PROTOTYPE THEORY, COGNITIVE LINGUISTICS AND PEDAGOGICAL GRAMMAR R .K.

Johnson Faculty of Education University of Hong Kong

Introduction This paper begins from a general theory of natural. categories, developed in the 1970's by Eleanor Rosch and her colleagues (Rosch et al, 1976) and referred to here as prototype theory . This theory has been applied to linguistic categories by George Lakoff (1982) under the heading 'cognitive linguistics' and can be shown to be consistent with the conclusions of a number of other 20th century linguists who have addressed semantic rather than purely formal aspects of language . It is suggested that prototype theory offers a principled. approach to the exemplification of form - meaning relationships within language and to the development of language teaching exercises which focus upon specific aspects of the language system and which offer 'enriched' input to the 'Language Acquisition Device' ; i .e . it offers a principled basis for the development and application of pedagogical grammars . Prototype Theory Rosch et al . offer a theory of the ways in which human beings and other organisms deal cognitively with their perceptions of the world 'out there' . "The world consists of a virtually infinite number of discriminably different stimuli . One of the most basic functions of all organisms is the cutting up of the environment into classifications by which non-identical (Rosch et al ., stimuli can be treated as equivalent ." 1976 :383) They claim that this process is principled and depends on the 'real-world attributes' of what is perceived, and also upon the characteristics of the perceptual . apparatus itself (i .e . We can only categorise on the basis of what we can perceive and, all things being equal, that which is more easily perceived will be of greater significance to the categorisation process .) Rosch and Mervis (1975) have shown "that the more an item is judged to be prototypical of a category, the more attributes it has in common with members of contrasting categories" (Rosch et al ., 1976 :433) . At the same time however, Rosch et al . (1 .976 :384) point out that :

"It is to the organism's advantage not to differentiate one stimulus from others when that differentiation is irrelevant for the purposes in harid ." There are then two basic cognitive principles operating : the first is to achieve maximum differentiation, with the prototypical. instance of

a category being that which distinguishes it most clearly from all other categories . The second is to avoid cognitive overload, which would result from over differentiating and a consequent loss in flexibility in grouping those things which share important characteristics, whilst being in other respects unlike . The principle of differentiation has been central to linguistic theory since de Saussure, who maintained that the language system is one in which "tout se tient" and in which "11 n'y a que des differences" (de Saussure 1953 :166) . The similarity between de Saussure's theory of language system, and the prototype theory of natural categorisation is well illustrated by de Saussure's account of the 'value' of linguistic elements, which I take to mean the cognitive or meaning value expressed by a linguistic form . These values are defined . . . . . " . . . non pas positivement par leur contenu, mais negativement par leurs rapports avec les autres terms du systeme . Leur plus exact characteristique est d'etre ce que les autres ne sont pas ." (op . cit .) The notion of negative definition is a difficult one to work with, but the principle of identification through contrast has been central to the synchronic description of language throughout the 20th century, regardless of the particular school of linguistics involved . Lakoff (1982) summarises Rosch's work in cognitive psychology and its application to - cognitive linguistics as encompassing entities - colours, events, actions, perceived spatial relations, causation, social institutions, syntactic entities (nouns, verbs, subjects, grammatical constructions) phonological entities, mental images, etc ." Lakoff (1982 :44) notes further : "This research has produced overwhelming support for prototype theory, or more properly for the need to develop further a theory of natural categorisation along the lines of Rosch's results ." In relation to each instances of prototypical where category membership context and communicative of the above areas of research, L!ikoff discusses category membership and boundary phenomena is unpredictable and dependent largely upon purpose .

