[back to list of essays]


Byron W. Bender

University of Hawai`i

par·a·dox n. 1. A seemingly contradictory statement that may nonetheless be true: the paradox that standing is more tiring than walking. 2. One exhibiting inexplicable or contradictory aspects: "You have the paradox of a Celt being the smooth Oxonian" (Anthony Burgess). 3. An assertion that is essentially self-contradictory, though based on a valid deduction from acceptable premises. 4. A statement contrary to received opinion. [Latin paradoxum, from Greek paradoxon, from neuter sing. of paradoxos, conflicting with expectation : para-, beyond; see PARA-1 + doxa, opinion (from dokein, to think...).]--American Heritage Dictionary.


It is my intention in this brief paper to raise what I consider to be some of the key questions in morphology today. In so doing, I will be anything but neutral, and perhaps even provocative, in the way I state them. If that is how I am perceived, I will have succeeded in my purpose, which is to arouse curiosity and stir debate. On many of them we have been floundering all too long; the time has come to face them head-on. The answers are not easy, and will not be found in this paper. At most, I will suggest a few avenues to explore.

[This is the written version of a talk on morphology given as a guest lecture in Ling 615, The Nature of Language, Fall Semester 1994. The course was being reorganized for entering graduate students by Professor George W. Grace to focus on the foundations of linguistics, and on subspecialties of linguistics--morphology in this case being the subspecialty under inspection. Lecturers were asked to touch on what the specialty sees as its central problem, and the underlying assumptions, current approaches, major discoveries, and obstacles to further progress in tackling the problem. In preparation for my lecture, students were asked to read Robins 1959, and were further encouraged to read Hockett 1954 and chapters 2 and 3 in Aronoff 1976.]

The seven paradoxes

Paradox No. 1. The word is central to language, yet resists definition.
Paradox No. 2. We know words as wholes, yet we also know them by their parts.
Paradox No. 3. The derivation-inflection distinction is crucial, yet not clear-cut.
Paradox No. 4. Inflection is both paradigmatic and rule-governed.
Paradox No. 5. The lexicon is where the words of a language are listed yet there are some that should not be listed.
Paradox No. 6. Composition, the Mother of Complexity.
Paradox No. 7. Universal aspects of the arbitrary sign.

[back to list of essays]

Paradox No. 1. The word is central to language, yet resists definition.

The first of these assertions is another way of saying that morphology is indeed an autonomous subfield of linguistics, and that its domain cannot be adequately covered by syntax, on the one hand, and phonology on the other, as was initially attempted by generative grammar. Word semantics is not simply the summation of morpheme semantics, and word form is often not phonetically motivated. Complex words are often more than clusters of linearly-arranged, agglutinative morphemes that can be manipulated by syntactic rule.

Word structure is a matter in and of itself, yet how is the word to be defined, either universally or for a given language? Occasionally a brave soul tries (Hockett 1958:148, 166-176); the wiser course seems to be to side-step the matter (Chomsky 1957:25-28, 96-97) or to recognize words as primes (Householder 1959 :233-236). The target is made even more elusive, we find, by incongruence between the two levels of patterning, giving us in many languages phonological words that are not always coterminous with grammatical words.

Or is this really a nonissue? Is it necessary to define words according to our usual expectations of definitions in order to be able to deal with the structuring of their various shapes?

Back to Top

[back to list of essays]

Paradox No. 2. We know words as wholes, yet also know them by their parts.

The evidence is mounting that, semantically at least, we know words as wholes. The number of lexicalists is increasing, as is the number of lexicalists who have concluded, with Halle and Aronoff, that Word Formation Rules (WFRs) are "once-only" rules, and are best regarded as lexical redundancy rules: "once a word has been formed, and registered in the dictionary, it can't be unformed" (Spencer 1991 :84). In everyday language use, we know (store, retrieve, and use) complex words as wholes, just like simple words, not as sums of their parts.

What then is the status of our knowledge of their parts, the lexical redundancy rules, or WFRs? To what use, if any, do we put them in everyday language use? Bernard Comrie (1981) has observed that a language that was completely suppletive, in which no word resembled any other word, would be impossible, especially if it were also highly synthetic (where the word approaches the sentence in size), and sentences therefore did not resemble each other either.

