Before presenting the actual material on cognitive linguistics, I thought I'd present an analogy that illustrates how the whole general approach of cognitive linguistics compares to the time-honored structuralist approach to linguistics. The reason I want to do this is because, most persons who've studied enough linguistics to understand what phonology, morphology, syntax, and semantics are, and can define and illustrate a phoneme, a morpheme, a syntactical rule, a dictionary definition, would probably ask themselves “so just how different can cognitive linguistics be from this stuff?” I mean, isn't a phoneme a phoneme, no matter what school of linguistics you study under?? This is an excellent question to ask, so I thought I'd start by showing how cognitive linguistics is different because it pretty much serves a different purpose than to describe and define phonemes, morphemes, etc.

The time-honored or, to use a convenient generic term, “standard” approach to linguistics is the approach that divides up human language into component areas and studies all the “parts” of each component, then later relates these components to each other in an increasingly complex manner that begins to give an overall picture of how language works.

Here's the analogy: Suppose you know nothing about cars and have never seen one, but now must learn how they work. The old approach would be to start with the frame, the axles, the wheels, the transmission, the drive train, the parts of the engine, how the engine connects to the drive train, the body, the steering column, the interior, the controls, etc. and “build” the car part by part to see how the whole functions. A “cognitive” approach would be to first study what humans need cars for and what needs should be addressed in determining how a car should be built and function. We'd start by looking at what roads are for, why transportation is necessary, why internal combustion is the approach that has developed (as opposed to steam, electric, or atomic power), what the technological limitations are on practically addressing those needs, why human beings value cars, their attitudes toward transportation, etc., THEN start with a whole car (not its individual parts) and analyze the whole car as to how it addresses and manifests the various needs, values, and attitudes we identified. Later, we can start taking the car apart (rather than putting it together) to determine how the “whole” car developed not just from the amalgamation of its parts, but also the synergy of its parts (i.e., where new structure “emerges” from the parts beyond the sum of those parts).

As you can see, the cognitive approach aligns much more closely with the “holistic” approach to science that is becoming increasingly mainstream these days, as opposed to the old approach which is definitely reductionist in nature (i.e., the assumption that a whole can be understood simply by understanding its parts). We will see how the cognitive approach goes about doing this when I begin posting the lessons.

Cognitive linguists believe Language is a reflection of what is going on inside the human mind, and therefore can tell us about the workings of the human mind. Because the seat of ?mind? exists in the brain, any conclusions drawn by cognitive linguists from linguistic data should be empirically consistent with the findings of neuroscience and psychology, and, in fact, linguistics and these other fields must all be considered not just related, but inter-related sub-fields of the larger field of cognitive science, the study of the workings of the human brain and its primary product, human consciousness and the phenomenon of mind.

As a preview of what we'll be looking at, I briefly list below some of the major findings/conclusions of cognitive linguistics that have no parallel in the ?standard? approach to linguistics:

1) The human mind is embodied, i.e., the way the mind works is fundamentally tied to the nature of the human body as a whole. Any ontological separation between mind and body, whether physically objective or spiritually subjective can be empirically shown to be nonsense. Cartesian dualism is bunk. Descartes was just plain wrong.

2) Human abstract thought is founded upon, and carried out by means of metaphor, operating on both sub-conscious and semi-conscious levels. Metaphor, far from its long-assumed “para-linguistic” role as simply a rhetorical device, is in fact fundamental to human abstract conceptualization. Essentially, if a concept or thought is based on something more complex than direct input from the five senses or basic bodily proprioception (pain, heat, cold, illness, physical well-being), it is conceived, thought about, and understood in metaphorical terms.

3) Aristotelian categorization (i.e., classical set theory) is bunk. Human beings create all subconscious categories (including the natural categories existing within the grammar and lexicon of natural languages) based on fuzzy logic, prototypes and family resemblances. This logic is very different from classical Aristotelian logic and set theory.

4) The lexico-semantics of human language exists within a subconscious semantical framework, specific to each language, which consists of rigid (i.e., rule-based), overlapping categories)which have a DIRECT impact on the morpho-syntactical and morpho-lexical structure of the language. The Chomskian notion that syntax exists separate and apart from semantics is na?ve and plain wrong. Semantics to an extremely large extent, dictates syntax.

5) The nature of Truth is fundamentally contextual and fundamentally relative, i.e., there is virtually no such thing as contextless, absolute truth.

There are other findings we will explore, but the above will be the topics of the first several lessons. I hope to have the first of these posted sometime late Sunday.

Simple. You want your conlangs and concultures to be realistic, right? For most of you that means, you want them to follow human universals (or deliberately violate them if your conworld/culture is non-human); at the same time, you want the specific overt and covert structures of your conlangs to be unique, right? I mean, after all, where's the fun and sense of authenticity if your conculture uses the same idiomatic structures as English such as “you're pulling my leg.” You want the equivalent in your conlang to be something like “your fondling my bum” or “breaking my elbow”, right? And yet, by not understanding how human languages, particularly your native language, have the same sort of language-specific structures at the subconscious level, you are at a great risk of simply building your supposedly “unique” conlang on a covert foundation that is simply a direct parallel of English, for example the idea that “going up” to Canada, means “travelling northward” or that all languages MUST have separate verbs for “to buy” and “to sell”, right, when in fact, cognitive linguistics shows that buy and sell exist within a subconscious semantic “frame” whose specific attributes and “edges” are peculiar to each language. And you also want to be able to decide whether your conlang needs to accomplish the same pronominal reference switching that I illustrated in the “If I were you, I'd hate me/myself” example, right?

So, by understanding a little bit about this stuff, you can begin to analyze how your conlang works at a deeper level than you ever tho

ught possible.

I will try to present this material as simply as possible, as I am only trying to provide an understanding of basic concepts rather than extreme detail. Those who find their appetites whetted can then proceed to explore more detailed material on their own. Several introductory textbooks have been mentioned above. The one I will be mostly relying upon for my material is David Lee's Cognitive Linguistics: An Introduction, which is less technical and, IMO, more coherently presented but also more elementary than the more comprehensive Cambridge Series book by Croft and Cruse mentioned in an earlier post. Anyway, shall we get down to it??

Let's start with a simple working definition of the word “cognitive” which will be sufficient for our purposes: “having to do with how the mind works”. And write my essay about it.

Cognitive linguistics arose during the 1970s essentially as a reaction to three things: (1) dissatisfaction with the existing linguistics paradigm of the time, Noam Chomsky's “generative grammar” due to the inability of generative grammar to provide explanations for an increasing number of problem examples and observations about language, especially when it was applied to non-Indo-European languages; (2) the failed attempts by Chomskian-trained linguists to create a “generative semantics”, i.e., the attempt to extend Chomsky's theory of generative grammar into the realm of semantics; and (3) the pioneering work on human categorization done by a psychologist named Eleanor Rosch, whose evidence strongly suggested that the subconscious human mind creates categories in ways previously unsuspected (although work by the philosopher Ludwig Wittgenstein had foreshadowed Rosch's findings, e.g., Wittgenstein's classic analysis of the German word “spiel” [English “game”]).

The first linguists to formally pursue a new non-Chomskian approach to linguistics were Charles Fillmore at UC Berkeley and Ronald Langacker at UC San Diego. Langacker, a former Chomskian, finally became so fed up with all the “exceptions” that had to be made in generative grammar the more he explored the subtleties of language, that he finally concluded Chomsky's theories must simply be wrong. Rather than try to fix generative grammar, he instead decided to sit down and re-think linguistics from scratch, irrespective of any theory, with the following guiding principles: that language is a direct reflection of the workings of the human mind, and that any theory of grammar and semantics must be consistent with the way the human mind functions and the human brain physically manifests the processes of thinking and conceptualization. He began publishing a series of papers on his new ideas in the 1970s, closely followed by George Lakoff, Leonard Talmy, Gilles Fauconnier, Fillmore and others. Langacker eventual encapsulated all his ideas in the monumental two-volume work Foundations of Cognitive Grammar, published in 1987 and 1991. It is generally perceived that the publication of this work, along with Lakoff and Johnson's Metaphors We Live By in 1980 and Lakoff's Women, Fire, and Dangerous Things in 1987, established cognitive linguistics on a solid academic footing which has now led to the generally worldwide acceptance of the new paradigm as nearly co-equal with (and in many universities now surpassing) Chomsky's generative grammar.

While cognitive linguistics was originally defined in terms of a rebellion against Chomsky's theories, in the last decade, cognitive linguistics has matured to be considered a fully autonomous linguistic paradigm in its own right. Nevertheless, for beginners, it is still convenient to introduce cognitive linguistics in comparati

e terms to Chomsky's theory of generative grammar.

Chomsky, whose theories evolved during the late 1950s through 1970s to replace the previous structuralist and behaviorist models of language, believes the structure of language is determined by an innate, autonomous formal system of rules (analogous to the predicate calculus for those of you who?ve been trained in formal logic, but much more intricate and sophisticated). This formal system of rules, called universal grammar (UG), is inherent within the human brain at birth and is largely devoid of any association with meaning. This UG is also independent of other human cognitive faculties, i.e., it operates on its own within the brain, independent of any other non-linguistic cognitive processes.

