The Oxford Handbook of Cognitive Linguistics Part 15 ppsx

two ships sailing from San Francisco to Boston a century and a half apart are
blended into a race between one ship and the ghost of the other.
35
Figure 4.14c gives
an example of an inadvertent phrasal blend, and figure 4.14d (adapted from
Kemmer 2003, which gives an excellent discussion of schemas as tools for analyzing
lexical blends) represents the graphicophonological pole of a purposeful lexical
blend. In each case, the blend clearly fits the pattern (schema) of 4.14a, which is
itself composed of schemas.
Sweetser (1999) and others (e.g., Fauconnier 1999) have stressed that the mecha-
nisms of blending must often be invoked for the analysis even of such everyday
grammatical structures as Adjective-Noun constructions. Having these structures
already analyzed in terms of schematicity relationships among components and
composite structure makes this sort of proposal much more natural and obviously
right than it would be under other theoretical models.
It is probably feasible to claim that all cases of blends consist of appropriately
configured arrays of schematic and partially schematic relationships among cog-
nitive structures, elaborating or differing in various ways from the prototype char-
acterized in figure 4.14a. Such a claim does not, of course, obviate the necessity of
specifying more fully what kinds of correspondences (figure 4.1c) are involved in
the partially schematic mappings which are so important to the blending or of ex-
plicating what kinds of emergent structures show up in the blended spaces and how
they do so. But at the very least, it seems clear that schematicity relationships are
crucially involved in the mechanism of blending.
Figure 4.14. Blends
110 david tuggy
5. Summary

The foregoing discussion is far from exhaustive: there are other ways schematicity
relations function in language and many other subtleties in the functions I have
discussed. It should be clear, however, that:

a. relationships of schematicity are pervasive in language;
b. recognition of them is crucial, at least under the Cognitive Grammar
model, to understanding many of the most central kinds of structures
which constitute the grammars of languages;
c. in particular, a number of other seminal and widely utilized concepts
within Cognitive Linguistics, such as image schemas (this volume, chapter
9), constructions (this volume, chapter 18), blends (this volume, chapter
15), metaphor (this volume, chapter 8), and (perhaps) frames, ICMs, do-
mains, and mental spaces (this volume, chapters 7, 14), are usefully seen
as particular kinds (subcases) or arrangements of schemas;
d. by recognizing schematicity in these different areas, the Cognitive Gram-
mar model achieves significant conceptual unification and appropriate
simplification of the theoretical machinery;
e. our understanding of certain long-standing problems for linguistic analysis
is considerably aided by adopting this perspective.
NOTES

1. The etymologically correct plural schemata is also used. I here follow the usage of
Langacker (1987a, 1991) and Lakoff (1987) in preferring schemas.
2. Dictionary definitions of the term are close to the Langackerian meaning we will use
in this chapter; e.g., ‘‘a summarized or diagrammatic representation of something, an
outline’’ (The American Heritage Dictionary of the English Language, 1978). The term’s use
in Cognitive Linguistics traces back, at least in part, to Rumelhart’s (1975) work with
computational schemas; see also this volume, chapter 9, section 2 for a fuller discussion
and references.
3. The other relationships mentioned in this regard are component-composite rela-
tionships, symbolization relationships between phonological and semantic structures,
and syntagmatic relationships between co-occurring forms. Of these, the component-
composite and syntagmatic relationships in their turn depend heavily on schematicity for
their characterization (sections 4.9, 4.10, and 4.11; Langacker 1987a: 73–75).

4. Although this point is relevant to the question of whether there exist linguistic
‘‘primitives’’ or ‘‘atoms’’ (e.g., Wierzbicka 1996), it is not exactly the same issue. For
instance, Wierzbickan-style primitives, while they are to be understood as being both
cognitively and linguistically universal, make no pretense of being cognitively, but only
linguistically, atomic. (They have a lot in common with Lakoffian ‘‘basic-level’’ and
‘‘image-schematic’’ concepts; see section 2.3.) A true conceptual (cognitive) atom would
schematicity 111
probably be something like a single neuron firing or (if we admit a slightly higher
neurological level) a message to contract a single muscle or the perception that a particular
single point of the skin has been touched. But it is improbable that such cognitive
structures are ever the meanings of any linguistic structures. Rather, it is much higher-level
patterns of such cognitive events that we are conscious of and use in our communications.
Such patterns are, by definition, schemas.
5. Although Langacker used italics instead of small caps, he clearly is talking about the
relationships between the concepts, the meanings which constitute the semantic poles of
the lexical items in question. I am following the tradition of representing such purely
semantic constructs in small caps.
6. It does not follow that there can be no difference between a mediated schematic
relationship A ? B ? C and a direct one A ? C. For instance, in figure 4.9a, an arrow is
represented directly from Small.Item-collector to butterfly-collector besides a relationship
mediated through insect-collector. This reflects my judgment that whether or not a
speaker activates the Insect-collector schema on a given occasion (or even has such
a schema), butterfly-collector is likely to be coactivated and compared directly with
Small.Item-collector and to be strongly sanctioned by it.
7. In particular, unless there is some special factor at work, casual comparisons which
yield few or no similarities are highly unlikely to ever become cognitively entrenched in the
first place, much less conventionalized among a group of speakers.
8. In Figure 4.1, A might be dog and B hyena; C would be a schema which we might
call dog-like carnivore, which would tend to become established by the mental activity
of construing hyenas as a (deviant) kind of dogs.