[His reasoning is as follows, "Here, we can take agglutination as the norm: clearly segmentable and invariant morphemes, and define the index of fusion as deviation from this norm. The extreme deviation from this norm would thus be suppletion, where there is absolutely no segmentability and no invariance, as with English went as the past tense of go. Thus a language [that] represented the ideal fusional type would have all of its morphology in terms of suppletion; if it also had an ideally high index of synthesis, then each sentence would simply be totally and unsegmentably distinct from every other sentence of the language. Given that a language consists of an infinite number of sentences, this is clearly a practical impossibility, which means in practice that as the index of synthesis gets higher, the ratio of agglutination to fusion must also increase; more radically stated, there can be no such thing as an ideal fusional polysynthetic language" (pp. 45-46).]

If he is right, we do need--in all but the most analytic of languages--for an appreciable number of our words to have recognizable, recurring parts. Further, our knowledge of these parts in such languages (languages that have a morphology) would seem to be crucial, even on an everyday basis--not just for coining new words, or understanding those not part of our prior experience.

The way out of this seeming paradox, though not easy, was pointed to by Robins already in 1959. [That he has gone unnoticed, by so many for so long, is in and of itself a paradox. For at least part of the explanation, see Newmeyer 1986 :6-7.] The morpheme is the minimal grammatical unit, but not the minimal semantic unit. That status must be accorded the word, a conclusion echoed by Aronoff (1976:14): "The sign gravitates to the word." The morpheme as formative is typically employed in a variety of functions, with quite different meanings for each function. That a suffix of the shape -s/-ez/-z marks both plurality in nouns and `3s' in verbs is not some strange quirk of English, but par for the course--actually something under par. At least four distinct functions can be identified for the -er suffix morpheme (morpheme in Robins's and Aronoff's sense, not as minimal unit of meaning) in writer (derives agentive nouns from verbs); older (comparative degree of adjectives); Londoner, New Yorker, etc. (`inhabitant, native of'); slipper, jumper, blazer (`article of clothing'). An excursion through Marchand (1969) reveals in similar fashion multiple functions for English de-, dis-, en-, re-, un-, -al, -age, -en, -le, -less, -ic, -ee, -dom, -ful, ing, and -hood, to give only a partial list. Robins gives parallel examples from Sundanese, Japanese, German, Latin, and Greek. Recalling Comrie's observation on our inability to deal with total suppletion, recurring partials such as those just cited should serve to facilitate our coping with complex word shapes, even while treating the words as wholes. These recurring familiar formatives thus constitute building blocks at a level above the phoneme--blocks that carry no inherent meaning, available for any function whatsoever.

Back to Top

[back to list of essays]

Paradox No. 3. The derivation-inflection distinction is crucial, yet not clear-cut.

Some recent theories of morphology purport to find no distinction between derivation and inflection. Spencer (1991:202), in summarizing one such theory, gives seemingly parallel subcategorization frames for two English suffixes, -hood (traditionally derivational) and -z (traditionally inflectional):

hood: [{[A], [N]} _____];   [N, +abstract]
z:    [[N]        _____];   [N, +plural]

The first says roughly that hood added to adjectives and nouns produces abstract nouns, whereas z added to nouns produces plural nouns. Aside from the ability of hood to combine with adjectives as well as nouns, there is nothing about this formalism to distinguish derivation from inflection. All that would be necessary to do so, however, would be to add subindexing to indicate identity or nonidentity of lexemes. To do so would bring out the information critical to all areas of grammar and lexicon that go beyond morphological form, that the two Ns in the first row will always be different lexemes (subindex them i and j, respectively), while those in the second row will always be identical (subindex them both i). Motherhood is a different word from mother, but boys is simply the plural of boy, to give actual examples.

But to maintain the importance of the distinction is not to assert the ease of making it (which is what our third paradox says). An example from Marshallese (Bender 1969:143-148):

  E-mijel       bok  eo.
  3sgSubj-thick book the.sg
  `The book is thick.' 

  Re-mmijel     bok  ko.
  3plSubj-thick book the.pl
  `The books are thick.'

  E-mâni        bok  eo.
  3sgSubj-thin  book the.sg
  `The book is thin.'

  Re-mmâni      bok  ko.
  3plSubj-thin  book the.pl
  `The books are thin.'