the same phenomena and have come to Other linguists have tackled similar conclusions . J .R . Ross (quoting Lloyd Anderson) formulated the question as follows : . "One should not ask 'Is the phenomenon in question discrete or non(Ross 1974 :121) discrete?' but rather 'How discrete is the phenomenon?'" and adopted the term 'squish' in papers on boundary phenomena amongst linguistic categories (Ross 1972, 1973) . W . Labov in experimental studies conducted over a ten year period, He concluded (1973 :143) : focussed like Ross upon boundary phenomena . "Instead of taking as problematical the existence of the categories, we can turn to the nature of the boundaries between them . As linguistics then becomes a form of

boundary theory rather than a category theory, we discover that not all linguistic material fits the categorical view : there is greater or lesser success in imposing categories upon the continuous substratum of reality ." The "greater or lesser success in imposing categories" is interpreted here in terms of degrees of prototypicality as defined by Rosch . Rosch's position is indeed the reverse of Labov's . Where Labov chooses boundary theory as a way of escaping from the problems associated with categorisation, Rosch proposes an approach which allows the cateory to be the main focus of attention while at the sane time accounting for boundary phenomena . In addition to the experimental studies he discusses, Lakoff suggests that the psychological validity of prototype theory receives support from the existence of linguistic terms which appear to indicate degree of prototypicality . He gives 'sort of and 'kind of as examples of expressions which indicate non-representative members of a category, while such terms as 'par exeelZenee' indicate prototypicality (1982 :44) ; Lakoff also quotes from Kay's (1979) analysis of 'ZooseLy speaking' and 'strictly speaking', amongst other linguistic 'hedges' . He concludes : "In short, if words can fit the world, they can fit it either strictly or loosely, and the hedges strietZy speaking and loosely speaking indicate how narrowly or broadly one should construe the fit ." Prototype theory can be extended beyond lexical and grammatical levels to discourse and textual levels of analysis . Brown and Yule (1984) summarise much of the recent literature on 'story ,grammars', 'frames', 'schema', 'scripts', 'scenarios' and 'schemata' . These notions like the work of Sacks and others on turn-taking, appear to suggest that all. 'grammars of expectancy' are based upon what Lakoff, borrowing from C . Fillmore, calls Idealised Cognitive Models (1982 :48) and which are essentially prototypical informational interactive structures as opposed to grammatical or lexical entities . However, while noting that the the theory may be wider, this paper restricts the discussion application of to the area traditionally associated with pedagogical grammars . Prototype theory therefore seems to suit the needs of some linguists very well . It provides a theoretical framework within which it is possible to solve at least some of the problems associated with adding the semantic dimension to linguistic investigation . The fact that a theory is relevant to linguistic theory does not of course make it necessarily relevant to language teaching and learning . The differences between these two enterprises, in terms of participants and goals, product and process, have been emphasised frequently and judiciously over the last twenty years . Applied Linguistics has turned increasingly towards Psycholinguistics and the processes of language acquisition for inspiration and for theoretical support . It is interesting therefore to note that prototype theory has also been extended to this area . In first language acquisition studies, the problems in determining a child's meaning are well-known, if too frequently ignored . The attempt to introduce notions of prototypicality in relation to an early stage of language development might therefore seem overambitious . Nevertheless, one study at least, by Labov and Labov, (1974) reported by Clark (1979) suggests that the child's lexicon, however exotic it may appear in its - 14 -

