Ronald W. Langacker
Foundations of cognitive grammar, Vol. 1, Theoretical Prerequisites.
Langacker, Ronald W.;
Foundations of cognitive grammar, Vol. 1, Theoretical Prerequisites.
Stanford University Press, 1987 [gbook]
ISBN 0804738513, 9780804738514
topics: | cognitive | language | linguistics
a rather dense text, hard to penetrate. but this is the pioneering work, that more than any other, re-established the primacy of meaning after the "miracle"-ridden the chomskyan quagmire.
don't worry, later books by langacker don't get any easier.
Excerpt
An obvious but seldom-made observation is that any polymorphic linguistic sign (this includes the vast majority of expressions) is nonarbitrary to the extent that it is analyzable. For example, given that staple means what it does, and that -er means what it does, it is anything but arbitrary that stapler is the form used in English for a stapling device. The principle is therefore limited largely to individual morphemes, but even here it must be significantly qualified. I refer not only to obvious cases of onomatopoeia, but more broadly to the pervasive influence of analogy and subtle forms of sound symbolism as constant contributory (though seldom determinative) factors in the gradual evolution of our lexical stock ([Rhodes and Lawler 81; Samuels 72; Haiman 83]).
[1: The form is conventional, inasmuch as another form could perfectly well have been chosen for this concept, but it is not arbitrary in the sense of being unmotivated, given the existence of other signs.]
MODULARITY:
There is no question that people have the capacity to learn a language
and that this involves innate structures and abilities. What is
contenious is whether some of these structures are unique to
language... In my opinion a convincing case has not yet been made for
a unique linguistic faculty. 13
DISCRETENESS:
CRITERIAL-ATTRIBUTE model: +/- against attributes that need to be
fulfilled to belong to category. Class membersship is thus an
all-or-nothing affair. 16
PROTOTYPE model: categories are organized around common,
run-of-the-mill members of the category. Psych: respondents
accept them as class members with shortest response latencies
==> Statistically, prototypical members are more likely to behave in
a prototypical way (depending on how central it is).
Dichotomies posited by linguists, that are often accorded a
theoretical status quite unjustified by the facts: 18
- Grammaticality / ungrammaticality: linguists themselves note the
inadequacy of this dichotomy, through ad hoc notations [that capture
various gradations of ungrammaticality] such as ?, ??, ???, ?*, *, **,
*** etc. 15
- synchrony / diachrony
- competence / performance
- grammar / lexicon
- morphology / syntax
- semantics / pragmatics
- rule / analogy
- homonymy / polysemy
- connotation / denotation
- norphophonemic / phonologica;
- derivational / inflectional morphology
- vagueness / ambiguity
- literal / figurative lg
INTEGRAL SYSTEMS - holistic description rather than decompositional:
the vowel [i] has the features [-consonantal, +vocalic, +high, +front,
-rounded] attempts to decompose something that is integral - these
features, while present, do not account for it alone. making these
movements without the correct timing will not generate the sound, ...
[hence these are not sufficient]
"while eating, my tongue accidentally assumes the shape and position
that would be appropriate for the sound [i]. I cannot claim to have
thereby implemented the phonological feature [+HIGH] or [+FRONT] -
the feature exists (in the sense linguists understand the term) only
in the context of its utterance. 21
similarly, uncle = [male, collateral, ascending generation] - but
in normal use of the concept, we manipulate the concept holistically,
without these separate attributes. But they are also there - no
conflict between a decomposition and the whole co-existing. 20
IDIOMS as unanalyzable whole - Tranformational grammar view: idioms
are inserted from the lexicon in a specific linear order, and
subsequently, transformational operations may be applied. examples
commonly cited for passive transformations include idioms such as
"Headway seems to have been steadily made". These assumptions make
the conclusion of transformational derivation inescapable [Fraser 70 -
the best TG study of idiom]. 24
but most idioms are analyzable to some
degree. Gorbet (1973) - many idiom chunks participate in anaphora -
(both pronomonial and ellipsis):
4a Anthony stole her heart and then broke it.
b First he broke her heart and then her spirit.
c After making no headway all morning we finally made some in the
afternoon. 24
[NOTE: Grammaticality is a matter of degree - I am not sure I agree
with c]
Example 5 indicates further analyzability:
5a. We didn't make the amount of headway that was expected of us.
EXCLUSIONARY FALLACY - that one analysis or explanation for a
linguistic phenomenon precludes another.
e.g. forms like "stapler" - means more than just 'something that
staples'.