Cognitive linguists, on the other hand, believe the structure of language is a direct reflection of human cognitive processes, and that there is no independent language faculty like UG in the brain. If there is, cognitive linguists generally believe it will eventually be found to be ultimately rooted in the general processes of human cognition itself (i.e., not peculiar to the phenomenon of language alone). The cognitivists believe that the grammatical structures of language are directly associated with the way people conceptualize (i.e., think about and understand) any given situation in the world. Syntax, morphology, even phonology are conceptual in nature, i.e., they are merely input and output of those cognitive processes within the human mind that govern speaking and understanding. This idea is generally encapsulated in a phrase coined by Ronald Langacker and often repeated by cognitive linguists: grammar is conceptualization.

The other big difference between Chomsky and the cognitivists is where knowledge of language in general comes from. Chomsky argues that infants know how to put language components together innately (because of their reliance on the UG), i.e., they do not (solely) rely on having to hear how to put words together correctly (i.e., syntax) from listening to their family and other sources such as television. Chomsky believes evidence exists to support this notion in his famous “poverty of the stimulus” argument (the one that Kirk has railed about in some of his posts in this and other threads), saying that children in general are “too good” at learning language so quickly, i.e., they don't get exposed to a sufficiently large corpus of language stimuli/data to work with to figure out so quickly how their native language works, therefore they must have an innate faculty (the UG) to subconsciously tell them about things like syntactic relations (e.g., case morphology), tenses, aspect, clause structure, grammatical transformations such as active-into-passive voice, etc.

The cognitivists, on the other hand, reject this argument entirely and do not believe in the “poverty of stimulus” argument. Cognitivists firmly believe that knowledge of language comes strictly from language use. Infants learn language by listening, observing, pattern recognition and pattern-matching, imitation and trial-and-error attempts to learn the grammatical rules of their native language. The reason Junior first says “Mommy drink” before he says “Mommy, I want a drink” is simply because the former is easier and therefore gets tried out and used first, while the more sophisticated (and “correct”) structure of the latter gets learned and used later on. In other words, language gets learned just like anything else gets learned. The use of language has nothing special about it that differentiates it from other cognitive processes. Rather, the human infant uses the same store of cognitive tools and processes to learn and use language as he learns to do anything else. Cognition is cognition. Learning is learning. Pattern-recognition and matching is pattern-recognition and matching; imitation and practice is imitation and practice, whether learning your native language or learning to ride a bicycle or select and put on clothes to wear.

Because cognitive linguists believe that grammar is conceptualization (see 1.3 above), the core area of study to date within the field of cognitive linguistics is semantics and morpho-semantics and the way these two components of language determine syntax (the way words are put together to create grammatically acceptable phrases and sentences). While cognitive linguists fully believe that the cognitive paradigm extends to more nuts-and-bolts units of language such as phonology and morphology, little work has yet been done in these areas during the brief quarter-century that the paradigm has existed. (Remember the analogy from my previous post about looking at the whole car first and driving it, THEN start taking it apart rather than starting with the pieces and parts and putting it together?) In regard to going the other direction beyond syntax into the linguistic areas of pragmatics and discourse analysis, many cognitive linguists believe that these two areas of linguistics actually don't exist. Rather, as cognitive analysis of language begins to delve more deeply as time goes on, the usage of language in everyday contexts which is the realm of pragmatics and discourse analysis will simply be found to be based purely on the same semantically driven rules of language structure that contextless or normal language structures are based upon. In other words, while linguists normally study the structure of sentences like “I have to urinate” rather than the semantically equivalent colloquial version “I gotta go”, the production of the two sentences by living, breathing English speakers is nevertheless analyzable by the same kinds of semantically driven, context-filled rules, INCLUDING rules to govern the speaker's very choice of using one sentence as opposed to the other. This area of language wouldn't be touched by a Chomskian with a ten-foot pole, whereas to a cognitivist, if people say it, it's fair game for linguistic analysis. Indeed, linguistics can't be considered complete until linguists understand why a person chooses to say it one way as opposed to the other. Needless to say, under this view of the scope of linguistics, the science of linguistics can be considered to be almost still in its infancy.

Anyway, enough by way of introduction.

Traditional generative grammar generally considers the following two sentences to be semantically equivalent, their differences explicable solely on the basis of the application of formal syntactical rules (e.g., deriving one from the other via transformational rules).

1a) John gave the book to Mary.
1b) John gave Mary the book.

The idea is that any situation can be mentally decomposed into parts, each of which bears a one-to-one correspondence with some element of language, i.e., a one-to-one mapping between the external world and a corresponding element of language, after which, simply application of syntactical rules takes over with no further impact on meaning or conceptualization. The pair of sentences above illustrate the well-known distinction between a “ditransitive” construction versus the presumably equivalent “complement” structure for indirect objects.

However, cognitive linguists do not believe in this sort of one-to-one mapping. Instead, any situation can be construed in different ways conceptually. Different construals are encoded in different sentence patterns. Thus, the differences in Sentence (1a) and (1b) above are indicative of a conceptual difference, i.e., a difference in meaning. Evidence for this exists in the fact that, in some cases, the ditransitive pattern is not allowed, e.g.,

2a) He brought the wine to the table.
2b) ??He brought the table the wine.

The ???? symbol indicates the acceptability of this sentence to native English speakers is highly questionable (i.e., it sounds weird). Sentence (2b) tends to be acceptable only if we consider the word table to be a metonymn substituting for “the people at the table” (metonymy is the substitution of one concept to represent another, such as when we use a place or object to represent the person associated with that object, as in The White House is threatening a blockade, or The ham-and-cheese wants a refill on his coffee.)

Here's another interesting example involving the ditransitive construction:

3a) Loretta gave Sue a wedding gift.
3b) Loretta gave a wedding gift to Sue.

Again, most generative grammarians would state that these two sentences are semantically equivalent. However, there is a subtle semantic difference between the two sentences. The first strongly implies that the wedding gift is for Sue, i.e., Sue is the bride and intended recipient. The second sentence, however, invites the possibility that Sue is only a temporary or circumstantial goal for the act of giving, but not the bride and intended recipient. For example, if Sue is merely a guest at the wedding and Loretta needed Sue?s help carrying an armload of wedding gifts, she might give a wedding gift to Sue, but that does not mean she would give Sue a wedding gift. This is an example of a conceptual principle cognitivists refer to as iconicity. The concept of iconicity is that the word order of a sentence reflects a conceptual distinction in the mind of the speaker. There are several types of iconicity. The type of iconicity distinguishing a recipient from a directional goal is an example of what is termed distance iconicity, because the two linked words are made more distant from each other in the sentence as a reflection of their more circumstantial association. Thus, the order of the words themselves reflects information about how we are to understand the utterance.

In English and other Western languages, the most common way in which iconicity is manifested is what is termed sequential order iconicity, the idea that the actual sequential order of words in a phrase or sentence reflects the sequential order of the events they describe. For example, the phrases “eye it, try it, buy it”, “I came, I saw, I conquered”, or “dine and dash” describe sequential events where the sequence of the words reflect the sequence of the events. What is most important is that re-ordering of the words either changes the meaning of the phrase or leads to semantic nonsense, e.g., “buy it, eye it, try it” implies that a different sequence of events actually takes place than “eye it, try it, buy it”. This can be more dramatically illustrated with the following pair of sentences.

4a) Jane got married and had a baby.
4b) Jane had a baby and got married.

In English, the ambiguous word “and” is interpreted as connecting a sequence of events, i.e., “and” is interpreted to mean sequential “then” (= “and following that”, “then next” or “then later”). As a result, the meanings of the two sentences imply very different social interpretations about Jane.

Besides the reflection of sequential order, other types of word-order iconicity are possible. For example, compare the subtle difference in meaning between the following two sentences:

5a) Sam painted the fence white.
5b) Sam painted the white fence.

In the first sentence, we do not know what color the fence was prior to being painted, or even if it was a new fence that had never been painted before. In the second sentence, not only do we know what color the fence had been, but also that it was not previously unpainted, however, we do not necessarily know what its new color is. This sort of iconicity is used to convey a resultative state of affairs, i.e., by placing the adjective “white” after the word “fence” (seemingly in violation of the usual adjective-before-noun word order used in English), we describe a resulting state of affairs.

2.3 Perspective Another big way that alternate construals of the same situation are manifested is via a difference in what cognitivists call perspective. Compare the following sets of sentences:

6a) The path descends steeply into the valley.
6b) The path climbs steeply out of the valley.

7a) John bought the car from Mary.
7b) Mary sold the car to John.

8a) The pen is on the table.
8b) ??The table is under the pen.

The two sentences in each pair above describe the exact same scene, yet they do not have the same meaning, and the actual position or physical viewpoint of the speaker is not what determines the difference. It's perfectly OK to say either (6a) or (6b) whether you're at the top of the valley looking down or at the bottom of the valley looking up. Either (7a) or (7b) can be used by any third party to describe the single commercial transaction that took place between John and Mary. It's all dependent on the perspective that the speaker chooses to emphasize. (Wouldn't it be interesting to have a specific morphological category in a conlang to automatically switch such a perspective on a verb, so that you didn't need to have separate lexemes for what are merely opposing or complementary perspectives?)

Sentence pair (8a) and (8b) illustrate that pragmatic factors can affect the acceptability of a sentence, even if there is nothing wrong with the sentence syntactically. It is not within the normal realm of conceptual perspective to position a table with respect to a pen, rather the normal practice is the other way around. Thus, actual human conceptualization based on external world experience ends up having a direct impact on grammatical usage, irrespective of the acceptability of syntax.

There are several other conceptual factors which play a role in construal, but most of these have to do with concepts of physical space. The human mind's conceptualization of physical space is one of the most important aspects of cognitive linguistics. Therefore, before continuing on with simply listing factors which impact construal, we will first devote our study to the conceptualization of space. Concepts we will learn about are landmarks and trajectors, as well as image schemas. The concept of image schemas will be very important in leading us later on into the study of metaphor and frames. Stay tuned for Lesson No. 3!