9. Figure 4.1c does not indicate, as 4.1a and 4.1b do, the direction of comparison; i.e., it
represents a comparison A
3
B as much as A
"
B.
10. In figure 4.2a, A and B might be man and woman,Cfeatherless biped,D
chicken, and E dog. Similarly in 4.2b, C might be biped and C’ featherless thing,
while the other identifications remain constant.
11. Figure 4.3 might represent meanings of the English word baby as follows:
P ¼ human infant,a¼ newborn/very young animal,b¼ youngest of a set of
siblings,c¼ girlfriend,d¼ girl or young woman addressed with familiarity,
e ¼ pet project,f¼ cherished object (e.g., car), g ¼ of smaller than normal size,
h ¼ male infant,i¼ female infant,j¼ puppy,k¼ chick,S¼ object of interest/
affection, S
1
¼ newborn/very young animate being,S
2
¼ human object of
tenderness/affection,S
3
¼ animate object of tenderness/affection,S
4
¼ (near)
youngest member of family,S
5
¼ inanimate object of interest/care,S
6
¼ (thing)
of smaller than normal size. Note that even S is not schematic for all the concepts, as it

does not include S
4
and b, nor S
6
and g, nor d.
12. Pinker (1994: 106), for one, holds that ‘‘a noun, for instance, is simply a word that
does nouny things; it is the kind of word that comes after an article, can have an ’s stuck
onto it, and so on.’’ The major problem with this statement for a cognitive grammarian is
the word ‘‘simply.’’ Pinker’s summary statement is that ‘‘a part of speech, then, is not a
kind of meaning; it is a kind of token that obeys certain formal rules.’’ My argument here is
that obeying such rules should be counted as a kind of meaning, but I also follow Lan-
gacker in contending, below, that there is other semantic material in the overall schema for
the category, and certainly in its prototypical subschemas.
13. Many cite Saussure’s ([1916] 1996) seminal notion of ‘‘oppositions’’ in this regard: a
category is defined not by what it includes but by what it contrasts with, and thus excludes.
Some substantive characterizations are also selective to the point of near vacuity. For
112 david tuggy
instance, the often-cited choice of bipedality and featherlessness as the criterial features for
defining humanity involves ignoring many substantive qualities of humans which are
intuitively more central, such as cognitive and particularly linguistic abilities, manual
dexterity and technological skill, facial appearance and general bodily shape, complex
social behaviors (again including language), and so on.
14. A type of phonological class which deserves special mention is that of phonemes.
They are usefully modeled as near-classical categories in which less salient elaborate
structures (allophones) are largely subsumed under highly prototypical schemas with little
overlap. Traditionally problematic issues, such as aberrant allophones, contextual neu-
tralizations, and ‘‘archiphonemes,’’ can be naturally and insightfully modeled in schematic
hierarchies including such structures. Similarly, phonological features can be modeled as
schemas, and their behavior, including those aspects that have been problematic for other
theories, fits the model well (Langacker 1987a: 388–94; this volume, chapter 17, section 5;

