Semantics - The Study of Word and Sentence Meaning


meaning.gif

 

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


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


‘The question is,’ said Humpty Dumpty, ‘which is to be master – that’s all.’

Meaning is at once the most obvious and most mysterious feature of human language. We can’t claim to know a language without knowing the meanings of words and yet more than 3,000 years of speculation by philosophers and linguists has failed to provide a theory of meaning. In this section we investigate whether Alice or Humpty Dumpty has a better theory of meaning. We devise several semantic tests for different theories of meaning. We end with some examples of a linguistic approach to the study of meaning.


Alice v. Humpty

What are the main features of Alice’s theory of word meaning?

What are the main features of Humpty’s theory of word meaning?


Reference Theories of Word Meaning


 

s-chair.jpg

 

cv_carb.jpg

 

Alice’s main concern is with the relation between meaning and reference. Ask anyone about the meaning of a common noun such as chair or carburetor and chances are they will point to an instance of such an object if it is handy. This type of ostensive definition satisfies our intuitions about what words mean by pointing to something the word denotes, and has prominent advocates such as John Stuart Mill. Does reference provide a valid theory of meaning?


Reference theories of meaning define the meaning of words or phrases in terms of the things (or actions, qualities, relations, etc.) that the word or phrase denotes. If we ask someone for a cup and they hand us a sponge, we are apt to think they did not understand the word cup. More precisely, we can equate the meaning of a noun with the set of things that are that object.


      CUP = the set of things that are cups

      SPONGE = the set of things that are sponges


Verbs, other words and even sentences can be defined in terms of their reference sets as well.


      RUN = the set of things that run

      BARK = the set of things that bark


      JOHN RUNS = true or false

      MARY BARKS = true or false


Reference theories of meaning define meaning in terms of the semantic extension of a word or phrase. The semantic extension of a word or phrase is the set of things, actions, etc. that define the reference or the word or phrase. The semantic extension of a declarative sentence is its truth value.


Just to keep things straight, I will put a word in italics when I mention it and will use capital letters to refer to the meaning of a word. So CUP stands for the meaning of the word cup. A cup is the thing referred to by the English word cup.


Testing Reference Theories of Meaning


Problem 1: Words without a Semantic Extension


A reference theory of meaning accounts for our ability to point to the things that words denote, but it has critical shortcomings. Not all words or phrases have a semantic extension in the real world, e.g. Hobbit, witch, phlogiston. The meaning of verbs would change continuously as, for example, some organisms stop running and others start running. We are investigating whether the word meaning has a semantic extension. Sentences create other contexts in which words are used in non-referential expressions.


      No dog misses a treat.

      Where did you find your cat?


How do such words and sentences invalidate reference as a theory of meaning?


Problem 2: Semantic Intension


The philosopher Gottlob Frege pointed out a critical flaw using the phrases morning star (Phosphorus) and evening star (Hesperus). These phrases have the same referent, but different meanings. Using them in a sentence makes this difference obvious.


      The morning star is the evening star.

      The morning star is the morning star.


The first of these sentences tells us something new, while the second sentence does not. A reference theory of meaning does not predict this difference. Why not?


Frege distinguished between a word’s semantic extension (reference) and its semantic intension (sense). The semantic extension of a word is the set of things the word denotes, while its semantic intension is the concept or meaning of the word. The meaning of a word determines the things it refers to, but it cannot be equated with its reference.


Problem 3: Opaque Contexts


Sentence meaning displays a similar distinction between extensional and intensional meanings.


      George is the best student in linguistics.

      I dreamed that George is the best student in linguistics.


The first sentence asserts that George is one and the same shining paragon of linguistics. In other words, it asserts that George and the best student in linguistics have the same semantic extension. If George is the best student in linguistics then the first sentence is true. Assuming the first sentence is true, however, does not guarantee the truth of the second sentence. The truth of the second sentence depends on what I dreamed and not on the truth of George being the best student in linguistics. We can usually equate the semantic extension of a sentence with its truth value, but we see that the truth of the embedded clause has no effect on the truth value of the second sentence. Propositional attitude verbs (dream, believe, want) create opaque contexts where the truth value of the embedded clause is unrelated to the truth of the whole sentence. We need something more than the reference of the embedded clause to understand meaning in opaque contexts.