Consider the physical nature of one's actions when inserting a foot into a stirrup in preparation to ride a horse. Compare this to the physical nature of one's actions when preparing to wear a ring on one's finger. Both actions involve placing a bodily extention into an enclosing metallic ring of metal and keeping it there, right? So why is the second sentence of the following pair generally unacceptable by native speakers of English?

1a) I put my foot in(to) the stirrup.
1b) ?? I put my finger in(to) the ring.

Conversely,

2a) ?? I put the stirrup on my foot.
2b) I put the ring on my finger.

If both situations involve the same kinds of physical action, why are they grammaticized differently using different prepositions (“in” versus “on”)?

Cognitive linguistics answers this question based on a powerful model of how human beings conceptualize space. Furthermore, our conceptualizations about space are so basic to our cognition that we subconsciously extend those spatial conceptualizations to describe far more abstract situations and relationships between entities.

Infants spend most of their waking hours observing things and people in motion, and manipulating toys and other items within their reach, placing them on top of one another, next to one another, inside one another, etc., as well as observing others manipulate objects. They come to understand how the world works at a physical level by grasping things, picking them up, dropping them, pulling and pushing them, hitting them, and throwing them, always watching how the object responds. These spatial relationships and concepts of motion, along with sensory input become well-understood even before the infant is able to use words to describe them. This sensorimotor knowledge is so cognitively fundamental that human beings naturally use space, motion, and the senses as domains for conceptually structuring less concrete, even entirely abstract aspects of our experience.

Research by cognitive psychologists and cognitive linguists support the conclusion that humans cognize spatial relationships between objects in terms of a landmark and a trajector (these are Langacker?s original terms; some authors use different terms for these). For our purposes, think of a landmark as an entity with respect to which some other entity moves. That other entity that moves with respect to the landmark is the trajector. So when screwing in a lightbulb, the bulb (trajector) moves with respect to the socket (landmark), whereas in the identical physical relationship between a jar and a lid, it is the lid (trajector) that moves with respect to the jar (landmark). Perhaps now you can begin to understand why we say “the bulb in the socket” but not *”the jar in the lid”, as well as “the lid on the jar” but not *”the socket on the bulb”, even though the physical relationship and connection between the two parts is highly similar, both involving a quasi-cylindrical glass object whose upper end bears a spiral groove pattern which connects via circular motion into a sparate ring-like metallic component also bearing spiral grooves.

Basic spatial relationships in English and other Indo-European languages are referenced by simple prepositions such as “in”, “on”, and “at”. Some of you might say at this point, “wait a minute, those little prepositions are far too vague in their meanings to really represent any sort of 'basic' spatial relationship”. And so it might seem at first, yet ask a native English speaker to give sentences using these three prepositions and nine times out of ten you will get sentences such as:

3a) The cat is in the house.
3b) The pen is on the table.
3c) My friends are at the supermarket.

as opposed to sentences such as

4a) The cat is in agony.
4b) The train is on time.
4c) My skills are at a new level.

Note that sentences (4a,b,c) are no less grammatical than (3a,b,c), nor are they more complex syntactically. Generative grammar would have no recourse but to explain the tendency via the social forces of pragmatics or discourse rules. But cognitive linguists have a more satisfying answer. Sentences (4a,b,c) represent extended or more abstract applications of the basic meanings of interiority, support against gravity, and locational reference, at the heart of the words “in”, “on”, and “at” so speakers are less likely to use examples like these. A child first learns to map a correspondence between his established sensorimotor knowledge of interiority to the word “in”. Later, of course, more exacting words such as “inside” and even “interiority” are learned, but the child retains a fundamental association with the first word mapped to the concept “in” with the raw sensorimotor concept ingrained during infancy. As a result, a prototypical or best example notion of what “in” means, is retained even into adulthood at a subconscious level. (As some of you might be guessing now, I am foreshadowing our eventual lesson on prototypes and fuzzy sets.) This explains why sentences (3a,b,c) are more likely examples of “in”, “on”, “at” than (4a,b,c), as they describe situations nearer to the prototype meaning.

As to just how and why simple words like “in”, “on” and “at” get extended to usages such as “in agony”, “on time” and “at a new level,” will be discussed in a future lesson. We'll also take a look at what's really going on in the “foot IN the stirrup” versus “ring ON the finger” examples.

We saw in Lesson No. 3 above that the choice of using one preposition over another in English is not necessarily based on the objective properties of the observed situation, but rather on the way in with the situation's various elements and their inter-relationships are construed. This is why it is possible to say “The fly on the ceiling” even though the actual physical situation should theoretically call for the phrase “The fly under the ceiling”. Here we start to see the problems inherent in traditional (Aristotelian) theory of categories (classical set theory) where all members must share a necessary feature. Examples like “fly on the ceiling” suggest that even members with diametrically opposed features can still qualify for membership in the same category.

Compare the following two sentences:

1a) Sam is in the bus/train/ plane/boat.
1b) Sam is on the bus/train/plane/boat.

I would ask all of you who are native speakers of English to stop and think to yourself what the semantic difference is between these two sentences. Although it is subtle, you definitely sense a difference, sentence (1a) probably seeming somehow more exact or concrete. The reason that English allows both constructions, and the reason they both carry distinct meanings, is because English allows larger vehicles such as busses, trains, planes, and boats to be construed either as containers, or as supporting surfaces against gravity. Essentially, we're back to our basic sensorimotor knowledge from infancy. Is a person taking the bus to work, being carried within a container? Or is he being moved over the ground while sitting upon a supporting surface? English spatial conceptualization allows both possibilities and encodes (i.e., maps) these possibilities to the prepositions “in” and “on”.

This idea of a complex object such as a bus or airplane being seen as a container and also as a supporting surface is a basic example of the concept of image schemas (while the correct plural is schemata, all authors in this area use schemas). An image schema is a subconscious conceptual metaphor, where basic spatial concepts from early childhood such as containment and support against gravity, become extended to apply to highly complex, even abstract situations, allowing us to subconsciously understand them in a simple way.

Actually, the “in” versus “on” a vehicle choice is even more subtle than described above. For example, if the vehicle is not moving, speakers are more likely to say Sam is “in” the train, as when someone on a station platform is asking for the whereabouts of Sam who is already waiting aboard the train which is yet to depart. On the other hand, if Sam's train is due to arrive in five minutes, we are more likely to say he is “on the next train” rather than “in the next train”. This is because we are more likely to subconsciously think of the train as a transport/supporting surface mechanism moving relative to the ground in this context, than as a container. The fact that people are contained inside it is seen as less important in this context than the fact that they are attached to and supported by it. The level of abstraction can become quite conceptually complex, as when we say “Sam is on the 8:45 from Boston”. Here, the idea of transport is paramount, to the point that the idea of containment becomes so irrelevant that encoding it into ??”Sam is in the 8:45 from Boston” sounds ludicrous to a native speaker.

Perhaps now you see that in the bus seems more “concrete” than “on the bus” because it is. “In” is conceptually closer to the idea of being inside a container such as a large vehicle, while “on” requires a further degree of abstraction, as it extends to the more complex situation of a supporting surface moving relative to the ground for the purpose of transportation.

There is a large amount of literature by cognitive linguists on the encoding of contrasting prepositions from basic to highly abstract meanings. Based on what we've discussed above, think about the way spatial conceptualization is being applied to abstract situations in the following sentences:

a) There are bubbles on the surface.
b) There are bubbles at the surface.

a) She has wrinkles on her skin.
b) She has wrinkles in her skin.

But compare…
c) That vase has cracks in it.
d) *That vase has cracks on it.

…and yet…
e) That vase has spots in it.
f) That vase has spots on it.

Also, try to come up with a cognitive explanation, based on the idea of the containment versus transport/support image schemas, as to why the “in” versus “on” distinction doesn't work with automobiles, and in an opposite pattern, does not work for motorcycles, e.g.,

a) Sue is in the car.
b) *Sue is on the car. [can only mean she is outside on top of the hood or roof of the car.]

a) *Joe is in the motorcycle.
b) Joe is on the motorcycle.

The idea of “in” representing a subconscious image schema involving seeing complex objects and even abstract ideas as containers and moving surfaces is very powerful, and begins to explain the extension of these basic spatial concepts to phrases like “in agony” (where “agony” is subconsciously construed as an enclosing place where one is trapped), “on time?”(where “time” is construed to be a linear moving surface like a roadway), or “at another level” (where one's skills are subconsciously construed as vertically arranged, shelf-like containers, on which things can be placed at varying heights relative to the ground).

We will be discussing image schemas in more detail in the next lesson. As you can see, we are working our way to our analysis of metaphor as perhaps the most powerful tool of human cognition. Stay tuned for Lesson No. 5!

Consider the English word “out”. Its basic concept is spatial in nature, specifically the idea of either static or dynamic exteriority, i.e., “being or moving to the outside of something”. So why does its usage extend to what appear to be contradictory examples? E.g.,

1a) The sun is out. The stars are out.
1b) The light is out. The fire is out.

2a) The sun came out.
2b) The sun went out. [e.g., in a sci-fi or fantasy context, or in a dream]

3a) Jim threw out a suggestion to the others.
3b) The others threw out Jim's suggestion.

The weirdness continues when we look at the following pair of sentences which a non-native speaker learning English might initially conclude to be opposites, but, in fact, are used to describe the exact same situation.