Nathan, this volume, chapter 23).
15. Note that to be true schemas for the lexical item Hilary Rodham Clinton, these
structures must be bipolar symbolic structures, with a signifie
´
/signifiant structure; that
is, a complete representation would have something like [Hilary Rodham Clinton /
'hil@ri rA:d@mklintn
"
] / [woman’s name / X] / [name / X] / [thing / X]. We will use
bolded lettering with initial capitals (e.g., Woman’s Name) to indicate structures of
this sort: too schematic to be lexical, but neither solely semantic nor solely phono-
logical.
16. thing may be thought of as the meaning of thing in contexts such as anything
at all.
17. The differences between the schemas thing, relation, and process, in Lan-
gacker’s view, are thus matters of construal (see this volume, chapter 3 ) rather than
necessarily of identity of the entities referred to. The verb or adjective parallel (both
relations, one processual and the other not) and the noun parallel(s) (a thing) can thus be
used of the same pair of lines; the differences are not differences of truth values or of what
situation is referred to, but are, nonetheless, semantic distinctions. Nominal/verbal pairs
denoting events or other processes (e.g., the noun love and the verb love, or the noun
distribution and the verb distribute) are handled similarly; the differences in meaning
consist of different construals imposed on a set of interconnected entities, designating
either the set as a whole or the interconnections (evolving over time) which help constitute
it as a set.
18. As usual, it is difficult to discuss these matters without recourse to the ‘‘content
metaphor.’’ We could perhaps reword this statement to say ‘‘the structures involved are
linked to definite cognitive routines which constitute their semantic poles.’’
19. ‘‘Spell-out rules’’ would be an exception: Cognitive Grammar holds that the re-
lationship between a meaning and the phonological structure associated with it is not one

of schematicity but of a different, associative rather than comparative, linkage. Not co-
incidentally, spell-out rules are one case where what is called a ‘‘rule’’ is not a general-
ization, but rather an idiosyncratic fact about a single lexical item.
20. NP, VP, and the like are of course shorthand for more substantive characteriza-
tions of the sort required under Cognitive Grammar, with the definition of an NP centered
on that of an N (i.e., it will have the schema thing as its semantic pole) and that of
VP centered on that of a V (i.e., the schema process as its semantic pole). Similar
substantive characterizations would be necessary for Aux, the uninflected verb represented
as be, the past-participial inflection represented as -en, the Determiner, the Locative
element schema, and so forth.
schematicity 113
21. In figure 4.6 and later diagrams, we follow the convention of using boxes with
dotted-dashed lines and rounded corners to indicate structures which are novel or near-
novel, that is, not yet established in their own right even though, as in this case, they may be
grammatical in the sense of being sanctioned—see section 4.5—by established schemas.
22. In figure 4.8, and not elsewhere in this chapter, an attempt has been made to
render the parameter of cognitive distance by physical distance between represented en-
tities: thus, in 4.8a the schema is much closer to its elaborations and they to each other than
in 4.8e.
23. The parenthesized perhaps is meant to indicate that there may well be a good many
speakers for whom the structures in question are not well established in their own right. To
the extent that any of them is established, it contributes its bit of legitimacy to beetle-
collector; if it is not, the lack of its sanction does not mean the novel structure is therefore ill
formed. The sanction from bug-collector is very nearly direct, since beetles are a proto-
typical kind of bugs; it is only the phonological specifications of bug and beetle that
conflict.
24. Presumably, this involved analogy with words like water-jet or air-jet. An analysis
similar to the one given in the text below would apply to the case of inkjet as well. The
sanction received from such high-techy words as ram jet, turbojet, Lear Jet, etc., or from
schemas derived from them, will not be further mentioned but is certainly a real factor in

the discussion that follows.
25. DeskJet, LaserJet, CopyJet, OfficeJet, PaintJet, QuietJet, ScanJet, and ThinkJet are all
trademarks of the Hewlett-Packard Company. Deskjet, and to a lesser extent laserjet, seem
to have achieved the marketing nirvana of being common nouns for the type of product as
well as specific names for the brand.
26. The historical order of these coinages is an interesting but nondeterminative
question. If the (historical) order was different from the one presented here, this order can
be taken as representing the experience of a hearer like me who first learned the terms in
the order given.
27. I will not pursue further the issue of whether a schema can consist of a disjunctive
‘‘either-or’’ structure or the closely related question of whether a list of alternatives may
function in certain ways as a schema would. I have argued elsewhere (Tuggy 1992: 254 –55)
that the answer to the second question, in certain instances at least, is yes.
28. ‘‘In practice, we are more likely to call a semantic structure a domain if there are a
substantial number of concepts profiled relative to that structure. The term ‘domain’
implies a degree of cognitive independence not found in a dimension’’ (Croft 1993: 340).
29. Lakoff and Turner (1989: 103) appear to use the words ‘‘domain’’ and ‘‘schema’’
interchangeably in discussing metaphor and metonymy: ‘‘In metaphor there are two
conceptual domains, and one is understood in terms of the other. Metonymy involves
only one conceptual domain. A metonymic mapping occurs within a single domain,
not across domains via metonymy one can refer to one entity in a schema by referring
to another entity in the same schema one entity is taken as standing for one other
entity in the same schema, or for the schema as a whole.’’
30. Since John Wayne is a human being, there is in the encyclopedic semantic
structure attached to his name a strong, though not particularly salient, expectation that he
eats food (and engages in other typical human activities). The toast, in contrast, contains
a clear and salient expectation that the designatum was produced in order to be eaten.
These specifications function as e-sites to which ate corresponds. However, (i) neither
is as salient within or central to John Wayne and the toast as ate’s e-sites are to it, and
(ii) there is little elaborative distance between them and ate. Thus, the dependence of the