Problem 4: Referential Change


The things that words refer to seem to be changing constantly. A good example of this change is the name of a town, e.g., Lawrence. Lawrence has been continuously expanding since its beginning. It was burnt to the ground once and rose from its ashes. Individual buildings in the town are constantly changing shape and color. The entity denoted by the name Lawrence is not the same from one day to the next. A strict interpretation of a reference theory of meaning would predict that the meanings of most names is constantly changing.


The philosopher Hilary Putnam pointed to an interesting case of semantic change in scientific theories. One of the major advances in physics occurred when Newton equated momentum with the product of mass and velocity. We say that Newton defined momentum as mass times velocity. This equation held true until Einstein predicted that it would break down for objects at relativistic speeds. Einstein redefined momentum by adding a relativistic adjustment to Newton’s original equation. Intuitively, we feel that Newton and Einstein were talking about the same concept, but a strict referential theory of meaning would claim that they were talking about different things.


Problem 5: Semantic Expertise


Putnam alleges that many people cannot pick out the referents for all the words they use. He claims that he cannot tell the difference between beech trees and elms even though he has used the words beech and elm for most of his life. A referential theory of meaning suggests that anyone would know the difference if they knew the words beech and elm. How would you define the difference between the meanings of the words walk, amble, shuffle, saunter, stroll, slouch, slink, slip and mosey?


Assessment


We have identified five significant problems with reference theories of meaning. These problems supply significant information about the nature of meaning and show that meaning is something more than reference. The text contrasts reference with sense, but never defines what sense is. One way to think of sense is as a mode of presentation or a small story. Two phrases with the same reference such as Barack Obama and the 44th president of the United States provide different presentations or stories about the same referent. In cases like Hesperus and Phosphorus, we may not know that the perspectives have the same referent. In cases like unicorns and phoenixes, we may have perspectives without real referents. In cases like elms and lychees, we have perspectives with vague referents.


These problems provide tests that any theory of meaning should meet. Any theory of meaning that depends on reference will have the same limitations. We investigate some other forms of referential theories next to see if they can respond to the five problems we discussed.


Other Reference Theories of Meaning


To this point we have evaluated a simple reference theory that equates meaning with semantic extensions in the real world. Other reference theories equate meaning with other types of semantic extensions. The following sections discuss a few of these other reference theories.


1. Mental Images


If Frege is correct then meaning is something more than reference. The philosopher John Locke proposed equating meaning with the pictures of objects in our mind. While this idea has a certain appeal, it also suffers from a number of problems.


Problem 1: Internal Reference


If we use mental images of objects as the basis for meaning then we are equating meaning with mental referents rather than external referents. To the extent that our mental images for morning star are similar to evening star then we just have a mental image version of a referential theory of meaning. To the extent that our mental images are different for these two concepts, we would need to add a new component to the mental images to explain why these phrases have the same referent.


Problem 2: Different Images


A mental image theory cannot insure that speakers of the same language carry the same mental image for any given concept. To the extent that one speaker’s mental image of a grandmother is different from that of another speaker, the theory cannot explain our ability to communicate via language.


A mental image theory predicts the possibility that every speaker has their own private language. The philosopher Wittgenstein pointed out that it would be impossible to prove that someone had a private language to the extent that it was private.


2. Prototypes


Wittgenstein offered his own version of a mental image theory built around prototypes. The idea is that we only require a family resemblance between objects to consider them the same. Wittgenstein pointed out that words like game refer to many different types of contests which lack any common features across their whole range of use. A game may involve multiple players or just one. The players may play strictly for enjoyment or profit. We recognize what counts as a game because it has one or more features of a game. Prototypical games have most of what we think of as game features.


Problem 1: Prototype Reference


Prototype theories of meaning are based on reference to a prototype. Prototype theories encounter all the problems that we discussed for reference theories. How would you apply Frege’s Morning Star/Evening Star critique to a prototype theory of reference?