4a) Tom filled in the form.
4b) Tom filled out the form.

Here's an example of the same phenomenon using a different pair of prepositions.

5a) They closed up the bookstore.
5b) They closed down the bookstore.

In other cases, the meanings of the supposed “opposites” in fact refer to situations which bear no apparent relationship to one another:

6a) The student dropped in. [i.e., came to visit us]
6b) The student dropped out. [i.e., quit school]

Still in other cases, two prepositions whose basic meanings don't seem related to one another, yet can be used to describe the exact same situation:

7a) A lot of people turned up for the rally.
7b) A lot of people turned out for the rally.

The above examples would appear to support the conclusion that the correct usage of simple prepositions in English is semantically chaotic and that the rules governing usage must be learned by rote for each preposition (I know this is how I?ve studied prepositions in French, Spanish, Italian, Portuguese and German, and I still make mistakes all the time!) It would appear that:

1. The same preposition can express opposite meanings
2. Opposite meanings can be expressed by semantically unrelated prepositions
3. Pairs of prepositions which normally express opposite meanings can express similar meanings
4. Pairs of prepositions which normally express opposite meanings can express unrelated meanings
5. Similar meanings can be expressed by unrelated prepositions

Generative grammarians would probably simply indicate that any given preposition in the lexicon must be accompanied by a myriad of rules relating to usage that are ultimately arbitrary and specific to each lexical item.

But cognitive linguists can do better. Careful analysis of these kinds of phenomenon using the concept of image schemas that we've already talked about can simplify this chaos significantly.

Let's diagram the image schema for English “out”. In cognitive linguistics, such diagrams utilize LM and TR to represent the landmark and the trajector, as well as additional symbols to convey other entities considered part of the image schema, such as observers, conceptualizers (i.e., a mental observer of an abstract concept), sources, goals, etc. Arrows and dotted lines are also used to indicate spatial relationships and motion.

The image schema for “out” indicates that a trajector entity is located outside of a container or container-like space. The relationship can be static or dynamic as in The cat is out of the house versus The cat goes out of the house. This basic meaning is extended to a large number of situations, both tangible and abstract, wherever the notion of exteriority can be construed.

And so we get examples such as The lava spread out in which an entity expands in size, thus taking up an area previously outside of the original landmark domain. Similarly, Hand out the brochures and Roll out the carpet describe situations where the area of dissemination or presence becomes greater than the original, i.e., the entity has moved “out” and beyond its original occupying space.

Abstracting further, we extend the concept to situations where a trajector entity (or a part thereof) moves outward from the landmark without any accompanying notion of “exit”, as in They set out on their trip or He reached out for it. Because such concepts can also be expressed without the notion of exteriority (e.g., using verbs like enlarge, increase, leave, etc.), it becomes clear that imposing a notion of exteriority upon them is a form of construal (see Lesson No. 2, if necessary).

In our “The sun is out” example, an added semantic notion comes into play besides exit from an original location: the idea that the landmark is a container that is inaccessible to the observer and that the trajector (the sun) emerges from this container into the observer's perceptual field. This interpretation is supported by the fact that we use the verb “come” when saying the dynamic equivalent of the sentence, “The sun came out”, which conveys movement towards the speaker or conceptualizer. Notice how those sentences such as “The light is out” or “The fire is out” convey an opposite meaning where the metaphorical movement involves the trajector moving away from the observer into the inaccessible landmark, evidenced by the fact we use “go” instead of “come” when saying the dynamic equivalent:

8a) The fire is out.
8b) The fire went out.
8c) *The fire came out. [this sentence is grammatical but not synonymous with 8b]

These examples show how the notion of perspective we studied in Lesson No. 2 comes into play when considering image schemas. Thus…

We now see why and how the same word “out” can be used in semantically opposite contexts. It is not arbitrary as we initially thought or as a traditional grammarian would state. In fact, we don't even have to change the meaning of “out”; its usage in the two contexts is based on the same underlying notion of exteriority applied to both contexts, but differentiated by perspective. These same schemas explain contexts such as The news is out or The secret is out, where the accessbility/innaccessibility is in relation to one's cognitive field as opposed to perceptual field, given that these sentences describe abstract contexts rather than tangible ones as with the sun examples. The principles extend to contexts such as “She speaks out”, “It all turned out okay”, “I've sorted it out”, in which the result or outcome of a process or action becomes known or comprehensible.

Examples similar to our “The fire went out” schema include “The music drowned out his voice”, “I want to blot out the memory”, and “The criminal is hiding out”, where there is metaphorical movement of the trajector into a landmark that is inaccessible to the coneptualizer's perceptual or cognitive field. This notion of landmark inaccessibility is further extended metaphorically to contexts involving the unavailability of tangible or intangible resources, e.g., “We're out of gas”, “The supplies ran out”, “I'm tired out”.

In the next lesson we'll continue our discussion of image schemas and their metaphorical extension to abstract situations, and will be discussing the implications of image schemas on conlanging. In the meantime, if you find yourself intrigued by this stuff, I'll give you an assignment (or a challenge, if you prefer). Using the ideas from the above discussion, provide a cognitive explanation as to why Sentences (4a) and (4b) above can be used to describe the exact same situation. HINT: Think about how the two sentences are subtly different in meaning.

See ya next lesson!

I'll spend this lesson explaining some of the puzzlers I left you with in previous lessons (since no one seems to want to offer their own attempts to explain).

Remember these?

1a) I put my foot in(to) the stirrup.
1b) ?? I put my finger in(to) the ring.

2a) ?? I put the stirrup on my foot.
2b) I put the ring on my finger.

The difference between “in” versus “on” here is the same as the light bulb + socket versus jar + lid example I gave earlier. While the physical action of inserting a body part into a ring-like object is the same for both situations, the notion of which is the landmark and which is the trajector is reversed. So we subconsciously identify the stirrup as the landmark (the part that doesn't move relative to the trajector) while our foot is the trajector (the moving part with respect to the landmark), whereas we see the reverse relationships with the ring and finger, the ring being the trajector against the stationary landmark of the finger. “In(to)” is semantically associated with insertion within a surrounding landmark, so is appropriate for the stirrup example, whereas “on” is semantically associated with placement of a trajector so that it surrounds or covers a landmark, thus making it appropriate for use with the ring example.

Next example: 3a) There are bubbles on the surface.
3b) There are bubbles at the surface.

In sentence (3a) the bubbles are conceptualized as a simple positional reference, while in (3b) they are conceptualized as being at a point along a path, that path being the path they have traversed while rising from beneath the surface. The choice between which sentence to use is one of construal, as described in Lesson No. 2.

Next examples: 4a) She has wrinkles on her skin.
4b) She has wrinkles in her skin.
4c) That vase has cracks in it.
4d) ??That vase has cracks on it.
4e) That vase has spots in it.
4f) That vase has spots on it.

Sentence (4a) and (4b) differentiate between conceptualizing wrinkles solely by visual appearance relative to a surface, versus being integrated within that surface as a feature of that surface. (4c) is like (4b) in this regard, but (4d) is questionable because “cracks” are conceived as by nature extending in depth into a surface, thus being integrated as a feature of that surface, not lying upon that surface. Wrinkles, on the other hand, can be clearly conceptualized as being on a surface, based on the analogous use of wrinkles on a flat two-dimensional surface such as a bedsheet or a shirt, in which there is no three-dimensional depth for the feature to be integrated into. Sentences (4e) and (4f) show that “spots” can be conceptualized either way, but with different meanings. In (4e) the connotation is that the spots are permanently fixed within, or visible underneath, the surface (e.g., under a transparent glaze), whereas (4f) implies the spots are temporary and not part of the surface (e.g., they can potentially be removed, say, with a cloth).

In the next examples, I asked, based on the idea of the containment versus transport/support image schemas, why the “in” versus “on” distinction doesn?t work with automobiles, and in an opposite pattern, does not work for motorcycles, even though it works for saying either “in” or “on” the bus/train/boat/plane.

5a) Sue is in the car.
5b) *Sue is on the car. [can only mean she is outside on top of the hood or roof of the car.]
5c) *Joe is in the motorcycle.
5d) Joe is on the motorcycle.

Sentence (5a) works because, like busses, trains, etc., cars can be conceptualized as containers. Conversely, (5c) does not work because motorcycles do not have an interior compartment relative to an external shell/body by which they can be conceptualized as containers, whereas (5d) works because a rider can be easily conceptualized as a trajector relative to the landmark of the motorcycle (same as the ring on a finger example). The interesting sentence, of course, is (5b). Intuitively it would seem that, if we can say “on a bus/train/boat/plane”, we should be able to say “on a car” with the same meaning (i.e., the transport/support image schema). It would appear, then, that there is more to the transport/support image schema than meets the eye. Apparently the schema subconsciously includes the idea of a very large size (or possibly length) of the landmark relative to the trajector in order to function semantically. The relatively small size ratio between a car and a person, as compared to boats, trains, plains, and busses, is insufficient to meet the salience necessary to permit use of the schema.

Lastly, I asked why opposite prepositions can be used to describe the same exact situation, as in:
4a) Tom filled in the form.
4b) Tom filled out the form.

Again, our answer is pure construal. Sentence (4a) conceptualizes a form as via the container schema, in which writing answers on it is seen as analogous to placing things inside of its internal compartments. Sentence (4b) on the other hand, conceptualizes the form in terms of an additive schema as something that grows in size or weight as answers are written on it. This is analogous to the image schema for “spread out” shown in an earlier lesson.