114 david tuggy
noun phrases on the verb is much less than the verb’s dependence on them. These rela-
tively subtle points are not represented in figure 4.12.
31. Although it is typical for standard transitive clauses in English to combine the
verb with its object, forming a verbal phrase constituent, before combining that phrase
with the subject, it is not necessary under Cognitive Grammar (and certainly not defini-
tional for subjecthood vs. objecthood, as in some other theories). Combination of the
verb with its subject first, as implied in figures 4.12b and 4.12c, will produce the same
composite structure and will in fact be favored in some syntactic environments. This
particular constituency is chosen here in the interest of expository clarity.
32. Where there is a great disparity between a highly schematic profile determinant
(typically an affix) and a highly elaborate nonprofile determinant (stem), linguists tend to
dispute whether headship should be accorded to the lightweight affix or to the semantically
heavier stem. Thus, it may be mooted whether assign or ment is the ‘‘head’’ of as-
signment. Under the Cognitive Grammar conception, profile determinance is a central
kind of semantic weight, and the prototypical head is both profile determinant and se-
mantically heavier than its syntagmatic partners. It becomes a matter of definition which is
‘‘head’’ when the profile determinance and overall semantic weight do not line up.
33. If transitivity is a need for further specification of the nature of the object (land-
mark) of a process, the elaborative link from the food specification to the toast ful-
fills that need, and further elaboration is likely to be unneeded, perhaps even problematic.
Yet another language, or even certain dialects of English, might well allow or require an
object nonetheless, giving something like the toast John Wayne ate it.
34. Most speakers will agree that English and horn are components of English horn,
and that eaves and drop are components of eavesdrop; that is, the participation of these
words in the construction is clear even though the nature of their participation is not. In
other cases, the participation itself is not clear: for instance, few speakers think of halter as
saliently composed of halt and -er, and fewer still would recognize the morphemes rue
and -th in the ruth of ruthless. Space precludes full discussion of such cases here, but
analyzing compositionality in terms of schematicity relationships automatically provides

for such variations in analyzability, allowing them to fit with perfect ease within the gamut
of constructional types (Langacker 1987a: 457–66).
35. The solidness of the boxes around the two input spaces and the generic space in
figure 4.14b is accurate only in the particular context in which the blend arose, where
those concepts were established in the minds of the author and most readers. They are not
widely established structures of English.
REFERENCES

Adams, Marilyn Jager, and Allan Collins. 1979. A schema-theoretic view of reading. In Roy
O. Freedle, ed., New directions in discourse processing 1–22. Norwood, NJ: Ablex.
Aitchison, Jean. 1990. Words in the mind: An introduction to the mental lexicon. Oxford:
Basil Blackwell. (2nd ed., 1994; 3rd ed., 2003)
Chafe, Wallace L. 1987. Cognitive constraints on information flow. In Russell S. Tomlin,
ed., Coherence and grounding in discourse 21–51. Amsterdam: John Benjamins.
Chomsky, Noam. 1957. Syntactic structures. The Hague: Mouton.
Coulson, Seana, and Todd Oakley. 2000. Blending basics. Cognitive Linguistics 11: 175–96.
schematicity 115
Croft, William. 1993. The role of domains in the interpretation of metaphors and me-
tonymies. Cognitive Linguistics 4: 335–70.
Croft, William. 2001. Radical construction grammar: Syntactic theory in typological per-
spective. Oxford: Oxford University Press.
Fauconnier, Gilles. 1997. Mappings in thought and language. Cambridge: Cambridge
University Press.
Fauconnier, Gilles. 1999. Methods and generalizations. In Theo Janssen and Gisela Re-
deker, eds., Cognitive linguistics: Foundations, scope, and methodology 95–127. Berlin:
Mouton de Gruyter.
Fillmore, Charles J. 1975. An alternative to checklist theories of meaning. Berkeley Lin-
guistics Society 1: 123–31.
Kemmer, Suzanne. 2003. Schemas and lexical blends. In Hubert Cuyckens, Thomas Berg,
Rene