`Thick' and `thin' are one of six pairs of opposite stative verbs that undergo consonant doubling when pluralized. [There is considerable variation in which of the consonants is doubled, or whether more than one is doubled, with remijjel, remmijjel, remânni, and remmânni also being heard.] Are they being inflected for the plural, or are we dealing with separate singular and plural verbs? If the former, why should only six pairs of verbs be inflected for number, and not all verbs, or at least all stative verbs? We expect inflection to apply to a much more general class. Alternatively, are we to gloss four separate verbs here, as for example:

mijel `thick (of single objects only)'

mmijel `thick (of multiple objects only)'


The distinction is not always as clear-cut as we would like. Another example can be taken from English, as being typical of many languages in which participles and gerunds have been problematic.

Following are two major expectations of derivation and inflection:

1. Derivations may or may not change the part of speech, but inflections do not. (A lexeme is a member of a given lexical category; it cannot change categories and remain the same lexeme.)

2. Inflections apply with great formal and semantic regularity across a part of speech, while derivations tend toward the opposite: to show irregularities in form, to have semantic idiosyncracies, and to be selective and difficult to predict with respect to the words to which they apply.

The expectations can be seen as intersecting in the following way:

  EXAMPLES    |    CONSTANT     |    CHANGE IN     
              |                 |
 REGULAR AND  |  typical        |  participles and 
  WIDESPREAD  |  inflections:   |     gerunds:     
              |                 |                  
              |  boys N+pl      | Tom's amusing the
              |                 | children with his
              |                 | stories . . .    
              |                 |  
IRREGULAR AND |  derivational   | typical          
IDIOSYNCRATIC |  prefixes (in   | derivations:     
              |  English):      |                  
              |                 |                  
              |  uneasy         |  falsehood N+abs 
              |  international  |
              |  coauthor       |  
              |  discontinue    |
              |  premature      |
              |  readvertise    |

That all four quadrants are clearly inhabited points up the sort of difficulties that have been encountered in attempting to enforce the inflection-derivation distinction as a dichotomy.

Back to Top

[back to list of essays]

Paradox No. 4. Inflection is both paradigmatic and rule-governed.

Presenting inflectional paradigms as paradigms, just as they are found in traditional grammars, is one way of avoiding the difficult choices to be made in segmenting a fusional language. Identifying the exponents within a paradigm permits us to go further toward pointing up sound-meaning correlations without performing questionable segmentations. Yet linguists generally are reluctant to do so, for they are by nature given to microanalysis. Where suffixation has clearly occurred, for example, we want to identify the suffixes and stems, even though their boundaries may be hopelessly obscured, conducting our analysis within the paradigm as if it did not exist.

Generative grammar is no better equipped to deal with the possibility that inflection is inherently paradigmatic than were its predecessors. In fact, there remains a strong note of skepticism, and a decided leaning toward the assumption that the paradigm is superfluous, simply "an epiphenomenon of the morphosyntactic feature system, and of no intrinsic interest" (Spencer 1991:191:224). Yet the evidence is mounting that inflectional phenomena often need something like a paradigm to explain them, phenomena that seem to involve paradigm structure conditions, paradigm leveling, paradigm mixture, and paradigm economy.

Perhaps our conception of what constitutes a linguistic rule is to blame. Morphological rules, or at least inflectional rules, may differ in kind from either syntactic or phonological rules, by being in some sense paradigmatic rules. A morphological rule may give us the shape of a single inflection, say the 3rd person singular present active indicative form of a Latin 1st conjugation verb ending in . . . at, another rule would give the 2nd person ending in . . . a:s; other separate rules would give parallel endings for a 4th conjugation verb ( in . . . it and . . . i:s). The paradigmatic dimension that is missing from individual rules such as these is the fact the first two go together (with each other, and with those for the other persons and numbers) with certain verbs, just as the last two do with other verbs, just as well. Paradigmatic rules are rules that work in tandem, or in concert. We have yet to develop the necessary formalisms to reflect this important fact of language.

Back to Top

[back to list of essays]

Paradox No. 5. The lexicon is where the words of a language are listed, yet there are some that should not be listed.

With the recognition that we know complex words as wholes, at least semantically (see Paradox No. 2), it is clear that each needs its place in the lexicon, first of all simply to register its existence. We know which are words in our language, and which are not. This is not a matter of potential words, but of actual words. For any particular candidate, we are prepared to take a position on its existence. This is not to say that we are never in doubt, but that we feel that the matter is determinate.