Time to discuss metaphor…

Most people think of metaphor (as do most linguists other than cognitive linguists) as a consciously applied rhetorical device, as in He must have wings on his feet to get here so fast! As such, metaphor prior to cognitive linguistics was considered to have little if any bearing on the relationship between semantics and syntax or morpho-syntax. Then in 1980, George Lakoff and Mark Johnson published Metaphors We Live By, immediately establishing one of the central pillars of cognitive linguistics, and creating implications beyond linguistics in the realm of philosophy and psychology which continue to be felt.

Lakoff and Johnson's premise is fairly straightforward on its surface, but profound in its implications upon further analysis: human beings structure their understanding of their experiences in the world via conceptual metaphors derived from basic sensorimotor, spatial, emotional, and other fundamental concepts learned during infancy and early childhood, in which these simpler, more basic concepts are used as a framework for conceptualizing more abstract experiences and situations. In fact, basic childhood concepts are the source domains which are in turn cognitively mapped to a target domain. Note that these target domains do not literally lend themselves to the source domain in any overtly physical or literal manner. Thus, these mappings are entirely metaphorical in nature. For some simple examples of what Lakoff and Johnson are talking about, observe the following pairs of sentences (those preceded by ?? are considered questionable in terms of acceptance by native speakers, i.e., they sound weird or even nonsensical):

1a) We're head up north.
1b) ??We're headed down north.
1c) We're vacationing up in San Francisco. [if said, for example, to a neighbor in Los Angeles]
1d) ??We're vacationing up in San Francisco. [if said, for example, to a neighbor in Vancouver, Canada]

2a) We're headed down south.
2b) ??We're headed up south. [EDIT: this example corrected Jan.19, 2006]
2c) We're vacationing down in San Francisco. [if said, for example, to a neighbor in Vancouver, Canada]
2d) ??We're vacationing down in San Francisco. [if said, for example, to a neighbor in Los Angeles]

3a) We're headed out west.
3b) ??We're headed out east. (This example is at least questionable in the U.S.; I'm uncertain about the U.K.)

4a) We're headed back east.
4b) ??We're headed back west. (This example is acceptable only if the intended meaning is ?we?re returning westward.)

From the above examples, the following conceptual metaphors can be elicited:

NORTH(WARD) IS UP(WARD)
SOUTH(WARD) IS DOWN(WARD)
EAST(WARD) IS BACK(WARD)
WEST(WARD) IS OUT(WARD)

There is no literal relation in the physical world between upward movement (i.e., toward the sky) and moving northward on (or parallel to) the ground, nor any literal relation between moving downward (i.e., in the direction of gravity) and moving southward, yet it is likely most native English speakers would need to think about this consciously before agreeing. This is because the above conceptual metaphors are so ingrained into the way we think of traveling long distances in the cardinal directions that they have become largely subconscious, used as if there is a literal relationship between the cardinal direction and the image schemas (see Lesson 5) for “up”, “down”, “back” and “out”.

Such conceptual metaphors exist in virtually every realm of abstract thought as expressed in language. Naturally, many of these are spatial in nature as seen in the above examples. Other spatial metaphors in English include:

MORE IS UP / LESS IS DOWN
GOOD IS UP / BAD IS DOWN
HAPPY IS UP / SAD IS DOWN
THE FUTURE IS AHEAD / THE PAST IS BEHIND

Here are a few other examples of the hundreds that have been identified in English.

SEEING IS TOUCHING/EYES ARE LIMBS
I can't take my eyes off her.
His eyes were glued to the the TV.
I was able to pick out every detail of the pattern.
She ran her eyes over his body.

EMOTIONAL EFFECT IS PHYSICAL CONTACT
Her death hit him hard.
I was struck by his sincerity.
She's a knockout.
She was touched by his remark.
I was blown away, dude.

LIFE IS A GAMBLING GAME
I'll take my chances.
The odds are against us.
It's a toss-up.
He's got an ace up his sleeve.
He plays it close to the vest.
Where were you when the chips were down?

LOVE IS MADNESS
I'm crazy about her.
You're driving me out of my mind.
He always raves about you.
Her fans are mad for her.
I'm just wild about Harry.

A key point to understand about conceptual metaphor is that these metaphors are not static idiomatic structures specific to the specific linguisitic expressions. Rather, it is the entirety of the source domain which is utilized as a means for understanding the target domain and many different elements from the source domain can be mapped to a corresponding element in the target domain. Thus, in mapping the source domain of “heat of a fluid” to the target domain of “anger”, various expressions associated with the source domain can be used to talk about and think about the target domain, thus:

I reached my boiling point.
She blew up on me.
His anger boiled over.
Their anger simmered during the presentation.
He erupted in anger.
I could feel my anger building up inside.

It is important to realize that Metaphor is linked with construal (see Lesson 2) because different ways of thinking about a particular event or situation (i.e., different construals of the event or situation) are associated with different conceptual metaphors. For example, it is common to conceptualize “intimacy” in terms of temperature, e.g.,

She finally warmed up to him.
He is a cold person
I treated her very cooly
Those two are hot for each other.

However, it is possible to equate intimacy with distance as well, e.g.,

I feel very close to you.
Her manner is very distant.
We?re drifting apart
He is very unapproachable.

An argument can be conceptualized as a building, e.g.,

That supports what I'm saying.
Your argument is crumbling.
I have evidence that buttresses her statement.
I'm building up evidence for my claim.

On the other hand, an argument can also be conceptualized as a journey, e.g.,

What are you driving at?
I want to take that point a little further.
That leads to the following conclusion.
I don't follow you.
You've lost me.
I'm not with you.

Many of the conceptual metaphors found in English can be shown to be more specific sorts of instances of more general conceptual metaphors. Some of the main conceptual metaphors for English are:

STATES ARE LOCATIONS
CHANGES ARE MOVEMENTS
CAUSES ARE FORCES
ACTIONS ARE SELF-PROPELLED MOVEMENTS
PURPOSES ARE DESTINATIONS
MEANS ARE PATHS TO DESTINATIONS
ACTION IS DIRECTED MOTION

Lakoff and Johnson state that conceptual metaphors do not simply involve how we talk about our experience, but also how we think about them. They and others have shown many examples of how this has political and social implications. One line of research in the use of conceptual metaphor showed that official government publications have almost always utilized conceptual metaphor when talking about nuclear weapons, using source domains which make the topic more palatable to the general public. Other research has shown how the description of the human body within the medical profession and medical schools has evolved from seeing the body as a machine where things are either broken or fixed to one where the human body is a homeostatic system capable of being in or out of balance. This paradigm shift has led to new treatments for disease and whole new lines of medical research and practices. Lakoff believes that “vast domains of our experience, understanding, reasoning, and practice are metaphorically structured”.

Hopefully, you are now beginning to see just how intricate, the subconscious structure of language can be in terms of the relationship between subconscious conceptualization and its correspondence with actual words.

In designing your conlangs, how many of you have ever thought about sentence pairs like those shown in Lesson 5? Have you ever asked yourself Should my language allow opposite locative constructions to refer to the same or similar situations, and the same locative structure to apply to opposite situations? If so, should I use the same pattern as English, or should I come up with a system of correspondence unique to my language? If I do, what conceptual logic will underly it? (In other words, how can I be sure I'm not just creating an idiom but instead am reflecting a straightforward cognitive extension of meaning from basic spatially-oriented usages to abstract, non-spatial contexts?) Are you now seeing that a lot of what you probably considered to be idiomatic constructions in language, are in fact simply the application of subconscious conceptual mappings from a basic domain of human experience and cognition to a more complex one.

What conceptual metaphors shall I use? When the time comes to translate an English sentence like “I'm crazy about her”?? are you simply going to go looking up your conlang?s word for “crazy” or are you first going to say, “Wait! The choice to conceptualize love metaphorically with madness is arbitrary to English (and other languages). Why can't I make my conlang's conceptual metaphor something different that says something about the personality of my conworld inhabitants' psyche and culture? Say, perhaps, LOVE IS FOOD ('She is my strawberry pie.') or LOVE IS DANCING,” etc.

Doesn't this make you want to grab your notes and start mapping out your conlang's image schemas and conceptual metaphors? I did image schemas for Ithkuil spatial roots (the language doesn't have prepositions). How about you? Do you really want your language to say that a poster hangs “on” a wall, or should it be like French which uses ??? as a reflection of a different image schema? Or a different image schema entirely? The Ithkuil root used to describe posters on a wall conveys a basic spatial meaning of vertical support against gravity, as opposed to the horizontal (or perpendicular) support against gravity of English “on”. Perhaps some of you might want to come up with even more interesting ways to designate such a spatial relationship.

Interesting stuff, eh?

If you were asked to define the word “eat” most of us would offer some variation of a simple dictionary type of definition such as “ingest food”. However, the word “eat”, in addition to having a simple definition, is associated with a semantic frame, a subconscious schematic representation of a particular type of situation together with a mental list of all the different participants, props, and other conceptual roles that are seen as components of such situations. Thus for “eat” we automatically conceive of associated concepts such as food, silverware, cups and plates, a heating source such as a stove, microwave or campfire, packages and cans of processed food, grocery stores, restaurants, menus, as well as more abstract concepts such as hunger. Those items that do not fit the frame, give rise to semantically unacceptable sentences such as *”The rock ate the candy bar” or *”Will you be eating the vacuum cleaner or the pencil sharpener for lunch?”