RULE/LIST fallacy: if it is listed in the lexicon, then it cannot be
part of the productive 'V + -er' derivational pattern.
But nothing intrinisically implausible about a poisition combining
both analyses. In addition to being an exemplar of this pattern,
stapler is also specialized in other ways ...
CLASSIC RULE OF DO-SUPPORT
auxiliary-like verb DO (chomsky 57), and also be / have, and their
counterparts in other languages (Bach 67]. In some constructions, do
bears a tense morpheme that would otherwise be strande:
7a. Do you like children?
b. They do not sound serious.
c. I did warn you.
d. Kittens like pizza, don't they?
GRAMMATICALIZATION:
Based on this predictability, it is concluded that do is
transformationally inserted. It is a prototypical grammatical
morpheme, being semantically empty ... Linguistic morpheme might
perfectly well adopt a morpheme with limited semantic content for
particular grammatical purposes; serving a grammatical function is not
inherently incompatible with being meaningful.
[**IDEA: refer w.r.t. kar in CP paper]
There is reason to doubt context-independence of grammaticality
judgment. Judgments about sentences are made relative to real or
imagined contexts. E.g. 11b is oten taken to be ungrammatical, and
constitutes the ground for the rule Equi-NP deletion:
11a. * I want me to be elected
b. I want to be elected
Yet the grammaticality of a is vastly improved (though still marginal)
when it is an answer to
12 who do you want to be elected?
when it is notably better than 11b.
SALIENCE
DATIVE SHIFT: supposedly derives structures like 13b from 13a:
13a he sent a letter to susan
b he sent susan a letter
these sentences have the same truth value and can be used
interchangeably to describe the same event but I suggest that they
differ semantically (cf Goldsmith 1980) - 13a emphasizes the path of
the letter, 13b the resulting state in which susan possesses the
letter ==> they differ in their IMAGES
X VERB Y to Z ==> path traversed by Y
X VERB Z Y ==> Z controls Y
14a The shorstop threw a ball to the fence
*The shortsop threw the fence a ball
semantic compatibility - fence as *possessor of ball. Reversed in:
15a ?Your cousin gave a new coat of paint to the fence
b Your cousin gave the fence a new coat of paint
here notion of path is far less salient.
Theories of AUTONOMOUS SYNTAX have a natural affinity for truth-value
semantics. 40
--
Lists or Rules? In reality there is a third choice - both rules and
lists. 42
[PROBLEM: Which is to be used when? Or must they address separate
domains? or separate degrees of familiarity? ]
I have learned that 13x13 = 169. But I can also use the principle of
multiplication to derive this product. Fixed unit and the ability to
compute it are consistent. 44
e.g. N+ -s rule of English - plurals are learned previously as fixed
units, and then they are generalized. Speakers do not necessarily
forget the forms they already know once the rule is extracted. Particular
statements (specific forms) coexist with general statements (rules accounting
for those forms) in a speaker's representation of linguistic convention, which
include a huge inventory of specific forms learned as units (conventional
expressions). Out of this sea o particularity speakers extract whatever
generalizations they can. 46
Linguists are also known to harbor false expectations at a higher level of
generality -- that of language universals. An example from an earlier phase
of generative theory is the notion that the apparent grammatical diversity of
languages is largely superficial: that as we penetrate beneath the surface to
more absrtact levels of representation, languages begin to appear much more
similar, perhaps even identical. According to this view, the underlying
syntactic uniformity of languages is obscured at the surface level by the
operation of grammatical rules, so it is at the level of underlying structures
that we find the most extensive grammatical universals. In contrast,
cognitive grammar claims that grammatical structure is mostly overt. [46]
Is semantic structure an universal? [L believed so once upon a time]
but I had not taken into account the pervasive importance of imagery,
ie our ability to construe a conceived situation in many diff ways
(seeing it from diff perspectives, emphasizing
certain facets over others, approaching it at diff levels of
abstraction etc). languages differ e.g. L1: I am cold, L2: I have
cold, L3: It is cold to me - though they describe the same experience, L's
differ in the images they employ to structure the same conceptual
content. Full universality of semantic structure cannot be presumed
even on the assumption that human cognitive ability and experience are
quite comparable across structures. 47
ABSOLUTE PREDICTABILITY - a statement pertaining to a class must be
valid for all and only the members of that class. Assumes that
language invariably or even typically lends itself to statements of
this kind. In fact, it does not. Such expectations are unreasonable
for NL and commonly lead to erroneous conclusions or dubious claims. 48
Chapter 2 FUNDAMENTAL CONCEPTS
A unit is a structure that a speaker has mastered quite throroughly,
to the extent that he can employ it in largely automatic fashion,
without having to focus his attention specifically on its individual
parts or their arrangements. Despite its internal complexity, a unit
for the speaker ... is effortlessly simple, and does not demand the
constructive effort required for the production of novel structures.