Complex words also need space in the lexicon for their semantics, those that may seem fairly transparent, as well as those that are clearly "lexicalized." However, at the ultra-transparent end, there may be an exception, pointed to by Aronoff (1976:35-45): derivational processes that know of no exceptions and whose semantics permit no irregularities. The one example given by Aronoff that seems to meet these qualifications fully is English -ness, capable of producing an abstract noun (whose semantics hold no surprises) from any adjective. Concerning the listing of nouns thus derived in the lexicon, Aronoff says:

We have found no reason to list them. On the assumption that only words [that] are arbitrary in some way must be entered in the lexicon, there is no reason to enter the #ness derivatives of Xous adjectives in the lexicon. The most productive classes never have to be listed. If the #ness forms are never listed, then they can never be blocked, and this is what we find. Nor will there be any sporadic gaps, since the concept of a gap presupposes a list, and we have no list. Nor will they drift semantically, since on our account semantic drift itself presupposes that the item [that] drifts be listed in the lexicon (p. 45). [ Aronoff credits Zimmer in a footnote as "the first person to suggest that productive and nonproductive classes could be distinguished by claiming that only members of the latter were listed in the lexicon."]

I have held this forth to students in the morphology class I teach as a worthwhile project, the full and careful testing of this hypothesis, but to date no serious counterexamples have been turned up: adjectives that cannot be bases for nouns in -ness, or that do so with idiosyncratic semantics. I have myself noted the existence of forgiveness, as seeming at first glance to involve a verbal base, but probably turning out to have been formed regularly on the participle forgiven, then truncated. As such, it now needs to be listed, but does not constitute a counterexample to Aronoff's putative fully productive process. Forgivenness continues as a viable alternate, and thus as the unlisted, regular output from the past participle as base.

Is -ness the only such example that can be turned up in English? Aronoff considers the -ly that creates adverbs from adjectives, but quickly finds gaps. Recently I examined in some detail English re-, which at first glance seems quite productive in the function `to V again.' What I found, however, is that the semantic detail diverges unpredictably (`to V again in identical fashion': relive, replay, reinvent, etc.; `to V again with a view to improvement': rewrite, reinvest, reconsider, etc.; `to V again and return to an original state': reestablish, reunite, reconstitute, etc.; `to V again (redundantly)': rehash, reiterate, refurbish, etc.; `to V again in reverse direction': rewind, repay, retrace, etc. Also, there would seem to be gaps: *refall, *restumble, *reslip, *regrab, *regrasp, *rehold, *redie, *readjourn, etc. At the moment, the best that can be reported is that the testing of -ness, and the quest for other candidates, continue on.

Before leaving this subject, one final note. Recalling the four-quadrant display of our inflection-derivation discussion under Paradox 3, -ness would seem to merit a place in the quadrant together with participles and gerunds as being both "regular and widespread" and involving a "change in part of speech." Deverbal adjective = participle : deverbal noun = gerund: deadjectival noun = ___? Is this simply a matter of tradition failing to supply us with a third name? Possibly, although simply supplying a name will not cause these words to fit any more happliy into a derivation-inflection dichotomy than participles and gerunds do.

But we may be close to a solution here, recognizing this new group (three members thus far) of "inflections" that change category, while also splitting derivations into those that change category and those that do not, along the lines of the four-quadrant display (repeated here, with modifications):

  EXAMPLES    |    CONSTANT     |    CHANGE IN     
              |                 |                  
 REGULAR AND  |  typical        | participles:     
  WIDESPREAD  |  inflections:   |  a rolling stone 
 (INFLECTIONS |                 | gerunds:         
     WITH     |  boys N+pl      |  Tom's amusing   
  BROADENED   |                 |  the children    
  DEFINITION) |                 |  with his        
              |                 |  stories . . .   
              |                 | "ness-nouns"     
              |                 |   playfulness      
              |                 |
IRREGULAR AND |  derivational   | typical          
IDIOSYNCRATIC |  prefixes (in   | derivations:     
(DERIVATIONS  |  English):      |                  
 OF TWO TYPES)|                 |                  
              |  uneasy         |  falsehood N+abs 
              |  international  |
              |  coauthor       |
              |  discontinue    |
              |  premature      |
              |  readvertise    |

For the moment I have given our new inflections the descriptive name of "ness-nouns." There is no more need to list them in the lexicon than there is to list other inflections, including participles and gerunds. Note that proceeding in this direction will make it necessary to revise the assertion made in connection with Paradox No. 3 above, where we said:

1. Derivations may or may not change the part of speech, but inflections do not. (A lexeme is a member of a given lexical category; it cannot change categories and remain the same lexeme.)