Often, a particular frame represents a situation which is associated with a simplistic, idealized view of reality which, in fact, does not lend itself well to modern real-life situations. An example of this is the word “bachelor”. Most people, when asked to define the word bachelor, would say something like “an unmarried adult male” or perhaps “an adult male human being who has never been married”. However, note the difficulty that bachelor presents in the following exercise:

Are the following persons “bachelors”?
1a) The Pope
1b) Tarzan
1c) An adult male living with his longtime girlfriend
1d) A male homosexual
1e) A male homosexual living with his longtime boyfriend

The idea of frames is important in that it goes against the classical or standard notion that the meaning of concepts is a plus-or-minus set of defining features for any given concept. Thus, “bachelor” is merely a concept consisting of a collection of semantic features such as [+MALE], [+ADULT], [? MARRIED]. Needless to say, the above examples belie this feature-based notion of conceptualization.

As can also be inferred from the “bachelor” examples above, frames are often associated with culturally-conditioned attitudes and expectations as well. Because frames operate mostly subconsciously, they can be used for clever social manipulation. In a recent book, Lakoff wrote about how the Republican party in the U.S. has co-opted the debate on taxes over the past decade or more by equating the “taxation” frame with the “relief” frame. The word “relief” conjures up thoughts of release from pain or an uncomfortable or unfair burden. By equating the two via the phrase “tax relief” and hitting the public over the head with the phrase again and again, they have succeeded in instilling a new frame into the minds of Americans, against which the Democrats have little defense. Even when the Democrats try to make a case for raising taxes, they subconsciously defend the idea within the “tax relief” frame. Lakoff suggests that the Democrats fight fire with fire by coining their own new frame and selling it to the public themselves, the frame being “tax responsibility” in which taxes would be seen as one paying his/her fair share for services and conveniences which one uses but could never create or build for oneself, e.g., roads, schools, and other public infrastructure.

Frames also overlap and borrow elements from one another, so the “commercial transaction” frame with its associated words like buy, sell, cost, price, money, credit card, purchase, customer, salesperson, store, etc. overlaps with frames associated with “money” and “product” and “economy”.

Frames have important ramifications for those conlangers whose conlangs are associated with a particular conculture/conworld. Semantic frames are specific to any given language (in fact, probably any given dialect group, given the differences in certain frames you can probably think of yourself between American versus British usage, e.g., “sports” frame and cricket anyone?)

Frame semantics was first conceived by the UC Berkeley linguist Charles Fillmore. He has devoted the past few years to creating FrameNet, a massive database and query system that attempts to identify the semantic frames within the English language, showing their associations and lists of words. The FrameNet website is absolutely fascinating and has much more information about frames than I have presented here. You can also look up a particular frame and use a tool called FrameGrapher to see a graphic representation of how it relates to other frames. The website is here. I would recommend that anyone visiting the site start with the FAQs, which give a good overview of the site's purpose, how to use it, and more information in general about frame semantics.

An excellent example of the implications of frame semantics is given by Lakoff and Johnson involving the word “gun”. (NOTE: Lakoff & Johnson do not use the term “frame” in their book, referring to the concept in a more descriptive way as “an experiential gestalt of inherent and interactional properties”.) Ask a person to define a “gun” and they will give a dictionary-style definition of a gun which mentions its inherent physical properties as well as its primary function use (i.e., to shoot bullets). But Lakoff & Johnson demonstrate how “gun” carries with it a subconscious semantic frame which is a necessary part of our conceptual understanding of the word. I say necessary because without the subconscious entailments inherent in the frame for “gun”, there are phrases and sentences which would make no sense to us. The reason such phrases or sentences do make sense to us is because of our subconscious understanding of the frame. Lakoff & Johnson demonstrate this by applying the adjective “fake” to render the form “fake gun”. On a conscious level, we generally consider that a “fake gun” is not a gun, applying an either-or, classical binary logic, i.e., a gun is either real or fake, with no other conditions necessary to understand what that means in context. However, observe the following about the phrase “fake gun”:

A fake gun has to look enough like a gun for the purpose at hand, e.g., as a prop in a play or movie, as a means of threatening someone with a weapon, etc. The props manager of the play can't hand you a dishtowel and say “Here's your fake gun for the scene, Mr. Bogart”. This implies that the object 'has to have the contextuallly appropriate perceptual properties of a gun' (to quote L&J).

You also have to be able to perform enough of the appropriate physical manipulations that you would with a real gun, i.e., be able to hold it or use it a certain way by which observers will acknowledge its appearance/use is as a gun. As L&J put it, it has to 'maintain what we might call motor-activity properties of a gun'.

There's also the point of why you have or are using a fake gun. It must serve at least some of the purposes that a real gun would serve, e.g., for threatening, for display, etc.

So, what makes the gun fake is that it cannot function like a real gun, i.e., it can't shoot bullets.

Lastly, it cannot have originally been manufactured to function as a real gun, i.e., a broken or inoperable gun is not a “fake gun”.

The fact that the above semantic entailments naturally emerge when we juxtapose the adjective “fake” with the noun “gun” means that the adjective “fake” actually serves to preserve some properties of guns and negates others. The properties it preserves are perceptual properties (it looks like a gun), motor-activity preoperties (you handle it like a gun), and puposive properties (it serves some of the same purposes as a gun), while the properties it negates are functional (it doesn't shoot bullets) and historical (if it was once a real gun, it can't be a fake).

The fact that “fake” can cause these factors to emerge from the concept “gun” implies that our subconscious conceptualization of the word “gun” contains at least five different semantic properties functioning as an experiential gestalt (i.e., the sum total of the components is readily understood as a single unit, more easily than the understanding of the individual components themselves), which would all be part of what Fillmore calls a frame. In plain terms, when we think of guns, we automatically think of what they look like, how one handles/manipulates them, what they?re used for, how they work, and whether they were ever genuine. What is important to note is that several of these properties are NOT inherent properties of the gun itself, but rather interactional properties, i.e., properties relating to how humans interact with guns.

Frames explain phenomena such as why the words “bachelor” and “spinster” are not full semantic counterparts, even though a classical feature-based theory of semantics would show them as such. Frames allow you to a systematic way of incorporatin cultural attributes into your conlangs.

So when creating words in your conlang, you should be asking yourself, Do I want to automatically carry over all of the frame surrounding this particular word from English (or your native language if not English), or should my conculture/con-inhabitants have a different frame for, what on the surface, looks like merely a translation of this word? So, for example, in your conlang, it might be possible to say that a broken gun is a “fake gun” because the frame is different and does not include the subconscious historical component. Or you might want to add a semantic component to objects like guns such as whether they are considered holy, so that saying they are “fake” would be a blasphemous utterance. You could even extend the idea into the grammar, so that certain classes of objects or the application of certain adjectives can never be spoken of in the past tense because it violates a parameter of the subconscious semantic frame. The possibilities abound.

The KILLING frame provides another example. The components of this frame in English are as follows (excerpted/modified from the FrameNet website):

Core Frame Elements:
CAUSE (of death), e.g., drowning, stabbing, shooting, falling, hitting over the head, etc.
INSTRUMENT [must be tangible entity], e.g., gun, knife, etc.
MEANS/METHOD, e.g., cutting off access to food, pushing off of a cliff, etc.
PERPETRATOR [must be sentient]
VICTIM [must be alive prior to the killing]

Non-Core Frame Elements:
DEGREE [with adjectives describes killing potential; with nouns indicates extent of effect], e.g., that poison is deadly, They're guilty ofmass murder.
DEPICTIVE [the state of the killer or victim during the killing]
MANNER, e.g., quietly, loudly, sloppily, competently, etc.
PLACE [the geographical location where the killing took place]
PURPOSE [the state of affairs the killer is trying to bring about by killing]
REASON [the preexisting state of affairs the killer is responding to]
RESULT [this is often redundant depending on the chosen verb, e.g., I beat him to death vs. * I decapitated him to death]
TIME

Looking at the above frame elements for killings in English, I can already think of two ways I might modify this frame for a conlang. I might add in a frame-element called BODY PART which specifies what part of the anatomy or anatomical system was the target of the instrument or the source of the cause of death. This would be manifested in grammatical sentences which, in English, would have to be awkwardly paraphrased such as “He stomached him (to death)”, meaning “He killed him by attacking him in the stomach (e.g., using a knife to stab him)”. Another example would be a sentence literally translatable as “I throat-killed him”.

The second way I can think of to modify the frame would be to add in an element called JUSTIFICATION. Maybe in your conculture, killing is considered moral in certain social or ritual circumstances and the word-forms used in the sentence carry with them a sense of such moral justification.

Another way to modify the frame for use in your conlang might be to eliminate certain elements that are part of the English frame.

I hope this helps to clarify how frame semantics can come into play when designing a conlang/conculture.

Our next lesson will be on prototypes and prototype theory, followed by a look at Fauconnier's theory of Mental Spaces. However, while we're on the topic of what considerations conlangers should give to all this cognitive stuff when designing a conlang, another example from Lakoff & Johnson regarding spatially-oriented conceptual metaphors is particularly useful for any of you who have constructed or are planning to construct non-humanoid conlangs.