Psychologists would speak of a "habit", or say that "automatization"
has occurred. [57]
PHONOLOGICAL UNITS
The basic sounds of a language are clearly units for fluent speakers.
e.g. French speakers fluently produce the sound \"u.
An English speaker learning French may be told to pronounce a vowel
like i, but with the lips rounded for u. Following these
instructions, he may pronounce it more or less correctly on the first
attempt, but doing so requires a constructive effort.
Phonological structures much larger than segments also achieve unit
status - including syllables, words, familiar phrases, and even larger
sequences. We can assume that any word in everyday use constitutes an
unit.
SEMANTIC UNIT = established concepts - novel concepts require a
constructive effort.
SYMBOLIC UNIT
The symbolic association between the phonological unit and the
semantic unit can itself become automatized, and gain unit status.
The simplest symbolic unit - morpheme - semantic and phonological
units participate in unalazyable wholes in a symbolic relationship.
Basic symbolic unis combine to form progressively larger symbolic
structures, which may themselves be mastered as units; the grammar
thus contains large inventory of conventional expressions (not
restricted to idioms or MWEs). Grammatical patterns are analyzed as
SCHEMATIC symbolic units, which differ from other symbolic structures
not in kind, but only in the degree of specificity.
When an idea coincides with the semantic structure in a symbolic unit,
then the phonological structure is automatically called to mind. When
no unit available, need problem-solving activity - assemble the
expression from smaller units. Because of the vast number of (larger)
units, the actual constructive effort is easily overstated. Say an
user has mastered the combination of [A], [B], [C], so as to form the
higher-order unit [A-B-C], as well as the combination [D] and [E] as
the unit [D-E], then creating the larger but non-unit structure
[A-B-C-D-E] demands the combination of just two units, by definition
manipulable as wholes. 59
coding: find ling expression for a conceptualization
target structure - the solution found 65
Nonverbal thought - consider the task of working on a jigsaw puzzle. 65
visual inspection to see if a piece fits involves decomposition into
protrusions, depressions, corners, sides etc - and syntactic operations to
check for matching - comparisons of shape / colour. How deep the semantic
structure is a function of the task [90]
Difficulty of distinguishing verbal vs non-verbal symbols: many sounds
used for non-linguistic tasks - e.g. humming uses voicing, rhythm and
pitch control). Similarly, many
gestures have become conventionalized in many cultures, but linguists
would balk at including gestures in the grammar of a language, despite
their symbolic character. Even limiting to symbols with a
phonological structure is problematic - ringing of a dinner bell? So
then sounds produced by human apparatus [but machine synthesized
speech?] - but what of meaingful noises like wolf whistles. Can say
that output has to be segmental in character - but what of clicks some
people use to direct a horse or call a cat? 60-61
But we may have already gone too far - intonation contours surely fall
withint the purview of this desription. Also excluded are certain
phenomena :
1a. This one is much better than that one. [difficult to utter without
accompanying gestures]
b When she saw the snake she went eeeeeeeee [SCREAM]
In b an imitative vocalization (can also be gesture or enactment) is
incorporated as an apparent constituent.
BETTER: prototypical linguistic symbols have for their realization
segmentally organized sound sequences produced by the human vocal
apparatus, whereas other kinds of symbols and symbolic systems that we
would hesitate to call non-linguistic
depart from the prototype in various ways:
- Am Sign Lang - visual mode
- Intonation - nonsegmental; abstractness of meanings
- Jabberwocky - content words have no semantic value 62
CONVENTIONAL LINGUISTIC UNITS: [e-language vs i-language]
what constitutes the set of conventional units? the set of units
mastered by all members of any speech community may be a rather small
proportion of the units constituting the linguistic ability of any
given speaker - so what is a grammar to encompass?
single person's cognitive organization - presumably of a
"representative" speaker.