earliest stage of development, can be interpreted in terms of prototypicality, or, from the standpoint of Labov's interest in boundary theory, degrees of non-prototypicality . In this study Labov and Labov recorded the over-extensions of the word 'cat' (one of two words in the child's total repertoire) and claimed to have identified a set of 'core' features . Animals fitting any or all of these core features were called 'cat' . However, the more core features were involved, the more confident the child appeared to be in her use of the word . Animals with none of the criterion features were never named 'cat' . In studies relating to second language acquisition in adults, the notion of prototypicality is more managable, since it is possible to assume a fully developed mother tongue language system which is mapped onto and realises a speaker's cognitive competence . In this context, recent discussions of transfer from the first language to the second language (e .g. Gass and Selinker, 1983) suggest strongly that prototype theory and theories of markedness and of universality in language have much in common . S . Gass (1984) reviews the evidence from a number of studies of transfer, in particular_ her own. and those conducted with J . Ard, those by E . Kellerman (1979, 1983) and by W . Rutherfo rd (1982, 1983) . Gass proposes that there are 'core' meanings, which I take to be equivalent to the prototypical categories discussed above, and concludes that : " . . . meanings which were closer to the 'core', that is, were more basic in meaning, were more likely to be transferred than those which were furthest from the core ." (Gass 1984 :129) Gass gives as examples, 'kick the bucket' and 'kick the ball' with the latter being perceived as the more transferable of the propositions . Similarly Kellerman (1978) showed that Dutch students were more willing to transfer the meaning of the Dutch verb 'breehen' to an English context such as 'He broke his leg' than 'The waves broke on the shore' though both are equally acceptable in Dutch and English . Hatch makes a similar point in her discussion of transfer (Hatch 1983) claiming that transfer is not random but systematic, with a tendency to be limited to core meanings . Hatch notes however that language 'distance' may be a factor . The more similar the languages appear to be, the more likely the learner is to extend the transfer ; the more distant, the more conservative the transfer will be . As was suggested earlier in the discussion of hedge terms, language users appear to have a strong intuitive 'feel' for the degree of The prototypicality of meaning - form relationships in various contexts . discussions of transfer referred to above suggest that the notion of prototypicality of meaning is related in interesting ways to theories of the nature of language universals, to marked/unmarked distinctions both within languages and across them (unmarked being more universal, less language-specific and more transferable) and therefore to speed and ease of learning and, using Hatch's analogy of a cognitive punch card system, to the question of whether 'meanings' for the different language forms require modification to existing cards or whether new cards have to be 'punched' .

The problems for the language teacher who attempts a direct approach to the teaching of specific aspects of the language system have been well documented . Theoretical and practical objections have been raised in profusion since the grammar-translation approach was condemned for, as Rivers succinctly expressed it, teaching about the language instead of teaching the language . Eliminating (or at. least radically curtailing) overt instruction about the grammar did not end the problem or the objections to 'structural' exercises . These have been seen to be irrelevant to the "necessary and sufficient conditions for language acquisition" (Newmark and Reibel, 1970), as indeed they are, and as a mere waste of time, which is more arguable ; as a cause of teacher-induced error (Corbluth 1974 ) (and most honest language teachers wince in recognition of the at least partial truth of this) ; as promoting "language-like behaviour" (Spolsky 1968), "structure-talk" (Dakin, 1973), & "usage" rather than language "use" (Widdowson 1978) . Yet the structural exercise lives on, deprived now of its behaviourist and structuralist underpinnings, it nevertheless seems to be an essential element within most language teaching programmes, and is regarded as such by most teachers . In practical terms, addressing specific aspects of the language system can be unsatisfactory, if not downright embarassing for the teacher who values clarity and consistency in explanation and/or exemplification . 'Rules' (i .e . the form-meaning relationship realised by a linguistic element) at times appear to operate consistently in differentiating the semantic functions of related elements (equated here with prototypical instances), at other times the distinction disappears (boundary phenomena) . The data presented below exemplify the problem and demonstrate the applicability of prototype theory to pedagogical issues of the linguistic system, first to show that prototype theory has general application, and secondly to substantiate the major claim of cognitive linguistics, which I take to be as follows : Every linguistic form expresses an underlying
cognitive entity (or meaning) which can be differentiated from azz other such entities where the realisation of those entities is prototypical .

The first example attempts to clarify what I am not talking about . It illustrates purely formal problems within the grammar . Sapir said that "all grammars leak", and Ross, from whom this example is taken, states that such sentences " . . . . fall between the cracks of the core system ."
1. Either Tom or the gir7.s (was were) responsible .