Human beings, by nature of how our bodies are constructed, naturally face in one direction, normally walk/run in one direction (frontward), and stand erect, perpendicular to the horizon and the ground. Because these orientational factors about one's body are recognized and understood in infancy, they are subconsciously ingrained as conceptual reference points which become the basis for many conceptual metaphors. To these are added other aspects of our early experience that form the basis for conceptual metaphors, e.g., we spend most of our time performing actions rather than doing nothing, and we tend to view our own selves as being naturally “good”. Because of all these factors, we tend to conceive of ourselves and others metaphorically as being more UP than DOWN, more FRONT then BACK, more ACTIVE than PASSIVE, and more GOOD than BAD. Similarly we spend our lives with a notion of knowing we exist in the place where we are located, in the temporal present, thus we metaphorically conceive of ourselves as being more HERE than THERE and more NOW than THEN. Thus the “canonical” person (to quote L&J) is UP, FRONT, ACTIVE, GOOD, HERE, and NOW, while a non-canonical (and therefore less desirable or less admired person) is DOWN, BACKWARD, PASSIVE, BAD, THERE and THEN. By metaphorical extension, these properties are applied to things and situations as well. This conceptualization is carried over into the overt syntactic form of language, as seen in sequential ordering of phrases. Thus we say “up and down,” “front and back”, “active and passive”, “good and bad”, “here and there”, and “now and then” while we are very unlikely to say ?”down and up”, ?”back and front”, ?”passive and active”, ?”bad and good”, ?”there and here”, or ?”then and now.” All of these orderings are in turn manifestations of a conceptual metaphor: NEAREST COMES FIRST.

So, for those of you with non-humanoid speakers of your conlangs, especially those with a different body symmetry other than a front-to-back bilateral symmetry, ask yourself how likely or at least arbitrary would it be for the same humanoid-bodily oriented metaphors to arise? Given what we know now from the lessons above, do you think that beings with radial body symmetry and 360-degree vision would actually say “The rock is in front of me”? What would “in front of me” mean to such a being? They'd use completely different orientational metaphors, which by extension would lead to a different set of conceptual metaphors for describing “canonical” things, situations, and people.

Even for those of you with human or humanoid speakers of your conlangs, the sorts of metaphors described above are culture-specific. Do you want your conworld to use the same metaphors for conceptualizing “canonical” events, things, and persons as Indo-European languages? (I suppose if your conlang is supposed to be IE-related you would, but otherwise…)

Evidence from cognitive psychology pioneered by Eleanor Rosch in the early 1970s and continuing to the present shows that human categorization does not function based on criteria as classical set theory (and Chomskian linguistics) would assume, i.e., membership in a category is not considered by the human mind to be an all-or-nothing affair. Rather, human categorization appears to follow the rules of fuzzy logic (also known as fuzzy set theory first formalized by Lofti Zadeh in 1965) in which membership in a category is a gradient phenomenon, where different members of a category display the defining characteristics or attributes of the set to different degrees. Additionally, human categories display prototype effects or centrality, where a particular member of a category may be seen as being the best example of a category, while other members of the category are viewed comparably to the prototype. In many categories, there is no prototype, as no one member of the category shares all (or even most of) the attributes of any of the members of the category; the members of such categories are related by what are called family resemblances and the category itself is known as a radial category.

To illustrate such categories, consider the human category “furniture”. Studies have shown that people tend to identify chairs, tables, and sofas as being prototypical (i.e., best) examples of the category, while clocks, radios, vases and ashtrays are seen as peripheral examples. Oranges and apples are generally given as prototypes for the category “fruit”, while strawberries and dates are less prototypical and tomatoes are seen as being highly atypical members; indeed, the tomato is often seen as also being an atypical/peripheral member of the “vegetable” category, following that fuzzy set theory allows for gradient overlap between two seemingly separate categories.

An example of a radial category would be Wittgenstein?s famous game, where no one member of the category manifests all the attributes found in each member and members at the extreme ends of the category share almost no attributes, e.g., bowling versus Monopoly versus hide-and-seek versus word puzzles such as anagrams.

Large amounts of research in cognitive linguistics show that the concepts of centrality, fuzziness, and family resemblances apply to linguistic categories as well, and at all levels including semantics, syntax, morphology, even phonology. Let's look first at morphology

Consider the English suffix -able, typically assumed to be used with verbs to transform them into an adjective meaning “able to be VERBed”, e.g., “solvable” = “able to be solved?” or “washable” = “able to be washed”. However, it soon becomes apparent that this meaning begins to vary as we apply it to other verbs. A “readable” book does not mean one that is able to be read (presumably ALL books are capable of being read) but rather it is easy or interesting to read. The very fact we can say a book is “very readable” would make no sense if “readable” simply meant “able to be read”. More peripheral meanings for -able emerge when we look at “payable” or “comparable”. A bill in the mail which reads “This bill is now payable” means it is due, not that it is able to be paid. When I ask a friend who is moving from the U.S. to Australia whether housing costs are “comparable” I am not expecting him to answer “Yes, they can be compared”. Why is this? Why the difference in meaning? It goes back to Frames and something called Foregrounding (which I've not previously discussed.. essentially, foregrounding means giving semantic prominence to part of, or a certain aspect of, a situation at the expense of other parts/aspects of the situation based on contextual relevance to the situation). In the case of “comparable” it makes little sense for the meaning of “comparable” to simply mean “able to be compared” as any two or more things in the world can always be compared. Rather, it is the purpose of the act of comparison that is semantically relevant, i.e., what the degree of resemblance is between the two things being compared, as opposed to the act of comparison itself. Similar in principle are “drinkable” where elements from the DRINKING frame determine the contextual/semantic meaning of the -able suffix here, in this case the element of “safely”, which explains why hydrochloric acid is certainly able to be drunk, but it is not “drinkable”.

A non-English example of a radial category in morphology is given by the diminutive in Romance languages such as Italian -ino, -etto, -ello. It can mean physical smallness, e.g., paese –> paesino “small village”; affection, e.g., mamma –> mammina “mommy” (“mummy” to our British contingent); shortened duration, e.g., sinfonia –> sinfonietta “shorter symphony”; reduced amount/strength, e.g., cena –> cenetta “light supper”; reduced scale, e.g., pioggia “rain” –> piogerella “drizzle”; bello “beautiful” –> bellino “pretty/cute”. With verbs, the diminutive suffix -icchiare/-ucchiare implies a process of intermittent or poor quality, e.g., dormire “sleep” –> dormicchiare “snooze”, lavorare “work” –> lavoricchiare “work half-heartedly”. Again, Frame and foregrounding operate to determine the contextual meaning of the diminutive. While it certainly feasible/possible for a language to have its morphological diminutive applied to the stem for “mother” to mean “small mother (in size)”, it is hardly relevant to most persons what size their mother is, whereas a convenient way of morphologically mapping the emotional bond a person feels as a small child toward his/her mother is more contextually salient.

The classic notion of “parts of speech” in which classes of words such as nouns and verbs are assumed to display all aspects of their set-defining attributes/behaviors is easily shown by cognitive linguistics to be rubbish. Examples abound of nouns that are nounier than others, as well as verbier verbs. For example, note that the agentive nominalisation of transitive verbs works in some cases and not in others:

1a) John imports rugs. –> John is an importer of rugs.
1b) John knew that fact. –> *John was a knower of that fact.

Our -able suffix likewise implicates some verbs as being less verby than others:

2a) That apple can be eaten. –> That apple is edible.
2b) That lighthouse can be spotted. –> *That lighthouse is spottable.

As does passivization:

3a) Sam kicked the ball. –> The ball was kicked by Sam.
3b) Sam owes two dollars. –> *Two dollars are owed by Sam.

The double-raising transformation for nouns works for some nouns but not others:

4a) It is likely to be shown that Bill has cheated. –> John is likely to be shown to have cheated.
4b) It is likely to be shown that no headway has been made. –> *No headway is likely to be shown to have been made.

The ability to use tag questions also implicates some nouns as nounier than others:

5a) Some headway has been made. –> Some headway has been made, hasn?t it?
5b) Little heed was paid to her. –> * Little heed was paid to her, was it?

Lakoff has shown that languages utilize various adverbs and adverbial constructions termed “hedges” to deliberately specify what are otherwise covert fuzzy relationships within lexico-semantic categories. Hedges in English include phrases such as “technically speaking”, “loosely speaking”, “strictly speaking”, “par excellence”, etc. Note how their use reveals centrality/prototype relations among various words in the following sentences:

6a) A sparrow is a bird par excellence.
6b) ??A chicken is a bird par excellence.

7a) Loosely speaking, an ashtray is a piece of furniture.
7b) ??Loosely speaking, an armchair is a piece of furniture.

8a) Strictly speaking, a tomato is a fruit.
8b) ??Strictly speaking, an apple is a fruit.

Research done in 1984 by Jaeger and Ohala showed that linguistically untrained laypersons psychologically perceive certain groups of consonants to be more voiced and less voiced than others on a gradient spectrum, rather than the either-or +voice/-voice that traditional phonetics describes. Specifically, the phonemes in question were grouped as follows:

VOICED CONSONANTS
Most voiced: /r, m, n/
Less voiced: /v, D, z/
Even less voiced: /w, j/
least voiced: /b, d, g/

VOICELESS CONSONANTS:
Less voiceless: /f, T, s, h, S/
Most voiceless: /p, t, k/

Starting with the psychologically-derived voiceless-voiced gradient spectrum listed immediately above, I see a definite possibilty for a totally new sort of six-level consonant harmony scheme analogous to the four-level vowel harmony scheme found in Turkish. Ideas for patterns of consonant mutation are implicit in the gradient scheme as well.

On the lexical scheme, you need to consider your lexical categories as to which members represent the prototypes of a given category or whether the category will instead be a radial category without a prototype. If radial category, what will be the common thread binding the category together in the absence of universally shared attributes? Every language is different here, so watch out that you don?t make yours like English (or whatever your native language).