Problem 2: Non-prototypical Examples


Although many experiments suggest that we recognize prototypical members of a category faster than we recognize non-prototypical members, and recall prototypical members faster as well, we still include non-prototypical members in every semantic category. The set of dogs includes Chihuahuas and Great Danes in addition to Labradors and retrievers. All prime numbers are prime numbers even though 3 and 7 may be prototypical primes. Prototype theory does not explain how we draw the boundaries between different concepts rather than just recognizing the most typical members.


Problem 3: Prototypical Features


Prototype theories typically rely on a list of features that speakers use to define the prototype for any concept. Prototypical features for a bird, for example, include a beak, wings, flight and feathers. We recognize a prototypical bird to the extent that it has most of the prototypical features. This process invites the question of how we recognize the prototypical features of birds. They would be features that we observe on prototypical birds. We then have a circular definition that relies on prototypical features to define the prototype, but also relies on the prototype to define its prototypical features.


Problem 4: Combining Features


A theory of meaning has to predict how the meanings of individual words combine to produce the meaning of a phrase. Prototype theories of meaning lack the ability to predict how to combine the meanings of words. For example, prototypes for the word pet would include dogs and cats. Prototypes for the word fish would include salmon and trout. But these prototypes do not predict the prototype for the phrase pet fish.


3. Semantic Features


If mental images do not supply the critical distinctions necessary for meaning another possibility would be that humans rely on a set of innate semantic features to construct meaning. The philosopher, Jerry Fodor, maintains this Language of Thought is the only explanation of our ability to communicate ideas. The innate semantic features would be akin to a table of atomic elements. Once we define each semantic element, we will be able to explore how they combine to produce meaning. Needless to say, the theory of innate semantic features also runs into reference difficulties.


Problem 1: Feature Reference


Semantic Feature theories must explain how speakers fix the reference of each feature. Lyons (1973) pointed out that semantic features never seem to provide enough power to explain word meaning. A theory that tells us the meaning of mare has the feature ANIMAL won’t take us very far if the theory doesn’t supply a meaning for ANIMAL. If the feature theory uses reference to fix the meaning for ANIMAL, it is just another type of reference theory with an intermediate mental vocabulary. We might assume, with Fodor, that ANIMAL refers to a mental concept or brain state. Does this version of feature theory evade Frege’s problem?


Problem 2: Feature Arbitrariness


Semantic feature theories have been criticized for their arbitrary nature (Burling, American Anthropologist 1964 ‘God’s truth or hocus pocus?’). Does the meaning of the word man contain a semantic feature [+MALE] or the feature [-FEMALE]. Either feature would allow you to distinguish the meanings of the words man and woman, but there is no reason to prefer [+MALE] over [-FEMALE].


Mathematicians have devised various definitions for the concept NUMBER. Dedikine proposed a ‘cut’ in the number line, while Russell & White proposed a definition using set theory. Both definitions are equally valid, but we have no reason to prefer one over the other.


Usage Based Theories of Meaning


We investigated several types of reference theories of meaning of the sort Alice would approve and found that they all face difficulties discussed by Frege, Putnum, Quine, Fodor and other philosophers. The basic problem is that reference-based semantic theories do not account for intensional meaning. Intensional meaning is based on the stories that we use to make sense of words. As Humpty Dumpty claims, our stories give us mastery over word meanings.


Frege offered a Humpty Dumpty or usage theory of meaning as an alternative to reference theories. A usage theory of meaning equates meaning with the ways that words are used. Dictionaries commonly employ a usage approach in their definitions of word meaning. Linguists and philosophers sneer at dictionary definitions, but they have yet to offer a viable alternative. Linguists have much to learn from exploring the practical approach that lexicographers use to construct definitions. We will explore some criticisms of dictionary definitions before looking at how a usage based theory meets the tests we used for reference theories.