There is no solution within English grammar to this problem of subject/verb agreement . Users of English simply have to avoid such constructions if they wish to avoid formal error . Formal problems of this kind would (I like to think) cause enormous difficulties for that hypothetical autonomous, context-free, sentence-making machine as it grinds on through eternity generating all and only the sentences of the language . Such problems do not concern us here . Example 2 illustrates what I will refer to as values, or 'boundary' phenomena . 'weakened' meaning

2.

They were both wearing the same hat .

There is a systematic distinction in English between same and Given the prototypical values of this distinction, (2) could possibly be the caption to a rather weak joke, or an entry in a fancy dress parade ; yet very few people would interpret (2) in. that way . It would be read as meaning that two people were wearing identical., or similar, hats . In (3), no joke interpretation is possible .
'identical' .

3.

(Lady entering a friend's flat)

We've both got the same stair-carpet!

There is nothing unnatural about (2) or (3) . No ambiguity or uncertainty about the meaning of the utterance results from this weakening of the meaning value of 'same' and the consequent blurring of the 'polar' distinction with 'identical' . (4) gives 'same' and 'identicaZ' their prototypical values . 4.
(Customer in a jewellery store) Is this the same gem stone that you . showed me before, or an identical one?

one might also speak of 'identical. twins' and 'non-identical twins' . The 'same twins' would have a quite different meaning, and no. change of context could render 'same twin' equivalent. t o 'identical train' . In this sense, prototypical values are context-free . Numerical reference might seem to be one aspect. of the language which would be unlikely to exhibit weakened or variable semantic values . The following characterises the prototypical. values of a part of that. system : All : more than two Both : two Either : one or other of two Neither : not either
5. They surveyed the land on either side of the river .

Example 5 means that both sides of the river were surveyed . Some second language speakers of English, however, understand (5) as meaning : 'the land was surveyed on one side of the river but not on the other', because they have applied the prototypical value of the word 'either', while native speakers do not . It is of course easy to illustrate the prototypical values in operation, as in (6) .
6. You can have either of these, but, don't, take, them both .

The semantic features which 'either' and 'both' share relate to 'twoness' ; and in particular contexts the prototypical value of 'either' weakens to express these shared features . In paired oppositions of this kind, one element may therefore be regarded as marked ( 'identical' and 'both' ), its value being capable of being subsumed by the unmarked 'either') . element ('same' and

The next set of examples investigates the relationship within the verb phrase between the perfective aspect and simple forms of the verb phrase . Example 7 illustrates the prototypical distinction between the present perfect and the past simple tense . 7. (Interviewer to interviewee) How Long have you Lived in .London? (a) (b) How Long did you Live in London?

In (8), the distinction, if it exists at all, is minimal, affecting neither the appropriacy of the question nor its communicative effect . 8. (Wife to husband as he is getting into bed) (a) Have you Locked the door? (b) Did you Lock the door?

The semantic feature shared by the past simple and present perfect tenses is 'pastness', the polar opposition between the two tenses expresses the relevance, or lack of it, of a past action to a present situation . In (7b), the choice of the past tense indicates a presupposition on the part of the speaker that a situation no longer exists (i .e . the In 7a, the choice of the present addressee no longer lives in London) . perfect tense indicates a presupposition that he or she does still live in London . In (8) however, the notion that a husband after :Locking up the house for the night, might unlock it again, would be too machiavellian for most readers, and most wives ; so the distinction becomes .redundant . Similarly, the past perfect tense may operate prototypically in opposition to the past simple tense as in (9) ; 9. (a) (b) When our guests finally arrived, we had eaten all the food . When our guests finally arrived, we ate a7.l the food.

Or it may not, as in (1Q) . 10. (a) (b) We Walked for ten miles before, we got, a Lift . We had walked for ten miles before we got a Lift .

The perfective aspect may even be considered only marginally acceptable in contexts such as (11) ; though its use would seem to be 'correct' judged by purely formal criteria . 11 . (a) (b) We were having a good time until, you arrived. We had been having a good time until you arrived .

