is to look for a ???fundamental particle???.  Even atoms are composite

>structures.  If we are going to take a bottom-up approach to constructing

>every possible cognitive unit, we need to build these units from something

>truly fundamental.

One point here is that this composition is anything but simple or obvious. 

 We often fall back on some ideas we consider foundational like ideas of Constituent Parts into Wholes like words in a sentence. 

We fall very naturally into a Katz and Fodor type of additive semantic feature model like (Male) + (Human) +(Mature) = Man.

 Even this structure semantics appraoch turn out to be pretty challenging and just a small area of what seems to be involved in semantic processes composing a new concept out of a starter set of primitives. After all what does + in such a model mean?

Gary Berg-Cross, Ph.D.  
SOCoP Executive Secretary
Knowledge Strategies    
Potomac, MD

On Sat, Aug 10, 2013 at 2:06 PM, Bruce Schuman <bruceschuman@xxxxxxx> wrote:

Thanks to all for the comments and the replies.


Just to restate this point again -- my use of the word "primitive" has probably been confusing -- since for almost everybody it means something like "a list of words".


"Aristotle's primitives, which he called categories, include Substance, Quantity, Quality, Relation, Time, Position, State, Activity, and Passivity. These are ultimate primitives to which all other concepts are supposed to be reducible."  http://originresearch.com/sd/sd4.cfm   Many others use the term “primitive” in similar ways, looking for some “best” or “correct” list of words.


What I have meant by the term “primitive” might be better stated as something like “fundamental algebraic distinction”.  For me – all words – including “primitives” – are composite abstract units with implicit/nested dimensional decomposition, and the word is the “name” for that composite abstract unit, assembled from distinctions.


It might be totally true that “everybody knows what ‘tree’ means” (item 23 on the Swadesh list) – but if we need to distinguish “tree” from “not-tree” for some reason, we might have to get precise.


“Consider a tree. It has no sharp boundaries between parts; yet words divide the tree into trunk, roots, branches, bark, twigs, leaves, buds, knots, flowers, seeds, fruit, and even finer subparts such as veins in the leaves and pistils in the flowers. Even the boundary between the tree and the environment may be indistinct: the tree may have started as a sprout from the root of another tree and may still share a root system with its parents and siblings; insects and animals may be living in the tree; a vine may be climbing up the trunk, moss may be on the bark, fungus may be growing on a dead branch, and bacteria in root nodules may be supplying nutrients.”  http://originresearch.com/sd/sd4.cfm




I did take a close look at the "Swadesh List” -- http://en.wikipedia.org/wiki/Swadesh_list -- and yes it is interesting, and seems a reasonable stab at "a basic common human vocabulary".  In the USA, it would be reasonable to expect that the Democrats and the Republics and the Tea Party and the Progressives would probably be comfortable with this pool of common meanings.  This is the basis of how we can talk to each other at all.  But throw a big complex messy abstract word like "Obamacare" into that list -- and you've immediately got a firefight -- and a huge splintered division over what the word means.  Without a precise “stipulation” of its meaning, any conversation immediately becomes hopelessly fragmented –- and maybe dangerous.  When conversations are necessarily constrained by psychological bandwidth and time limits, the use of broad abstractions in a public context is very problematic.


As I was reading through the Swadesh list, I was reminded of a constraint mentioned by Smith and Medin (1981) in their important book "Categories and Concepts" -- where at the beginning, they clarify the meaning of the broad term "concept" to generally mean what they call "object concepts" -- things that can be named simply, the kind of things that fit into taxonomies.  They are not talking about "concept" as the term might be popularly used (just about anything anyone could think of) -- but in a narrower way that reduces the complexity.


On the Swadesh list, the verbs don’t start until item 54.  Most everything listed before that are basic objects, or the basic dimensions that describe objects.




John: I agree that the search for good distinctions and defining terms is useful.  But the world is a continuum, and the range of "games" that people play with words is open ended and constantly varying.  The common features among the games are fuzzy, rough, and squishy approximations, not primitives in a mathematical sense.


Bruce: This thought that “the world is a continuum” is basic to any understanding of concept formation – and is connected to the essentially ad hoc nature of most word usage.


“Concepts are inventions of the human mind used to construct a model of the world. They package reality into discrete units for further processing, they support powerful mechanisms for doing logic, and they are indispensable for precise, extended chains of reasoning. But concepts and percepts cannot form a perfect model of the world, -- they are abstractions that select features that are important for one purpose, but they ignore details and complexities that may be just as important for some other purpose. Leech (1974) noted that "bony structured" concepts form an imperfect match to a fuzzy world. People make black and white distinctions when the world consists of a continuum of shadings.”  http://originresearch.com/sd/sd4.cfm


The mystery is – why do we choose to parse experience as we do?  The answer has something to do with motivation – what is our point, what are we trying to accomplish? As a “continuum”, the “world” or “reality” could be parsed in an infinite number of alternative ways – but “for some reason” – we pick some particular way.


Your comment that “the games people play are open-ended and constantly varying”, as I understand it, has something to do with immediate context of communication – everything in the situation, everything I know about you and think you know about me.  If I want to successfully communicate with you, I use words I think you will understand – based on any clues at my disposal….