Problem 1: Dictionary Definitions Use Words in Definitions for Other Words


A common complaint about dictionary definitions is that you have to know the meanings of the words the dictionary uses before you can understand the meaning of the word you are looking up. The American Heritage College Dictionary provides the following definitions:


opossum

1. Any of various nocturnal, usually arboreal marsupials of the family Didelphidae, ... of the Western Hemisphere, having a thick coat of hair, a long snout, and a long prehensile tail.

2. Any of several similar marsupials of Australia belonging to the family Phalangeridae.


diapir

An anticlinal fold in which a mobile core, such as gypsum, has pierced through the more brittle overlaying rock.

 

opossum.jpg

 

diapir.gif

 


The philosopher Willard Van Orman Quine proposed a dictionary type theory of meaning as a basis for his thesis of semantic holism. Semantic holism assumes that the meaning of every word depends on the meanings of other words, tying word meanings into a semantic net. The more stories that we learn, the more we know about the meaning of each word. A change in our understanding of a word will affect our understanding of the stories we tell. Quine’s semantic theory reflects his view of science where a single discovery can radically transform our understanding of everything.


A usage theory provides a dynamic theory of meaning. It recognizes that we do not know everything (a state Putnam labels logical omniscience). New discoveries lead to changes in our stories. We need to discover that Phosphoros and Hesperus refer to the same planet. A usage theory explains why we can use words such as elm or beech the way other people use them without being able to identify their exact referents. A usage based theory ties meaning to a linguistic community rather than to the mind of individual speakers.


Problem 2: Dictionary Definitions Include Function Words


Dictionary definitions cannot avoid using common function words such as the, of or to in their definitions. This practice seems to create a problem of accounting for our understanding of these words. Function words actually provide strong evidence in support of usage theories since these words lack obvious referents and differ considerably in use from language to language. Some languages lack articles or prepositions altogether. Quine observed that the meaning of articles like the depends on whether languages also have plural markers or noun classifiers.


Linguistic Semantics


It is possible to use linguistic methods to investigate semantics rather than philosophical arguments. The primary method linguists can deploy is to describe the semantic distinctions speakers use their words to make akin to the phonemic distinctions that speakers observe. The main difficulty is to discover a language-neutral way to describe a semantic domain. The meanings of English words are different from the meanings of words in other languages.


Berlin & Kay’s (1969) work Basic Color Terms provide a classic model of the linguistic approach. The color domain seems easy to describe. The visible spectrum is contains an infinite number of different wavelengths that languages typically divide into a more manageable finite number of basic color terms. Linguists distinguish between basic color terms such as green and red from nonbasic color words such as burnt umber and olive drab. Basic color words are known by every speaker and can be easily identified. They are not restricted to certain domains (as are the English words brunette and sorel), they are not included in the range of another term (as are the English words lemon and scarlet), and their meaning cannot be derived from other basic color terms (bluish).


Compare how the basic color terms of English, Shona and Bassa divide the basic color space:

color.jpg











English: red    orange       yellow       green               blue                                   purple

Shona:       cipswuka         cicena       citema                         cipswuka

Bassa:                               ziza                       hui


Berlin & Kay (1969 Basic Color Terms) discovered that although languages have different numbers of basic color terms, they add terms in a definite progression:

 

      light                                  green

      dark          <   red      <   yellow       <   blue     <   brown


They also found that the focal color for each term was very similar across languages.


Space


Spatial language is another domain that can be explored linguistically. The expression of spatial location would appear to be one of the most fundamental semantic notions common to all humans. After all, we seem to have a good idea of where we are most of the time. Spatial language turns out to be anything but a simple reflection of an objective reality. Recent cross-linguistic research has revealed a surprising degree of variation exists in spatial expressions between languages.


Languages locate objects in relation to some reference point. Herskovits (1986) refers to the located entity as the located object. She calls the noun phrase that specifies the location the reference object. Talmy (1983) refers to these notions as figure and ground, respectively. Spatial locations are essentially relations between the located object and a reference object:


1.   The cup is on the table.

      Located     Reference

      Object       Object

 

      Figure       Ground

meaning1.gif

 


Talmy ([1983] 2000: 184) figure and ground as:


“The Figure is a moving or conceptually movable entity whose site, path, or orientation is conceived as a variable the particular value of which is the relevant issue.”