Another aspect: sociolinguistic connotation is part of the speakers
linguistic knowledge - e.g. L's (English) knows that ain't
is stigmatized, sir indicates the relative social status of speaker
and addresse, deja vu is a borrowing from French, and the terms
subjacency and Move-Alpha are restricted to a professional subset.
To the extent that speakers learn the sociolinguistic status of
conventional units, it constitutes an aspect of their linguistic value
and is thus a proper concern for linguistic description. 63
[NEW WORDS:
idiolect - ??particular individual's language
chiller - the rack in the fridge just below the freezer
- have a role in communicating these ideas to others, but also in
encapsulating it so that the concept can be further enriched.
fr. Gk idios - "one's own", + lect fr Gk legein "speak." (dialect) ->
lexis speech, word ==> lexicon
LECTURE (N.) :: 1398, "action of reading, that which is read," from M.L. lectura
"a reading, lecture," from L. lectus, pp. of legere "to read," also "to
gather, collect, pick out, choose" (cf. election), from PIE *leg- "to pick
together, gather, collect" (cf. Gk. legein "to pick out, choose," also "to
say, tell, speak, declare;" lexis "speech, diction;" logos "word, speech,
thought, account"). To read is to "pick out words." Meaning "action of
reading (a lesson) aloud" is from 1526. That of "a discourse on a given
subject before an audience for purposes of instruction" is from 1536. The
verb is attested from 1590.
IDIOM :: 1588, "form of speech peculiar to a people or place," from
M.Fr. idiome, from L.L. idioma "a peculiarity in language," from Gk. idioma
"peculiarity, peculiar phraseology," from idioumai "I make my own," from
idios "one's own, of a particular person, personal." Idiomatic is first
attested 1712.
IDIOSYNCRASY :: 1604, from idios "one's own" + synkrasis "temperament, mixture of
personal characteristics," from syn "together" + krasis "mixture."
Originally in Eng. a medical term meaning "physical constitution of an
individual." Mental sense first attested 1665.
IDIOT :: c.1300, from L. idiota "ordinary person, layman," in L.L. "uneducated
or ignorant person," from Gk. idiotes "layman, person lacking professional
skill," lit. "private person," used patronizingly for "ignorant person,"
from idios "one's own."
"Reader, suppose you were an idiot. And suppose you were a
member of Congress. But I repeat myself." [Mark Twain, c.1882]
Idiocy (1487) is perhaps modeled on prophet/prophecy.
[IDEA: Rule that generates idiocy?]
]
FEATURES OF COG GRAMMAR:
DOMAINS -
semantic units characterized relative to a cognitive domain - any
conceptual knowledge system can serve as a domain - social
relationship, conception of speech situation, existence of various
dialects, etc.
ENCYCLOPEDIC
any semantic unit involves many conceptual domains - some more
central to its value than others.
[**IDEA: Encyclopedic - use in Arjun paper
Decontextualization - key idea in Kiran model - how does it differ
from Pust "decomposition"? ]
DECONTEXTUALIZATION -
key to acquiring "units". If a property (e.g. relative social
status of speaker and hearer) is constant to the context whenever an
expression is used, the property may remain a part of the semantic
specification of the unit.
GRAMMAR AS PROCESS - generativity - series of operations that gives
well-formed sentences as output. 64
Generativity: one cannot assume without question that generativity is
a reasonable requirement to impose on linguistic descriptions, or
that explicitness and rigour demand it.
If generativity is interpreted to mean that the grammar itself must
fully and explicitly enumerate all and only the well-formed
expressions of a language, including their semantic structures,
then an encyclopedic model will be ruled out. 64
[?DISCUSS But is this due to the Finite grammar assumption - does encyclopedic
mean infinite? ]
Grammar as Autonomous formal system
Cannot exclude from the grammar (and hence from serious consideration)
linguistic usage and figurative language - which are pivotal to
understanding linguistic structure. 64
[COMMENT: Giving well-formed sentences as "output" - but the autonomous
grammar does not do that. What is the input? It is only analytic, not
synthetic - only analyzes linguistic fragments, there is no other input!]
ELEMENTS OF GRAMMAR:
- specified symbolic units - morphemes, polymorphemic lexical items,
and larger conventional expressions.
- schematic symbolic units - established patterns for assembling
complex symbolic structures out of simpler ones.
To the extent that a target structure accords with the conventional
units in the grammar, these units are said to sanction this usage.
Sanction is a matter of degree and speaker judgment.