The future perfect tense also may maintain a clear meaning distinction between itself and the future simple ; 12 . (a) (b) We will have finished when you get there . We wiZZ finish when you get there .

or the distinction may be weakened to the point where it no longer operates .

13 . (a) (b) We will finish before you arrive . We will have finished before you arrive .

The past perfect and future perfect . tenses express the relationship in time of one event to another . However there are other ways of expressing sequence ; the order in which events are recounted is assumed to be the order in which they occurred and it is only when this assumption is incorrect or when other markers such as 'before', 'after', are absent that these perfect tenses express their polar values (i .e . in 9a and 12a) . Within the noun phrase system, determiners, quantifiers, countability and plurality are used below to illustrate this same feature of the rules and their relationships . The polar meaning values may be reduced to the point where the choice of one or other of the related forms is equally natural and in, no way changes the meaning : e .g . (14) and (l5) for 'a/the' and 'some/any' respectively .
14 . (a) (b) I asked a. taxi driver who brought us here . I asked the taxi driver who brought us here .

15 .

(Butcher to customer) Would you like any sausages today, Mrs . Brown? (a)
(b) Would you like some sausages today, Mrs.

Broom?

In (16) and (17) the polar values are in full opposition, and the meanings expressed are therefore clearly different .
16 . (a) (b) (a) (b) I'm Looking for a wife . I'm looking for the oife . Some of my friends would lend you the money Any of my friends would lend you the money .

17 .

In polar opposition with 'a', 'the' expresses such semantic features as 'known', 'specifiable', and 'unambiguous in terms of reference' as in (16) . It could be objected that (16) is a highly colloquial usage ('the wife'), but there is no reason why colloquial usage should require a set of The meaning values rules distinct from formal usage ; quite the contrary . usage conform precisely to those indicated above, and are paralleled of this by such other examples as 'the pub', 'the church', 'the car', 'the shop' However, (7 .8) provides a more 'standard' and for. that matter 'the old man' . illustration showing both weakened and polar values of 'a' and 'the' .
18 . As I was crossing (a/the) busy road in town, dodging cars and bicycles . to catch (a/the) bus that was just leaving, (a/the) driver slammed his brakes on and then swore at me .

In the three instances in (18) where 'a' or 'the' may be inserted, the choice is semantically empty in the first two . A native speaker might use, or accept the use of, either, and the meaning of the message as a whole would not be affected in any way . The shared features of 'a' and 'the' seem to be the purely formal ones associated with their status as determiners : i .e . for each of these noun phrases there is a formal Either 'a' or 'the' can fulfil that requirement for a determiner . requirement . In the third instance, however, the polar opposition is crucial to the identification of the swearer ; 'a driver' would signify any one of the many using the busy road at that time ; 'the driver' must refer to the driver of the bus, the only specified vehicle, and therefore by implication, the only specifiable driver . Similarly 'some' and 'any' have a definite/indefinite polar opposition as in (17), which may be weakened to the point where either is merely a generalised marker of quantity (15) . The feature of 'countability', which distinguishes 'mass' nouns from 'countable' nouns in English, has a clear semantic value in distinguishing 'stone' the material from 'a stone' ; 'glass' from 'a glass' ; 'cloth' from 'a cloth' . But in some contexts the distinction is weakened to the point where either form may be used .
19 . (a) (b) The mixture of gas in these containers is explosive . The mixture of gasses in these containers is explosive .

The relations discussed above are essentially paradigmatic ; i .e . they illustrate the differences in meaning which may, or may not, result from substituting one element in an utterance for another element, depending upon whether the values expressed by the opposition between the elements is operating strongly or weakly . Constraints also operate syntagmatically ; e .g . determining whether a non-finite verb should be realised as an infinitive or as a present participle in a particular embedded clause . The verb 'want' (20) is said to co-occur with or 'take' an infinitive :
20 . I want to talk to her at, the party .

while 'enjoy' co-occurs with the

'-ing'

form of the embedded verb (21)

21 . I enjoyed talking to her at the party .