´
Dirven, and Klaus-Uwe Panther, eds., Motivation in language: Studies in honor
of G
€
uunter Radden 69–97. Amsterdam: John Benjamins.
Lakoff, George. 1987. Women, fire, and dangerous things: What categories reveal about the
mind. Chicago: University of Chicago Press.
Lakoff, George, and Mark Turner. 1989. More than cool reason: A field guide to poetic
metaphor. Chicago: University of Chicago Press.
Langacker, Ronald W. 1987a. Foundations of cognitive grammar. Vol. 1, Theoretical pre-
requisites. Stanford, CA: Stanford University Press.
Langacker, Ronald W. 1987b. Nouns and verbs. Language 63: 53–94.
Langacker, Ronald W. 1991. Foundations of cognitive grammar. Vol. 2, Descriptive appli-
cation. Stanford, CA: Stanford University Press.
Pinker, Steven. 1994. The language instinct: How the mind creates language. New York:
William Morrow.
Rumelhart, David. 1975. Notes on a schema for stories. In Daniel G. Bobrow and Allan M.
Collins, eds., Representation and understanding: Studies in cognitive science 211–36. New
York: Academic Press.
Saussure, Ferdinand de. [1916] 1996. Cours de linguistique ge
´
ne
´
rale. Ed. Eisuke Komatsu.
Trans. George Wolf. Oxford: Pergamon.
Schank, Roger C., and Robert P. Abelson. 1977. Scripts, plans, goals and understanding.
Hillsdale, NJ: Laurence Erlbaum.
Sweetser, Eve. 1999. Compositionality and blending: Semantic composition in a cognitively
realistic framework. In Theo Janssen and Gisela Redeker, eds., Cognitive linguistics:
Foundations, scope, and methodology 129–62. Berlin: Mouton de Gruyter.

Taylor, John R. 1995. Linguistic categorization: Prototypes in linguistic theory. 2nd ed.
Oxford: Clarendon Press. (3rd ed., 2003)
Tuggy, David. 1992. The affix-stem distinction: A cognitive grammar analysis of data from
Orizaba Nahuatl. Cognitive Linguistics 3: 237–300.
Tuggy, David. 1993. Ambiguity, polysemy, and vagueness. Cognitive Linguistics 4: 273–90.
Wierzbicka, Anna. 1996. Semantics: Primes and universals. Oxford: Oxford University
Press.
116 david tuggy
chapter 5

ENTRENCHMENT,
SALIENCE, AND
BASIC LEVELS

hans-jo
¨
rg schmid
1. Introduction

One of the basic tenets of Cognitive Linguistics is that the human capacity to
process language is closely linked with, perhaps even determined by, other fun-
damental cognitive abilities. This chapter is concerned with possible manifestations
of such abilities—most notably among them perception, memory, and attention
allocation—in linguistic competence and use. It deals with mechanisms that in-
fluence the storage of concepts and constructions in long-term memory and with
factors involved in the retrieval and activation of concepts and constructions from
memory during ongoing language processing.
This chapter falls into seven sections. Following this introduction, section 2
illustrates the use of the notions of entrenchment and salience in Cognitive Lin-
guistics and provides initial definitions. Section 3 deals with the role of entrench-

ment in the emergence, sanctioning, and blocking of linguistic units. More specific
linguistic effects of entrenchment and salience in the lexicon are discussed in sec-
tion 4. Section 5 reviews an attempt to measure the relative entrenchment of cate-
gories in lexical taxonomies. Section 6 deals with effects of entrenchment and sa-
lience in the area of syntax, and section 7 offers an outlook on future research in
this area.
2. The Notions of Entrenchment
and Salience in
Cognitive Linguistics