Rather than the term "grammaticality", I will refer to an expression's
degree of "conventionality". 66
FULL SANCTION: Single unit: If a speaker requires a term for the
general category of 3-sided polygon, the lexical unit triangle most
likely satisfies and is an optimal solution.
Typically however, she may be unable to find a sym.unit that meets the
full, detailed specification needed in the desired target structure,
even in what would be considered straightforward language use.
Linguistic expressions almost invariably underspecify the
conceptualizations they encode [LAKOFF 77: Linguistic Gestalts].
e.g. "This is a triangle" - has as a concept a particular triangle, of
given edge-lengths/angles, drawn on paper etc, which the
decontextualized ling expression cannot code.
CONVENTION:
units - in square bracket or rectangle. non-units in parenth or round-box
semantic unit - upper case ; phonological unit - lower case
mapping from sem to phon - indicated by "slash": [ [TRIANGLE]/[triangle] ]
(here the outer square parenth indicates the unit status of the whole)
[ANALOGY: Perception Gestalt - on a corridor - big man and small man
at diff distances vs same distance - perceived in context as a
gestalt.
IDEA**: However, decontextualized expr has some constraints on what the sem
can be - can't be a square dog, say. Sim the image of a man remains that of a
man, only size is changing.]
SCHEMATICITY : category hierarchy: solid arrow - prototypical;
dashed arrow - non-prototypical (TOMATO in class FRUIT)
[usually "schematicity" ==> degree of categorization in grammatical concepts]
CODING: symbolic unit (sanctioning structure) is coded into a target
structure:
[ [TRIANGLE]/[triangle]] ==> ((TRIANGLE')/(triangle'))
the target is semantically more detailed (TRIANGLE') and does not have unit
status. Similarly the vocalization is more specific and not an unit.
[NOTE: units are not instances but generalizations]
PARTIAL SANCTION: frequently, conflict between specification of sanctioning
and target structures ==> strain. E.g. calling a cone with a writing tip as
"pencil":
[ [PENCIL]/[pencil] ] ---> ((PENCIL'}/(pencil'))
is a non-prototypical coding, an EXTENSION.
In figurative usage, e.g. ostrich for a person who ignores reality -
[ [OSTRICH]/[ostrich]] ---> ((PERSON'}/(ostrich'))
the semantic structures are incompatible in the great majority of their
specifications - this is why we call the usage "figurative". However, even
such usage can achieve unit status if used often. 71
So partial/full sanction distinction may be theoretically interesting but hard
to exercise in practice. 72
Full sanction - Target Structure TS is more specific than Sanc Str SS, i.e. TS
is an instantiation of SS - and "swallows it up" [p.93]
[IDEA**: What aspect of the semantics is transferred? pig, lion, fox, tiger,
rat, dog - all used for people. What characteristic is it that is
chosen for each of these? See discussion on ELABORATION re: "tube"
for subway p.72]
CREATIVITY: two kinds
- rule-governed creation of novel expressions by correct
application of grammatical rules.
- figurative language etc - using lexical items in new situations -
involving willful violation of grammatical rules 72
Grammar as gradation => not much distinction between these two meanings.
E.g. child does not like pie because it is too "apricoty". Usage refers to a
schema [ [N/...]-[Y/y]] is an unit, which coexists with its instances salty,
nutty, spicy, peperminty - but the mapping does not have unit status:
([ [N/...]-[Y/y]] ==> ([APRICOT/apricot]-[Y/y]))
i.e. the rule generating apricoty is not entrenched as an unit in the grammar.
Grammar = vast inventory of units structured in hierarchies that overlap and
intersect on a massive scale. Three basic kinds of relations between its
components:
- Symbolization [ [ DOG]/[dog]]
- Categorization ==> schematicity FRUIT==>APPLE, FRUIT ---> TOMATO
Schematic network - the hierarchy in grammar
- syntagmatic combination
SYNTAGMATIC is combination, (horiz) [unlike schematic (category, vertical)],
e.g. adding the plural morpheme
[ [DOG]/[dog]]-[ [PL]/[z]]
The basis for syntagmatic combination is INTEGRATION, which may involve
semantic adjustments (ACCOMMODATION) - e.g. "run" as applied to humans is
to be modified when discussing a horse, say.