“The Ground is a reference entity, one that has a stationary setting relative to a reference frame, which respect to which the Figure’s site, path, or orientation is characterized.”


Talmy introduces the notion of relevance into his definition of figure since the speaker’s perspective determines what is picked out at the figure. In many cases, either object in a spatial relation can be picked out as the figure:


2a. The lamp is over the table.


  b. The table is under the lamp.

meaning2.gif

 



The Geometry of Linguistic Space


Objects can be located through topological or projective relations. Topological locations are invariant with respect to changes in the reference object. The three basic topological locations are proximity, containment, and exteriority. They are topological because they remain the same no matter what angle the located object is viewed from. Projective locations change with the perspective of the speaker. Two basic projective locations are left and right.


Containment denotes the inclusion of a located object in the reference object. Containment may be: i. either partial or total, ii. apply in any dimension, and iii. be either real or virtual:


      The books are in the box.

      There’s a crease in the bedspread.

      What do you have in mind?


The books may be completely or only partially contained by the box. The box may also be opened on its top, or on its side. What matters is whether the books are in contact with the box’s interior in some fashion. Two-dimensional objects such as paper and blankets have interiors within their boundaries. Their interiors are simultaneously potential supports. What counts as an inherent part of a located object is not obvious:


      There is a blemish on your skin.

I found a scratch on my car.


Cross-linguistic Encoding of Topological Relations


Melissa Bowerman pioneered the cross-linguistic study of topological encoding. She initially looked at the encoding of in and on in English, Berber, Dutch and Spanish.

 

meaning3.gif












 

ole.gif












 

meaning4.gif












 

meaning5.gif














Functional Relations


I. Nyoman Aryawibawa (2008) explored the semantics of spatial relations in Rongga, Balinese, and Indonesian. These languages employ unmarked prepositions to express normal relations between objects and marked prepositions to express abnormal relations.



meaning6.gif  

meaning7.gif  

meaning8.gif  

meaning9.gif  


Rongga

  

Kain meja one meja

Li’e munde one mok

 

cloth table on table

that orange in bowl

 

‘The table cloth is on the table’

‘The orange is in the bowl’


Changing to an abnormal relation between the figure and ground results in the use of a marked preposition. The marked form is used if the table cloth is folded and then put back on the table or if a ribbon is put in the bowl instead of an orange.


  

Kain meja zheta wewo meja

Pita zhale one mok

 

cloth table on table

ribbon in bowl

 

‘The table cloth is on the table’

‘The ribbon is in the bowl’


Aryawibawa contrasts Levinson et al’s topological categories with those of Rongga, Balinese, and Indonesian:

 

meaning10.gif  

meaning11.gif  

meaning12.gif  

meaning13.gif  

Levinson et al:

attachment

superadjacency

full containment

subadjacency

 

meaning14.gif meaning15.gif meaning16.gif  

meaning17.gif  

Aryawibawa:

function

location


Projective Locations


Projective locations depend upon the viewer’s perspective or properties of the reference object. A ball could be in front of a desk from one perspective, beside the desk from another, and behind the desk from a third perspective.

 

space7.gif

 

Some objects have conventional features that speakers treat as fronts or backs. The back of a chair provides back support. The front of a house is marked by the main entry way. Objects may be located in relation to the fronts, backs or sides of reference objects. Speakers must decide whether to use their own perspective or the features of reference objects to encode projective locations.