2.1. Entrenchment
When speakers encode their conceptualizations in words and sentences, they uti-
lize their competence, that is, the linguistic knowledge of phonological, semantic,
grammatical, and collocational properties of words and syntactic structures. This
knowledge is stored in their long-term memory. It is fairly unlikely, however, that
speech processing is always carried out in a creative, generative fashion in the sense
that language users always have to actively, or even consciously, search their mem-
ory for means of encoding what they have in mind or decoding what they hear or
read. Presumably, a lot of what speakers say is available in memory in some kind of
prepackaged, ready-made format. Convincing evidence for this claim are the words
of a language, since these represent nothing else than conceptualizations that have
been fossilized by convention in a speech community. We hardly ever stop to think
what language would be like without prepackaged concepts readily encodable by
words. To refer to a dog that we see running across a meadow, there is no need to
consciously construe an appropriate conceptual unit from scratch, because words
like dog or poodle are readily available. The question of how to name this entity will
not reach a level of conscious awareness, and the activation of concepts matching
our experience of the dog will hardly require cognitive effort. The reason is that
familiar concepts like ‘dog’ or ‘poodle’ are deeply entrenched in our memory so
that their activation has become a highly automated routine.

When we are faced with a more exotic animal, say a tapir in a zoo, the
situation will be different, because the cognitive processes relating the perceptual
input that determines the target conceptualization to the corresponding phono-
logical unit are less well entrenched. We are likely to need more time to identify
and categorize the animal by considering some of its most prominent attri-
butes before we can even begin to search our mental lexicon for a word matching
this cognitive category. Clearly, then, the conceptual unit ‘tapir’, which is rep-
resented by this cluster of attributes, is less well entrenched than the cognitive
unit ‘dog’.
Cognitive units come to be entrenched and their activation automated to the
extent that they have been used before. According to Langacker (1987: 59), there is a
continuous scale of entrenchment in cognitive organization. Every use of a
structure has a positive impact on its degree of entrenchment, whereas ex-
tended periods of disuse have a negative impact. With repeated use, a novel
structure becomes progressively entrenched, to the point of becoming a unit;
moreover, units are variably entrenched depending on the frequency of their
occurrence.
118 hans-jo
¨
rg schmid
Langacker conceives of entrenchment as being fostered by repetitions of cog-
nitive events, that is, by ‘‘cognitive occurrences of any degree of complexity, be it
the firing of a single neuron or a massive happening of intricate structure and large-
scale architecture’’ (1987: 100). As a result, the degree of entrenchment of a cognitive
or linguistic unit correlates with its frequency of use. Geeraerts, Grondelaers, and
Bakema (1994) argue for a more refined version of this idea (see section 5). On their
account, it is not frequency of use as such that determines entrenchment, but fre-
quency of use with regard to a specific meaning or function in comparison with al-
ternative expressions of that meaning or function.
Entrenchment of concepts or constructions not only depends on the frequency

of activation by individual speakers (and in that sense is not a completely private
matter), but it also applies to languages as such and whole speech communities,
because the frequency of occurrence of concepts or constructions in a speech com-
munity has an effect on the frequency with which its members are exposed to them.
The (tacit rather than explicit) implication is that this results in some kind of
collective automatization effect, which makes it possible to talk of the degree of
entrenchment of a concept or construction in a given language.
In short, the notion of entrenchment is thus used in Cognitive Linguistics—
and especially in Langacker’s influential framework of Cognitive Grammar (1987,
1991; this volume, chapter 17)—to refer to the degree to which the formation and
activation of a cognitive unit is routinized and automated.
2.2. Salience
The notion of salience is employed in Cognitive Linguistics in two closely related
ways, yet distinct enough to call for differentiation.
The first usage, called ‘‘cognitive salience,’’ concerns the activation of concepts
in actual speech events. Cognitive units must be activated when they are required
for speech processing, and this may result from either one of two mental processes:
the activation of a concept may be controlled by a conscious selection mechanism,
whereby the concept enters a person’s focus of attention and is being processed in
current working memory (Anderson 1983: 118–20; Deane 1992: 35); alternatively, a
concept may be activated through spreading activation, which occurs when the
activation of one concept (e.g., ‘dog’) facilitates the activation of others (e.g., ‘bark’,
‘tail wagging’, ‘fur’, ‘poodle’, ‘alsatian’, ‘collie’, etc.) (see Collins and Quillian 1969;
Collins and Loftus 1975; Anderson 1983: 86–125; and Deane 1992: 34). Irrespective of
how a cognitive unit has been activated, it is said to be salient if it has been loaded,
as it were, into current working memory and has thus become part of a person’s
center of attention. Since the use of concepts that are already activated requires
minimal cognitive effort, a high degree of cognitive salience correlates with ease of
activation and little or no processing cost. Currently inactive concepts, on the other
hand, are nonsalient.

entrenchment, salience, and basic levels 119