[ SYNTAGMA, SYNTAGM -- (a syntactic string of words that forms a part of some
larger syntactic unit); SYNTAGMATIC -- (related as members of a syntagma;
"syntagmatic word associations")
NOTE: use of word "schematic" to indicate something like "superordinate"
category, relates to use of schematic as "simplified"]
]
2.2 GRAMMATICAL STRUCTURE
Cognitive Grammar posits just three basic types of structures: semantic,
phonological, and the symbolic structure that combines the two, consisting of
a semantic pole and a phonological pole, and the association between them.
semantic space - field of conceptual potential within which thought and
conceptualization unfold. A semantic structure is a location or
configuration in semantic space.
phonological space - field of phonic potential - often thought about in
spatial terms - e.g. vowel chart showing high-low and front-back
parameters is an attempt to map out one domain in phonological
space. Similarly the speech pictogram, plotting energy distn along
time and frequency - and a sound can be defined as a configuration
(e.g. a particular formant structure)
conceptualization unfold. A semantic structure is a location or
Mapping out the various domains of semantic space and their
interrelationships is clearly a prerequisite to any kind of definitive
semantic analysis.
CODING: sanctioning structure - > target structures
____________________________________________________________________
| |
| Semantic Space |
| |
| ------------------------------ --------------------------| |
| |Grammar (Ling. Convention) | | Usage Event | |
| | | | | |
| | ---------------------| | | | |
| | |Symbolic Unit | | | | |
| | | | | | ________________ | |
| | | --------------| | | | | | | |
| | | | Semantic | | | | | Conceptual- | | |
| | | | Unit |----- coding --------| ization | | |
| | | | | | | | | | | |
| | | |_____________| | | | |______________| | |
| | | | | | | | | |
| | | | | | | | | |
|---|--|---------|----------|----|-----|--------------|----------|--|
| | | _______|_______ | | | _______|_______ | |
| | | | | | | | | | | |
| | | | Phonological| | | | | | | |
| | | | unit |----- coding ---------| Vocalization| | |
| | | |_____________| | | | | | | |
| | |____________________| | | |_____________| | |
| | | | | |
| | Sanctioning | |_________________________| |
| | Structure | |
| |____________________________| Target |
| Structure |
| Phonological Space |
|___________________________________________________________________|
Even the articulatory facets of speech sounds are properly regarded as
conceptual - e.g. the segment [i] - [speakers can] actually hear the sound,
or they can simply imagine hearing it, i.e. they can activate an auditory
image of it (as in silent verbal thought). [Analogously for articulation] a
speaker can implement the articulatory routine and produce the sound, or he
can simply imagine implementing it, i.e. mentally run through the motor
routine without [translation] into muscular gestures. [78]
So, phonological space is part of semantic space.
_________________________________ _______________________________
| | | |
| Semantic Space | | Semantic Space |
| _______ | | |
| | DOG | | | |
| |_____| | | |
| | | | |
| __________________|___ | | ______________________ |
| |Phonological ___|___| | | |Phonological ______ | |
| | Space | dog || | | | Space |NOISE| | |
| | |_____|| | | | |_____| | |
| |_____________________| | | |_____________________| |
|________________________________| |______________________________|
Onomatopeic words like "clang" may have the semantic unit also inside the
phonological space. Here a correspondence can be measured directly in
terms of the phonology of the unit [clang] and the sound it represents. [90]
Some other situations like "The boy went [NOISE]" where
NOISE may be a sound like "ga ga ga" - which is used directly for itself - it
is its own self-symbolizing map - it also resides in the ph space.
GRAMMAR AS SYMBOLIZATION [2.2.2. p.81]
Grammatical Morphemes: later chapters
Grammatical Classes: e.g. class of nouns: [ [THING]/[...]] ==> [ [TREE]/[tree]]
involves [ [THING]==>[TREE]] at semantic pole, and ... ==> tree (phon)
Grammatical Constructions: Syntagmatic combination of morphemes and larger
expressions - symbollically complex, involving a grammatical
schmema which incorporates syntagmatic integration inside it:
[ [THING]/[...]]-[ [PL]/[z]]] ==> [ [ [PIN]/[pin]]-[ [PL]/[z]]]
Integration involves a complicated set of relationships. Syntagmatic
combination opereates on both semantic and phonological poles (BIPOLAR):
- phonological structures - [pin-z] - i.e. [z] is attached to the
outermost consonant - and also instantiated differently in vowel-endings from
consonant-endings, e.g. "boys" or "peas" vs "pins" or "tables".
- semantics: [PIN-PL] - [PL] designates an indefinite num of replications (in
type) of a discrete entity.