Front


The ‘front’ of an object depends on properties of the reference object and/or the perspective of the viewer. Languages chose different vantage points to determine a viewer’s perspective. English uses the side of the reference object facing the viewer as the ‘front’. Hausa uses the opposite side of the reference object, the side facing away from the viewer as the ‘front’ (Hill 1974, 1982). In the situation shown below, an English speaker would say the spoon is in front of the pumpkin. A Hausa speaker would say (Hill 1982:21):


            Ga cokali can baya da k’warya.

            look spoon there back with pumpkin

            ‘There’s the spoon behind the pumpkin’

 





Viewer’s


          ————>


Perspective

spoon.jpg

 

pumpkin.gif

 

There is more variation between languages in the attribution of ‘front’ and ‘back’ to inanimate objects. The front of vehicles is the part facing the forward motion. Even though ships have bows rather than fronts, we can still locate an object in front of a ship. The fronts of chairs and houses are determined by the convention point of access. This convention applies to appliances such as tvs and radios whose access point – the ‘on’ button, defines the front.


Other languages impose an anthropomorphic image on inanimate objects to a much greater extent. Languages that make use of relational nouns rather than adpositions are especially prone to attribute fronts to things. The Mayan language K’iche’ uses a single preposition in combination with a set of relational nouns to express spatial location (Kaufman 1990:77):

 

Prep.

Rel. Noun

English Equivalent

 

chi +

‘at, on’

xee’ ‘root’

chii’ ‘mouth’

wi’ ‘head’, ‘hair’

iij ‘back’

wach ‘face’

‘below’, ‘beneath’

‘next to’

‘above’, ‘on top of’

‘behind’, ‘outside’

‘in front of’


K’iche’ and English speakers agree on the features for many objects. Both would define the ‘front’ (English)/‘face’ (K’iche’) by reference to the front door (u:-chii’ jah ‘its-mouth house’ in K’iche’). Both would agree on a located object being ‘beneath’(English)/‘root’ (K’iche’) a chair or table. But, K’iche’ speakers extend their relational nouns in ways that English speakers cannot predict. Rather than saying something is ‘on’ the ground, K’iche’ speakers say it is on the face of the ground (chi+u:-wach uleu). Rather than saying something is on top of a house, K’iche’ speakers say it is on the back of the house (chi+r-iij lee jah).


Other projected locations include back/behind, below, above, and left/right.


Absolutive Systems


Levinson (1996) describes two other types of frames of reference: intrinsic and absolutive.

Language like English use a relative system of spatial reckoning for projected locations. A relative system is based on the speaker’s or hearer’s perspective or the features of the reference object. These projections may be either relative or intrinsic.


 

meaning18.gif

 

Levinson (1996) provides evidence that speakers of other languages employ an absolutive system of spatial reckoning. An absolutive system is based on a fixed set of spatial coordinates, eg., north, south.

 

meaning19.gif

 

Tzeltal Mayan has an Absolutive system of spatial reckoning based on the predominant uphill/downhill lay of the land (along a South (uphill)-North (downhill) axis):

 

 

Verb

Position

Relation

 

UP

DOWN

mo ‘ascend’

ko ‘descend’

kaj ‘be above’

pek’ ‘be low down’

ajk’ol ‘uphill’

alan ‘downhill’

            e.g., ‘The rain is descending’ (i.e., coming from the south)

                          ‘It (a puzzle piece) goes in downhillwards’


Tzeltal children begin to use the Absolute vocabulary in the one- and two-word stage to refer to vertical relations (with verbs of falling and climbing) as well as horizontal relations (movement between houses or of toy cars on the flat patio) (Brown 2001).


The core semantics for the children’s verbs mo/ko are restricted at first to local places (particular houses in the local compound). The children generalize these verbs to novel contexts such as moving objects into trees, onto beds, up onto the roof, and to and from particular houses.


Marquesan, a language of French Polynesia, has speakers that live almost exclusively on islands. As a seafaring people, the Marquesan speakers use the directionals “Seaward,” “inland,” and “Across” (or to:place name). The interesting distinction between this spatial reference system and that of Tzeltal is the while in Tzeltal “Uphill” will always denote one direction, “seaward,” though still an absolute spatial reference (as the sea isn’t going anywhere), can mean a different direction based on where you are on the island. The semantic implications of this are fascinating, in that a speaker must constantly know exactly where they are with respect to the ocean in order to have any frame of reference for where anything else is. Marquesan speakers use this system in both large scale and small scale directional referencing.