These restrictions are generally considered to be purely formal ; i .e . they do not mark any semantic value, and examples such as (22) appear to support the notion that no semantic value is involved, and even to raise questions about the centrality in the language system of the meaning -form relationship .
22 . (a) (b) He likes talking to people at, parties . He likes to talk to people at parties .

Example (23) however shows that there is a meaning distinction when the prototypical values operate .

- 20 -

23 .

(a) (b)

He's too drunk to remember to post the letter . He's too drunk to remember posting the Zetter .

The infinitive seems to express or relate to unfulfilled conditions or future actions, while the participial form seems to express fulfilled or completed actions . The fact that the verb 'want' can only be associated with the infinitive form tends to confirm that the choice is semantically based since 'want' implies an unfulfilled condition (20) . Enjoyment, it seems, can relate only to what has been experienced or is being experienced, and Thus co-occu .rence with the infinitive not to unfulfilled conditions (21) . or participial forms seems to be determined semantically and not formally . However, this leaves for consideration verbs said to take either form . Cases such as (22) can be accommodated within the theory being presented here as 'boundary' phenomena, but further investigation shows that such verbs are in fact affected by the same semantic criteria and that prototypical instances do arise . 24. (a) (b) I would like to borrow your car, if I may . '*I would like borrowing your car, if I may .

In (24), where 'like' has many of the features of 'want', (a) is In (25) the situation regarding clearly acceptable, and (b) is not . acceptability is reversed . 25 . (a) (b) I liked driving your car very much last weekend . '*I Ziked to drive your car very much- last weekend .

In (25), 'like' has much the same semantic value as 'en,~oy', and in this case the '-ing' form is acceptable, and the infinitive is not . Thus, the essence of the theory of language presented here is that it is a 'meaning-driven' system . Linguistic rules exist, to use Sapir's Linguistic forms are expression, in order to keep meanings apart . discrete entities, but the meanings these forms express are not discrete in any sense . Their values are determined by the nature of-the relationships obtaining with other elements within the system . These relationships are not constant, but may be strongly expressed through opposition of prototypical meaning values, or these values may be weakened in non arbitrary ways depending upon context . The prototypical values discussed. and illustrated in this paper are clearly not the basic units of meaning . These have been referred to here rather informally as 'semantic features' or 'cognitive elements' . Linguists, psycholinguists cognitive psychologists and philosophers, anyone in fact who might be interested in the study of the nature of meaning, must However, the level clearly focus upon these abstract underlying elements . of abstraction chosen for discussion here is that which is considered most appropriate for pedogagical purposes ; i .e . the level at which meaning value can be related most transparently to its realisation as linguistic, form. As will be obvious by now, the claim being made in this paper is that prototypical instances of form-meaning relationships offer 'enriched' input to the language acquisition device .

In an important sense, the ways in which the mind processes and The stores input are irrelevant to language teaching and learning . 'language acquisition device' may be relied upon to do whatever it does, provided only that the necessary and sufficient conditions for language acquisition are met . The question then arises whether it may be possible to enhance the necessary and sufficient conditions by 'enriching' the input to the L .A .D . in various ways . The question has not been addressed directly in this paper, but the assumption here is that intervention is possible and desirable, and promotes more rapid and more effective language learning than would otherwise be possible . The particular type of intervention under consideration here is the 'structural exercise', involving the identification of a particular 'rule' within the language system (i .e . a particular meaning/form relationship) and the development of a series of learning experiences which will enable the learner to focus upon, gain access to, and finally integrate that rule into the internalised language system ; thus moving the learner forward along the interlanguage continuum .