- SYMBOLIC : the sem integration corresponds to the phon integration
[NOTE: The resulting object may be a unit on its own [PINS] if it is suff
conventionalized, or a non-unit (e.g. (UNITS)). For some plurals, the
resulting object involves restrictions - e.g. [EYES] or [LEGS] - often seen as
pair]
[Also - not fully productive; e.g. oxen in forward mapping; "trousers",
"scissors" in reverse]
Similarly in [TALL-BOY] the ph. integration involving the linear ordering of
[tall] and [boy] symbolizes the sem integration - [TALL] pertains to the
feature "height" in [BOY]. The integration here involves fixing the norm in
the scale on which [TALL] will be measured in terms of the norm for
height in [BOY].
[IDEA: Quine Observational Predication: learn from repeated adjectival usage -
black dog, tall boy - that the structure mod-N has this behaviour]
syntax cannot be "autonomous" - if the semantic pole is suppressed, there is
no basis for recognizing morphemes (or larger lexical units).
[i.e. morphemes are the smallest "meaning-carrying" elements, phonemes also]
2.3 COMPONENTIALITY [86]
Componentiality must be Cognitively plausible: e.g. [CIRCLE] - can be defined
in terms of [RADIUS] and [CENTER] but this does not seem cogn plausible -
children learn [CIRCLE] as an entrenched object without focusing attention on
the distance from some particular distinguished point. [CIRCLE] is more
likely learnt as a shape gestalt: [RADIUS] is then learned as a secondary
concept. [IDEA: children's drawing of circles / closure - progression over ages]
NO PRIMITIVES: Cog Grammar has no notion of "primitives" - indivisible further
- not of interest. Units are simply well learned, and require no constructive
effort. Internally however, they can be complex. Semantic units in any event
are defined relative to extensive and complex knowledge structures.
NONREDUCTIVE: the properties of any complex may be more than its
parts. (e.g. stapler - the concept is more specific than something that
staples - ask someone to draw a "stapler". ) 87
HIERARCHY INVERSION: to define HAND, we need ARM; thus ARM is a conceptual
component in HAND, and HAND in FINGER, not the other way around.
UNIPOLAR vs SYMBOLIC (BIPOLAR) components : [poor naming?]
Phonological example:
[tables] = [ [tey]-[blz]] ("l" is the syllabic nucleus)
==> is unipolar, since no semantic connotation.
[tables] = [ [tey]-[bl]]-[ [z]] corresponds to the [TABLE-PL] semantic
compound and is bipolar. Here the integration must specify how [z] is
added to the end of the last syllable, forming [blz]
Semantic example: Peas and Corn - both refer to small discrete
objects replicated to form a mass.
Thus at one level, the (unipolar) components of these concepts are
a) replication and b) small discrete objects.
But symbolically, peas is lexicalized as an individual pea, corn at the level
of the mass, unanalyzable further. Thus the "small discrete object" in corn
is SUBLEXICAL (needs periphrastic "kernel of corn"). Hence
"peas are" vs "corn is", "many peas" vs "much corn" etc. 89
[periphrastic - roundabout -
A periphrastic study in a worn-out poetical fashion,
Leaving one still with the intolerable wrestle
With words and meanings. - T.S.Eliot ]
FIGURATIVE LANGUAGE: Example: CAT OUT OF BAG
The entire string has unit status, as has each of its components. The literal
sense [ [ [CAT] [OUT OF] [BAG]] / [cat out of bag] ]] is the sanctioning
structure that sanctions the figurative [ [ [INFORMATION] [OUT OF']
[CONCEALMENT]] / [cat out of bag]]]. The sem/ph parts relates separately, and
the whole also. The ph parts are identical "<--->", but the sem part is a
very partial sanction "--->" as is the overall symbol. In the sem
sanction, [CAT]==> [INFORMATION} is mysterious ("the conflict is blatant"),
but [BAG] has something of [CONCEALMENT] in it, and [OUT OF] undergoes a minor
shift - from spatial to abstract domain.
Chapter 3 COGNITIVE ABILITIES [99]
Semantics is defined in terms of mental processes. Events become entrenched
through repetition.
[INTENTIONAL repetition?]
Similarity may involve comparison of parts, or selections of domains (e.g. in
music, sounds evaluated based on pitch, but also quality - e.g. can separate
out the trumpet in an orchestra).
Short-term memory as a trace before its decay - needed for comparing events
across times. Vectors for characterizing semantics. 104
RECOGNITION: proclivity for interpreting new experience in terms of previous -
difficult not to recognize a familiar object or not to read text, or not to
realize that "........" constitutes a line. example: CLOSURE phenomenon -
where degraded input is nevertheless recognized and categorized.