The revision would have to state that "both derivations and inflections may or may not change the part of speech" and that "under certain 'de-part-of-speech' conditions, a lexeme may change lexical categories and remain the same lexeme."

So has Paradox No. 5 been resolved? Perhaps.

Back to Top

[back to list of essays]

Paradox No. 6. Composition, the Mother of Complexity.

Compounding is sometimes referred to as composition, and that is what I intend here: the composing of a new word by combining two existing words--at the very least, two roots. This is, I think, the most difficult area of morphology in which to work, because in a sense, it's where it all starts, where parts of phrases turn into words, parts of which may later turn into affixes, all of which may later fuse, on the slow march toward complexity. We have difficulty determining just when this process has begun, and if it has, whether its results should be registered in the lexicon (see Paradox 2), especially if it is ultra-productive (see Paradox 5).

As an example, leaving spelling aside, there is a shift of stress to let us know when a main street has become Main Street (and a proper noun as well as a compound noun), but that phonological clue is not available for the first avenue we come to. While the semantics of such proper nouns are relatively simple, because they have referents, and the referents are unique, can the same be said of common nouns--that their semantics are transparent, and need not be registered in the lexicon? After the fact, when it fits with one of our cultural artifacts, we think the meaning of bathroom towel-rack should be obvious, but without that fit, would we be so sure? Towel-racks could equally well be racks made of less flexible toweling material, or racks for stretching towels. Bathrooms could be chambers for removing the dirt from tubers on their way to becoming fries, or spaces for the application of fixing chemicals in a photo-processing machine, not to mention rooms in modern houses containing only a toilet and a washbowl (with or without a shower cabinet). Or to take an example not quite so close home, are we sure, nowadays, that a water mill is a mill powered by water? Could it not equally well be a mill for producing water, or for analyzing water, or just a mill near the water? [This latter example is from Scalise (1986:90-91), quoted by him from Allen 1978 and Allen 1980 .]

Back to Top

[back to list of essays]

Paradox No. 7. Universal aspects of the arbitrary sign.

It used to be thought that the imitation of sounds in nature was an instance in which languages are less than fully arbitrary in correlating sound with meaning. The idea barely survived the much-maligned bow-wow theory of language origin, and was still endangered at mid-century, when cross-linguistic studies seemed to show that there is considerable diversity and stylization, even in our supposed imitation. [I remember especially one of a series of educational films on linguistics in which Henry Lee Smith, Jr. dropped a chunk of wood into a large tub of water and had speakers of a number of languages give their words for the sound. Few of them sounded much like "splash," as I recall.] But sound symbolism extends well beyond onomatopoeia. The correlation between high front vowels for the eensy-weensy, and those at the opposite corner of the quadrangle for the gross and humongous, seems to be fairly universal. [See Bolinger 1975:24-25, 274-275.] And we continue to wonder why terms from early childhood, such as mama and papa, have such high incidence from language to language. Sound symbolism may extend even to the phonestheme (see the examples below, for most of which I am indebted to Bolinger 1975:218-220). Is there anything universal about the `carefree abandon' of jamboree, jubilee, whoopee, etc., or the sleaziness of sleaze, or is their emotive power felt only by those who know a specific language, English? Is there anything natural about any of these types of supposed sound symbolism?

These remain open questions. Those concerning the phonestheme have received as little attention as they have partly because we have not known how to deal with the phonesthemes within a morphological theory based on the morpheme. For example, are words such as lag, flag, sag, drag morphologically simple or complex? To say that they are simple would seem to deny something they clearly have in common. If we decide that they are complex, and identify their -ag as a morpheme, what are we to do with their residue, l-, fl-, s-, and dr-? Must each bit of residue from every phonestheme be counted as a separate cranberry morph?

Fortunately, the Word and Paradigm approach to morphology enables us to ask different questions, and proceed toward less contradictory answers. The phonestheme may yet become nothing more than a subspecies of the exponent. [For which see Matthews 1991:175-183.]