But my thought about “mathematical definitions” and why they are important emerges when the conversation becomes uncertain.  Item 52 on the Swadesh list is “heart”.  Could usage of that term be confusing?  Do you/I mean the physical heart that actually pumps the blood, or some sense of emotional heart or center (“she’s got a lot of heart”), or maybe “the heart of the town”?  If we get confused on that point, somebody might have to stipulate something.  Item 93 is “hot”.  The question might immediately arise, “How hot?”  Maybe all we need to know is that somebody thinks it’s hot.  But if it matters – I’m a mechanic fixing a car, or a doctor talking to a mother about a sick baby – I might have to know exactly what is meant – and that might require a dimensional stipulation.  In this sense, “hot” is a (brief/abstract) ordinal value for a numeric quantity that might have to be specified exactly – or specified within a defined boundary value range.


So – for me, when I see this idea that there are “games” going on with word meanings, I’m thinking two things: 1) the exact meaning is context specific, and hence plastic and fluent and context-dependent, so the “game” is figuring that out, and 2) if we are unsure, or feeling some tension, we might have to “negotiate” the meaning of the word – “hot” – or “cold” – or “dry” – or indeed, many other words on the Swadesh list.  Maybe that negotiation process is a kind of “game” as we settle in to an agreement on what we are talking about (for me, generally settling on a mutually-acceptable boundary-value range).


John: Words with similar meanings are the result of common aspects of human experience.


Bruce: This is the foundation of basic communication in language – and most of these Swadesh concepts have a very limited range of possible uncertainty.  Generally, “we know what they mean”, and their meanings are not highly context-specific.


Gary: On this view the meaning is not in, or embodied in, the word symbol. Rather the personal meaning of a word-symbol is the totality of what gets activated in a cognitive agent.


John: I agree.  As Wittgenstein said, the meaning of a word is its use.


The multiple word senses are the result of the many ways that a word is used.  For each individual, each word is linked to a growing and changing network of experiences.  Word senses are classifications that some lexicographer derived from citations of word uses.


Bruce: And what I would say is – that each use of a word – regardless of how widely understood or common the word might be – is essentially stipulative.  When I use a word in an act of communication – it means what I want it to mean at that moment – and I HOPE you understand it – drawing from all cues available to you, certainly including the common pool of meanings that you and I almost certainly share.


See Lewis Carroll on Humpty Dumpty talking to Alice in Through the Looking Glass:


    "I don't know what you mean by 'glory,'" Alice said.

    Humpty Dumpty smiled contemptuously. "Of course you don't - till I tell you. I meant 'there's a nice knock-down argument for you!'"

    "But 'glory' doesn't mean 'a nice knock-down argument'," Alice objected.

    "When I use a word," Humpty Dumpty said, in rather a scornful tone, "it means just what I choose it to mean — neither more nor less."

    "The question is," said Alice, "whether you can make words mean so many different things."

    "The question is," said Humpty Dumpty, "which is to be master — that's all."




Bruce (previously): For me – the word-symbol is a label for a composite and implicit block of distinctions and “meanings”.  For reasons of psychological economy, people use brief word-labels for complex ideas – leaving out many specific details and implications because we don’t have time for the details (“the weeds”), and we hope we can get away with skipping over them.


John: The stable words on the Swadesh list are good examples.  They're easy to recognize, they're associated with a frequently occurring experience, and they're islands of stability in the growing and changing network of associations.


Bruce: Yes.  A lot like the “concepts” discussed by Smith and Medin.


John: Nouns dominate the list of stable words on the Swadesh list.  The verb 'drink' is the most stable verb, but it's 38th in relative stability.


But note that emotional terms are absent from that list.  The word 'good' is the only evaluative term, and it's 79th in stability. People certainly use emotional and evaluative terms, but their meanings are not stable -- i.e., the ways they are used tend to change, and the word forms aren't preserved over the long term.


Bruce: Yes, these “qualitative variables” – as I might want to call them – are very problematic.  So, my approach tends to be “dimensional” – with the concept of dimension generalized to include qualitative (or “nominal” or “ordinal”) values – “good”, “bad”, “better”, “worse”, etc. – with the precise value in some particular usage entirely context-specific.


The framework to make this kind of stipulation meaningful is absolutely context-dependent.  At this level of abstraction, usage is totally local and dependent on intent.  And we might have to get into a bunch of “word games” (or “boundary-value negotiations”) to come to an agreement.



Message Archives: http://ontolog.cim3.net/forum/ontolog-forum/
Config Subscr: http://ontolog.cim3.net/mailman/listinfo/ontolog-forum/
Unsubscribe: mailto:ontolog-forum-leave@xxxxxxxxxxxxxxxx
Shared Files: http://ontolog.cim3.net/file/
Community Wiki: http://ontolog.cim3.net/wiki/
To join: http://ontolog.cim3.net/cgi-bin/wiki.pl?WikiHomePage#nid1J

Message Archives: http://ontolog.cim3.net/forum/ontolog-forum/  
Config Subscr: http://ontolog.cim3.net/mailman/listinfo/ontolog-forum/  
Unsubscribe: mailto:ontolog-forum-leave@xxxxxxxxxxxxxxxx
Shared Files: http://ontolog.cim3.net/file/
Community Wiki: http://ontolog.cim3.net/wiki/ 
To join: http://ontolog.cim3.net/cgi-bin/wiki.pl?WikiHomePage#nid1J    (01)

<Prev in Thread] Current Thread [Next in Thread>