“For a speaker of Marquesan it is not unusual to say that the plate on the table is inland of the glass or to localize a crumb on another person’s cheek as being on the seaward or inland cheek.” (Cablitz p.41)


Semantic Fieldwork


The basic procedure for semantic investigation is to pick a semantic domain and investigate the words that a language uses to mark semantic contrasts within the domain. One early paper that illustrates this approach is Labov’s study of cups (‘The Boundaries of Words and Their Meanings’, in R. Fasold, ed. 1973). Labov explored the semantic boundaries of the words cup, mug, bowl and vase. He provided subjects with line drawings of containers that varied in width and height.

 

cups.jpg

 

He also presented the drawings to his subjects in two contexts. In the first, or ‘neutral’ context, he merely presented the drawings. In the second, or ‘food’ context he presented the drawings and said they contained rice or mashed potatoes.

 

cups7.jpg

 


References

 

Aryawibawa, I Nyoman. 2008. Semantic Typology: Semantics of Locative Relations in Rongga (ISO 639-3: ROR). Masters Thesis. University of Kansas

Bowerman, Melissa. 1999. Learning How to Structure Space for Language: A Crosslinguistic Perspective. In P. Bloom, M. A. Peterson, L. Nadel and M. F. Garrett (eds.), Language and Space, 385-436. Cambridge, MA: MIT Press.

Brown, Penelope. 1994. The INs and ONs of Tzeltal locative expressions: The semantics of stative descriptions of location. Linguistics 32.4/5.743-790.

Brown, Penelope. 2001. Learning to talk about motion UP and DOWN in Tzeltal: is there a language-specific bias for verb learning? In M. Bowerman and S. C. Levinson (Eds.), Language Acquisition and Conceptual Development, 512-543. Cambridge: Cambridge University Press.

Cablitz, Gabriele H. 2002. The acquisition of an absolute system: learning to talk about SPACE in Marquesan (Oceanic, French Polynesia). The Proceedings of the 31st Stanford Child Language Research Forum, pp. 40-49.

Dopkins, Stephen & Gleason, Theresa. 1997. Comparing exemplar and prototype models of categorization. Canadian Journal of Experimental Psychology 51(3).212-230.

Frawley, William. 1992. Linguistic Semantics. Hillsdale, NJ: Erlbaum.

Fodor, Jerry A. 1975. The Language of Thought. Cambridge, MA: Harvard University Press.

Herskovits, Annette. 1986. Language and Spatial Cognition. Cambridge: Cambridge University Press.

Hill, Clifford. 1974. Spatial perception and linguistic encoding: A case study of Hausa and English. Studies in African Linguistics (Supplement) 5.135-148.

Hill, Clifford. 1982. Up/Down, front/back, left/right. A contrastive study of Hausa and English. In J. Weissenborn and W. Klein (Eds.), Here and There, 13-42. Amsterdam: Benjamins.

Kamp, Hans & Partee, Barbara. 1995. Prototype theory and compositionality. Cognition 57(2).129-191.

Kripke, Saul A. 1980. Naming and Necessity. Cambridge, MA: Harvard University Press.

Levinson, Stephen C. 1996. Frames of Reference and Molyneux’s Question: Crosslinguistic Evidence. In P. Bloom, M. A. Peterson, L. Nadel and M. F. Garrett (eds.), Language and Space, 109-170. Cambridge, MA: MIT Press.

Levinson, Stephen, Sérgio Meira, and The Language and Cognition Group. 2003. ‘Natural concepts’ in the spatial topological domain—adpositional meanings in crosslinguistic perspective: An exercise in semantic topology. Language 49.485-516.

Burling, Robbins. 1964. Cognition and componential analysis: God's truth or hocus-pocus?

            American Anthropologist 66.20-28.

Putnam, Hilary. 1988. Representation and Reality. Cambridge, MA: MIT Press.

Quine, Willard Van Orman. 1960. Word & Object. Cambridge, MA: MIT Press.

Wierzbicka, Anna. 1980. Lingua Mentalis: The Semantics of Natural Language. Sydney: Academic Press.