FORMS   EXAMPLES              SEMANTICS            

-ag     lag, flag (v.), sag,  `droopy' and `flabby'
        bag, drag                                 

-amble  amble, ramble,        `locomotion'         
        scramble, shamble                         

-ash    gnash, crash, trash,  `breaking' or        
        smash, bash, dash     `fragments'          

-ee     shivaree, jamboree,   `absence             
        jubilee, spree,          of restraint'     
        whoopee, whee,        `carefree abandon'   
        free, yippee, glee
-eazy   sleazy, greasy,       `the sleaze factor'

-ician  logician, beautician, `practicioner'
        mortician, magician, 

-icious delicious, luscious,  `sensual indulgence,
        scrumptious,          `appealing to the 
        nutritious,              senses'
-ump    hump, rump, bump,     `heavy masses'
        crump, lump, stump
-ush    lush, slush, gush,     `moist' and
        flush                  `oozy'

-utter  mutter, stutter,       `utter'
        sputter, splutter 

-urry   scurry, worry,         `haste' and
        flurry, blurry,        `confusion'

ob-     objectionable,         `objectionable'
cr-     crawl, cringe,         `bent'
        creep, crumple 

fl-     flitter, flow,         `phenomena
        flicker, flurry           of movement'

gl-     glitter, glow,         `visual
        glare, gleam              phenomena'

sl-     slip, slide, sled,     `horizontal
        slime, sledge, sleigh,    movement',
        sluice                 `lack of 

sw-     swoop, swell, swish,   `flourish',
        swoon, swagger         `sweep'
tor-    torture, torment,      `pain'

tur-    turbulent, turment,    `violence'

tw-     twist, twirl, tweak,   `twisting motion'
        twine, twinge 

vi-     vituperative,          `intense ill-temper'
        vitriolic, vindictive, 

Back to Top

[back to list of essays]


Fortunately, I began thinking about my assigned topic not long after teaching the department's introductory course in morphology for the eleventh successive spring semester, while my mind was still filled with loose ends from the course. One or two seemed to me then so overwhelmingly paradoxical that they led naturally to the theme for this paper. I wasn't sure how many there were when I started, but it wasn't too difficult to find these seven, or to present them as paradoxes. But who knows, what to one person seems paradoxical may to the next person be an instance of clarity, or at least the threshold of discovery. Be my guest.

Back to Top

[back to list of essays]


Allen, Margaret Reece. 1978. Morphological investigations. Ph. D. dissertation, University of Connecticut.

------. 1980. Semantic and phonological consequences of boundaries: A morphological analysis of compounds. In Juncture, ed. by M. Aronoff and M. L. Kean, pp. 9-27. Amna Libri. Back Up

Aronoff, Mark. 1976. Word formation in generative grammar. Linguistic Inquiry Monograph 1. Cambridge, MA: The MIT Press. Back Up (to first ref.) Back Up (to second ref.)

Bender, Byron W. 1969. Spoken Marshallese: An intensive language course with grammatical notes and glossary. Honolulu: University of Hawai`i Press. Back Up

Bolinger, Dwight. 1975. Aspects of language. New York: Harcourt, Brace, Jovanivich, Inc. Back Up

Chomsky, Noam. 1957. Syntactic structures. The Hague: Mouton. Back Up

Comrie, Bernard. 1981. Language universals and linguistic typology. Chicago: University of Chicago Press. Back Up

Hockett, Charles F. 1954. Two models of grammatical description. Word 10:210-234. Back Up

------. 1958. A course in modern linguistics. New York: Macmillan. Back Up

Householder, Fred W. 1959. On linguistic primes. Word 15:231-239. Back Up

Marchand, Hans. 1969. The categories and types of present-day English word-formation. München: C. H. Beck. Back Up

Matthews, P. H. 1991. Morphology. 2nd ed. Cambridge: Cambridge University Press. Back Up

Newmeyer, Frederick J. 1986. Linguistic theory in America. 2nd ed. Orlando: Academic Press. Back Up

Robins, R. H. 1959. In defence of WP. Transactions of the Philological Society 1959:116-144. Reprinted in Diversions of Bloomsbury: Selected writings on linguistics, by R. H. Robins, pp. 47-77. Amsterdam: North-Holland. Back Up

Scalise, Sergio. 1986. Generative morphology. 2nd ed. Studies in Generative Grammar no. 18. Dordrecht: Foris. Back Up

Spencer, Andrew. 1991. Morphological theory. Oxford: Basil Blackwell. Back Up (to first ref.) Back Up (to second ref.) Back Up (to third ref.)

Back to Top

[back to list of essays]