How nouny should a particular noun be in your language, and how verby each verb? Should each noun be allowed to operate grammatically in all possible grammatical constructions? Not very likely if your conlang is supposed to be a natlang-style, realistic language. So how to decide whether a particular noun/verb should be permissible in a given construction? Think about its Frame and contextual usage, i.e., what situations it needs to be able to refer to in the real world. Maybe in your conlang, “headway” is something that is likely to be shown to have been made, “heed” is something to which a tag question can apply, and two dollars is something that CAN be owed by someone. Or, maybe you might want to go the opposite direction, so that fewer nouns/verbs are permissible in a given construction. The thing to be aware of is why it is so or why it isn't so from a standpoint of Frame, Construal, Foregrounding, Image schema, etc.

On the morphological front, what sort of categorization scheme for your affixes should you come up with? Should diminutives convey size, affection, scale, shortened duration, etc. as in Romance languages? Or do you want to eliminate one or two of these from your diminutive, split these among two or three different affixes, or add additional semantic distinctions beyond what is found in Romance languages, e.g., relevance, e.g., “person” + diminutive = “irrelevant person” or connotations of duplicity or mendacity, e.g., “speak” + diminutive = “to lie; tell falsehoods”?

Remember the “If I were you I'd hate me/myself” problem way back at the beginning of this thread? Well, it's time to analyze what's going on. Consider the following sentences:

1a) *The girl with blue eyes has green eyes.
1b) In that photo, the girl with blue eyes has green eyes.

2a) *I'm taller than I am.
2b) John thinks I'm taller than I am.

Sentence (1a) and (2a), although syntactically well-formed, are unacceptable because they are semantically anomalous in that they contain contradictions. According to formalist theories of grammar (e.g., Chomsky's), sentences (1b) and (2b) should likewise be unacceptable. But instead, they are acceptable. How can this be?

Let's look at our curious example from early on:

3a) *I would hate me.
3b) If I were you I'd hate me.
3c) If I were you I'd hate myself.

Sentence (3a) is unacceptable because it is grammatically ill-formed (the co-referential pronoun at the end of the sentence should be “myself”). Chomsky and other formalist theories have no simple explanation as to how this ungrammatical sentence can magically transform itself into being grammatical when preceded by the adverbial if-phrase in sentence (3b). The situation becomes more bizarre when we see that the sentence is distinct in meaning from sentence (3c), and that the co-referenced party within each sentence is different for each sentence, i.e., in (3b) “me” refers to the speaker, but in (3c) “myself” refers to the listener.

Cognitive linguistics offers elegant answers to the above puzzlers via Gilles Fauconnier's theory of Mental Spaces. To understand the idea of mental spaces, it is perhaps easiest to first look at the simpler linguistic phenomenon of metonymy.

We discussed metonymy briefly earlier in this thread, but let?s look at it more closely. Consider the following sentences:

6) You'll find Hemingway on the top shelf.
7) Van Gogh fetches enormous prices these days.
8) The White House has announced that all Iraqis are happy.
9) The ham-and-cheese wants a refill on his coffee.

The above four sentences all utilize substitution of one concept in place of another. Sentence (6) does not mean Ernest Hemingway, the person, is sitting on the top shelf, but rather a book or books written by him. Likewise, in sentence (7), Van Gogh the person is not being sold into slavery, and in sentences (8 ) and (9) it is neither a white-colored building whose address is 1600 Pennsylvania Avenue in Washington, D.C., nor a sandwich that has magically learned to talk. Metonymy shows that human beings have the cognitive ability to refer to a concept by substituting a different concept as long as the substitute concept is conceptually linked in some way, e.g., creator-for-creation, place-for-person, consumable-for-consumer, etc.

The entity that is the normal referent for a metonymic construction is termed the “trigger” while the entity to which the predicate applies is termed the “target”, so that in sentence (6), Hemingway the person/author is the trigger and the book by him is the target. Metonymy is yet another example of how important the knowledge of frames are to understanding language usage. Consider what the above sentences would mean to someone who had never heard of Hemingway, Van Gogh, the White House, or a ham-and-cheese sandwich.

At any rate, the important lesson of metonymy to apply to the theory of mental spaces is the notion of trigger and target.

Examine the pair of sentences above in (1a/b) again. Sentence (1b) becomes grammatical because the phrase “in that photo” serves to create a hypothetical reality, which functions as a parallel mental space, allowing us to divide up the entity named in the sentence into trigger and target. The trigger (the entity who exists in our reality) is “the girl with blue eyes”, while the target is the figure pictured in the photo, an entity to which the predicate “has green eyes” applies. The reason why “the girl with blue eyes” can be used to refer to the target (the figure in the photo) is that there is a pragmatic function linking that figure to a real person, i.e., the fact that the figure in the photo is a representation of the girl with blue eyes in the real world. The sentence relies on an understanding common to all human beings that an image can function as a representation of something in the real world. The relationship between the trigger (reality) space and the target (photo) space can be diagrammed.

Based on the above analysis, the phrase “in that photo” is termed a space builder because it serves to create an alternative, hypothetical mental space (i.e., hypothetical parallel reality) in which elements from reality can be mapped in a one-to-one correspondence with different but conceptually linked elements in the mental space. The creation of this alternative mental space by the phrase “in that photo” allows for a girl with blue eyes and a girl with green eyes to co-exist and yet be one-and-the same entity, explaining why sentence (1b) can be semantically acceptable while sentence (1a) is not (because it has no space-building word or phrase to create the hypothetical parallel reality).

Similarly, the phrase “John thinks” is a space builder in sentence (2b), creating a split between the trigger (reality) space where I am, and the target (hypothetical) space where John's beliefs reside. In the reality space, I have my actual height (h), while in the target space my height is greater than that (H). Once again, the space-builder phrase allows the two halves of the contradiction to exist separately in alternate spaces, one real, the other hypothetical, allowing the sentence to be semantically acceptable while its simpler, single space version (2a) is not.

As for our pesky trio of sentences (3a,b,c), the if-phrase is the space-builder which allows us to explain the distinction between (3a) and (3b), similar to the examples above. But how to explain the distinction between (3b) and (3c)? Here, we need to bring in our old friend, conceptual metaphor, to help us. In English (and presumably other languages), a conceptual metaphor exists which separates an individual human being into what has been termed the “Subject” versus the “Self”. The Subject is essentially the inalienable seat of our rational/moral judgments, while the Self is the quasi-alienable part of ourselves that interacts directly with the world. Evidence for this Subject-versus-Self conceptual metaphor is seen in sentences such as I couldn't stop myself, You got carried away, He's a very together person, She can't help herself when it comes to chocolate. Given this conceptual split between Subject and Self, we can now diagram these sentences. First sentence (3b):

The diagram above shows “If I were you I'd hate myself” which is uttered in the situation where I strongly disapprove of the way you behave. Here, the if-phrase sets up a mental space (or hypothetical world, if you prefer) with counterparts of both I and you except that the counterpart of you contains my Subject instead of yours. This means that in the mental space, your bad behavior (as judged by my Subject occupying your Self) would cause you to experience self-hatred. Since the hatred is self-directed (it is experienced by the constructed you against your Self), it is appropriate to use a reflexive pronoun as per the normal grammatical rule of co-referentiality.

On the other hand, “If I were you I'd hate me” is uttered when I disapprove of my own actions (even if you apparently do not disapprove), so that the constructed you has grounds for hating me (the speaker). Here, the hatred is not self-directed, thus there is no co-reference and so the appropriate accusative pronoun me is correctly applied per the normal grammatical rule.

Mental space theory can also easily explain sentences involving referential ambiguities such as:

10) Jean thinks she wants to marry a Norwegian.

This sentence has three different meanings. It can mean that there is a specific Norwegian whom Jean believes she has a desire to marry, or it can mean Jean believes she has a desire to marry some Norwegian (but she hasn't found any eligible ones yet), or it can mean she wants to marry to a specific person she believes to be Norwegian but in fact he isn't Norwegian at all. The mental space diagrams for all three interpretations are quite distinct from one another (trust me, cause I'm too lazy at the moment to draw them).

Interesting enough, some languages (e.g., the Romance languages) are able to grammatically distinguish between some of these types of sentences using the indicative versus subjunctive mood distinction. For example in French:

11a) Jeanne veut ?pouser quelqu'un qui est Norv?gien.
11b) Jeanne veut ?pouser quelqu'un qui soit Norv?gien.

Sentence (11a) implies a particular, identified Norwegian exists whom she has in mind, whereas sentence (11b) implies that no particular Norwegian has been identified and she may never find one for all we know.

This should be giving you ideas for your conlang as to just how or whether your conlang should be grammatically identifying contexts involving hypothetical mental spaces. Should you use different moods on the verbs like the Romance languages or leave it all ambiguous like English? How about a different verbal voice? Or maybe suffixes on the noun participants to indicate they are hypothetical representations of real-world counterparts. Perhaps even different counterpart lexemes altogether!? (As for Ithkuil, it has an entire morphological category called “Essence” that identifies such hypothetical/representational contexts.)

Well, I think I am going to bow out now regarding continuing these lessons. Needless to say, there are lots of other fascinating aspects of cognitive linguistics which we have not discussed (and we have only scratched the surface of the topics that have been mentioned), but unfortunately I am burning out on preparing them, and I want to get back to work on the Ithkuil lexicon and the Il?ksh website. So, for those of you who have found this thread interesting, I leave you to take it from here on doing your own research, study, etc. on the topic.