Summary and ConeZusion

This paper has discussed a particular theory of cognitive categorisation, prototype theory, which has been applied to linguistics under The data presented the general heading of cognitive .linguistics . illustrate the variable nature of the semantic realisation of linguistic rules showing that meaning-form relationships may be realised strongly . It is my contention i .e . prototypically, or weakly depending upon context . that prototypical instances offer precisely that enriched data which should The nature of facilitate the work of the language acquisition device . the teaching and learning activities which would best. exploit the potential of such instances must be left to a future paper . The question whether such intervention is in fact of any value must be tested empirically .

REFERENCES Bailey, C-J .N . and Shuy, R .W .

Variations in English.

1973 . New Ways of Analysing (Eds .) Washington, D .C . : Georgetown University Press .

Discourse Analysis .

Brown, G . and Yule, G . University Press .

1983 .

Cambridge : Cambridge,

Clark, E . 1979 . The Ontogenesis of Meaning . Verlagsgesellschaft Athenaion . Corbluth, J .D .

Teaching 28(2) :118-25 .

Wiesbaden : Akademische
English Language

1974 .

'Remediation or development' .

Dakin, J . 1973 . The Language Laboratory and Language Learning . London :Longman . 1983 . Language Transfer in Language Gass, S . and Selinker, L . (Eds .) Learning . Rowley, Mass . : Newbury House . Gass, S . 1984 . 'A Review of Interlanguage Syntax : Language Transfer and Language Universals' . Language Learning 34(2) :115-132 . 1983 . Psycholinguistics : a Second Language Perspective . Hatch, E .M . Newbury House . Rowley, Mass . : 1979 . Kay, P . Revisited .' 'The Role of Cognitive Schemata in Word Meaning : Hedges Berkekey Cognitive Science Program, ms .

'Giving Learners a break : native language intuitions Kellerman, E . 1978 . Working Papers in as a source of prediction about transferability' . bilingualism 15 :59-92 . Kellerman, E . 'Transfer and non-transfer : where we are now' . Studies in Second Language Acquisition 2 : 37-57 . 1979 . Now
yoi3

Kellerman, E . 1983 . 'Now you see it . L . Selinker (Eds .) q .v .

don't' .

in S . Gass and

1973 . Labov, W . 'The Boundaries of words and their meanings' in C-J .N . Bailey and R.W . Shuy (Eds .), pp . 340-73 . 1982 . Lakoff, G . 'Categories and Cognitive Models' Series A, No .96, : Linguistic Agency University Trier . Trier 1970 . 'Necessity and Sufficiency in Language Newmark, L . and Reibel, D .A . Learning' in M . Lester (Ed .) Readings in Applied Transformational Grammar . New York : Holt, Rinehart and Winston, pp . 220-244 . Rosch, E . and Mervis, C .B . 1975 . 'Family resemblances : studies in the internal structure of categories' . Cognitive Psychology 7 : 573-605 . Rosch, E ., Mervis, B ., Gray, W . D ., Johnson, D .M . and Bayes-Braem, P . 1976 . 'Basic objects in natural categories' . Cognitive Psychology 8 : 382-439 .

- 23 -

'The category squish : endstation hauptwort' in Papers Ross, J .R . 1972 . for the Eighth Regional Meeting of the Chicago Linguistic Society, pp . 316-328 . Ross, J .R . 1973 . pp . 96-140 . 'A fake NP squish' in C-J .N . Bailey and R .W . Shuy (Eds .)

'Markedness in Second Language Acquisition' Rutherford, W. 1982 . Language Learning, 32 :85-108 . Rutherford, W . 1983 . 'Language typology and Language Transfer' in S . Gass and L . Selinker (Eds .) Saussure, F . de . 1953 . Cours de Linguistique Generate . Paris : Payot .

1966 . 'Critique in programmed F .L . instruction' . Spolsky, B . International Review of Applied Linguistics 4 :11 .9-129 . 1978 . Teaching Language as Communication . Widdowson, H .G . Oxford University Press . London :