IMAGERY
The term has been used for figurative language, but also for mental imagery.
Here it has a somewhat diff meaning - a more detailed depiction that differs
in terms of salience of subparts, figure-ground, level of abstractness,
or other aspects.
e.g. All these may describe the same situation, but involve diff imagery:
1a The clock is on the table.
b The clock is lying on the table
"lie" ==> alignment of clock to horiz plane of table.
c The clock is resting on the table.
"rest" - static-ness
d The table is supporting the clock. 110
figure-ground shift(?)
[IDEA: doesn't this imply that the clock is a heavy one]
imagery may be peripheral or autonomous [poor choice of terms - sensorimotor =
peripheral, autonomous is internal, indep of senses]. Imaginary worlds have
as much relevance. Aspects of attention ( subsection 3.2.3, but then section
3.3 deals with "Focal Adjustments" - i.e. linguistic focus) 111-115
LINGUISTIC FOCUS [POLYSEMY / NEAR-SYNONYMY]
- selection - predicates selects particular domains
2 big blue plastic cup: DOMAINS: size / colour / material
3a close to the garage / 4a near the garage : space
b close to christmas / near Christmas : time
c This paint is close to the blue we want / ?near the blue we want : colour
d Steve is close to his sister / *near his sister : emotional
("emotive")
each invocation profiles different aspects of its conceptual space
- the synonym near does not conventionalize as well for the
profiles for c and d
Some predicates may be restricted in SCALE, e.g. "minute"
5a galaxies close to one another / ??minute galaxy
b San Jose close to Berkeley / ?minute nation
c The runner is staying close to first base / minute diamond
d The sulfur and oxygen atoms are quite close in this molecule /
minute molecule
Nominal predicates may differ in scale - cove/bay, town/city 118
PROFILE - entity designated by a predication - maximally prominent, a
focal point.
BASE (= scope) - context needed for characterizing profile
(Profile: projection against a base)
A linguistic expression intrinsically evokes a knowledge structure,
some facet of which is "profiled"
thumb - profiled against concept of a human hand
Base cannot be too distant, as in these "have" constructs (Bever/Rosenbaum
70, Cruse 79):
6-7a A body has two arms / ?two elbows
b An arm has an elbow and a hand / ?five fingers
c A hand has five fingers / ??Arm has five fingernails and 14 knuckles
d A finger has three knuckles and a fingernail / body has 28 knuckles
- perspective - position from which scene is viewed
- figure = part of scene that is perceptually or attentively
salient - e.g. moving object is salient / compactness or
integralness of a region - linguistically: figure may be a
"new" object, against old. But can get over-ridden.
- viewpoint = perceptual view - objects are presented in canonical views,
(e.g. house viewed from ground, not from above), unless
vantage point is otherwise defined as in 9a
9a I was about to take a picture of the house when a blimp landed right
in front of it [may also be rear of house]
c A blimp landed in the street in front of the house [canonical front]
[DeLancey 81] - effect of viewpoint on aspect, voice, and grammatical
relations - ASPECTual relations - imperfect event (unfinished)
from Point of view of speaker - event (finished)
voice: also refers to the p.o.v
- deixis - ground = speech event+ participants+ setting
deictic - many expressions are perceptually grounded;
"epistemic predication" - deictically grounded
- tense : most FINITE verbs are deictic (inflected for
tense or person)
- pronouns / nouns - "the cat"
Sometimes viewpoint may shift to a viewpoint diff from the speaker
11b (Mother to child:) "Don't lie to your mother."
12a That's me in the top row (said while looking at a photo)
- subjectivity vs objectivity :
- abstraction - level of specificity:
fig 3.7 : tall ==> over six feet tall ==> about six foot five inches tall
==> exactly six feet five and a half-inch tall
thing - animal - mammal - rodent - squirrel - ground squirrel
move - locomote - run - sprint
Challenges to image models for non-basic prototypes [Kempson 77] - how do
we form an image of a dog encompassing both alsatian and poodle?
[NOTE: may be linguistically motivated and not pre-conceptual;
This section can be enriched considerably, seems theoretically weak]
bookexcerptise is maintained by a small group of editors. get in touch with us! bookexcerptise [at] gmail [dot] .com. This review by Amit Mukerjee was last updated on : 2015 Mar 18