[back to list of essays]

Morphological Processes[Note 1]

[back to list of essays]


It is possible to distinguish between phonological processes, which function to make words easier to pronounce or more salient to the listener, and morphological processes, which function to make one word or inflection distinct from another. To the former belong assimilation and dissimilation (including haplology); apocope and syncope; prothesis, epenthesis, and metathesis; fusion, liaison, and breaking; vowel harmony; rhotacism, and probably several others I am overlooking. What does a listing of the latter look like?

The answer to this question, which is the aim of this paper, implies the following question: What are the kinds of resemblances two wordforms can have to each other? I am assuming that a semantic resemblance between the two exists, and am focusing on the variety of formal resemblances that may accompany it­­the formal mechanisms used in either inflection or word formation.

Certainly the most common type is affixation.[Note 2] One wordform may have an affix where the other has none (boy, boy-s), or where the other has an affix that is different (Latin urb-s (nominative), urb-is (genitive) 'city'). The extreme prevalence of affixation has created problems for morphological theory. It has overshadowed other legitimate processes to such an extent that either (1) they are ignored, or (2) there is an attempt to reinterpret them as affixation, albeit strange and deviant, and difficult to deal with.

To ignore the other types is to assume that there is only affixation. This was one of the chief factors leading to the theory of the morpheme that has dominated linguistics through much of this century, the expectation that all morphological contrasts consist of segmentable material correlating on a one-to-one basis with meaning contrasts--segmentable because they result from either suffixation or prefixation (rarely, infixation). The expectations that all languages have segmentable morphemes as their minimal meaningful units and that affixation is the only available morphological process go hand in hand.

When other processes are taken into account, all too often their results are reinterpreted as affixes in the quest for segmentable morphemes. If this could be done easily and well, and without distortion or obfuscation, there should be no complaint, and affixation could well hold claim to being the universal morphological process. But there are few who would maintain that any of the five alternative analyses Hockett considers for the vowel modification of English took is either appropriate or convincing. [Note 3] On the other hand, there are many who would see nothing wrong with his IA reinterpretation of reduplication as "an affix ... that is morphophonemically a chameleon" (388); it is quite common to view reduplication as a special type of affixation. Yet, as we shall see, to do so may blind us to evidence that it represents a quite different kind of gesture on the part of the language user.

What follows, then, is a preliminary attempt to rescue other morphological processes from oblivion. Our purpose here is simply to provide a catalog of the range of types that need further investigation if morphological theory is to move forward. [back to outline]


One of the questions to be given special attention is the type of gesture each process represents for the speaker, or for the hearer through reenactment under a motor theory of speech perception. Gesture is to be interpreted at a full range of levels, from the psychological to the physical.[Note 4] If inflections are tunes that are played on words, what are the different mental and motor skills required by each type of tune? [back to outline]



Prefixation and Suffixation

It is not often recognized that two quite different gestures may be involved in affixation--especially in suffixation, and quite possibly also in prefixation in some languages--depending on whether the affix is added to the base, or whether it replaces an affix that is already "built into" the base. This is Bloomfield's (1933:224-226) distinction between word-inflection and stem-inflection. In the former, "a paradigm consists of an underlying word (itself a member of the paradigm) and some secondary derivatives containing this underlying word . . ." The unaffixed word is itself a member of the paradigm. In the latter, "none of the forms in a paradigm can conveniently be viewed as underlying the others," and none is unaffixed. A bound stem is seen as occurring throughout the paradigm. An English speaker adds a suffix in inflecting the 3rd person singular, whereas a German speaker replaces:

  I laugh     ich lach-e  
  you laugh     du lach-st  
  she laugh-s     sie lach-t  

Even the citation form (the infinitive lach-en) must bear an affix in a stem-inflected language such as German. In this type of language, to affix is always to replace--which is actually a subtype of Modification.As such, it constitutes a different gesture from true affixation.

I will have little more to say here on these two most common processes, other than to suggest that the difference in the gestures required for each may be greater than often assumed: anticipation vs. perseveration, preposing vs. postposing, etc. Also, their metrical consequences seem quite contrastive, just as do the factors that lead to their origin (and that may account for a language's favoring one over the other or using the two for quite different purposes). [back to outline]


Since it often disturbs the integrity of words at their very roots, interrupting them as it were, infixation seems at first glance a process totally other from the more common types of affixation in the performance and perceptual skills it requires, although obviously still well within the range of human capability.

Yet as manifest in Austronesian languages it occurs early in the root and alternates under certain dissimilatory conditions with prefixation. Thus in Tagalog, the verb linisin 'to clean something' is inflected for the Perfective Aspect alternatively with an -in- infix or a ni- prefix: lininis or nilinis. (The ­in suffix on the basic form of the verb disappears from these inflected variants.) [back to outline]


This process has no rival for the variety of alternate names that have been used to describe it: identity operation, zero operation, zero derivation, null adjunction, and implicit transposition, to mention at least some.[Note 5] It is what takes place when man the noun is used as man the verb, with no overt change in form but obvious change in function. As this example may suggest, the process is usually identified only in derivation.[Note 6] [back to outline]


This process can be classified according to the amount of a form that is duplicated, whether complete or partial, and if the latter, according to exactly which part. Several such types may function side by side in a given language. So, for example, in Marshallese, one finds at least three types of partial reduplication:

Initial C: liw lliw
  scold someone be angry
Initial CVC: yetal yetyetal
  go walk
Final CVC: takin takinkin
  socks wear socks
Combination: kijdik kkijdikdik
  rat be infested with rats

It is not uncommon to find that more variation is tolerated in the output of reduplication than of other morphological processes, such as affixation. For example, the plurals of certain stative verbs in Marshallese have the initial C of the verb stem reduplicated, but there are variants in which the next consonant, or both, are reduplicated (ye- is the 3S subject marker, re- 3P):

Singular Plural
ye-maniy re-mmaniy
it is thin they are thin
ye-kadiw re-kkadiw
it is short they are short

Examples such as these give rise to the analogy with "playing a tune" used in the discussion of gesture above. Reduplication especially, with output such as these in contrast with the usual precision of affixation, can be seen as a special kind of gesture that one plays upon a form. It matters little whether the force and timing of the gesture are such as to affect a given segment or its neighbor, or both.

Another example of the tolerance of reduplication for variation in output can be seen in Tagalog verbs that include either the pag- or ka- prefix in their formation. The initial CV of either the prefix or the verb root may be reduplicated in inflecting for the Contemplated Aspect. The same variation occurs in the Imperfective Aspect where -in- infixation occurs as well:[Note 7]

pagaralan papagaralan pinapagaralan
  pagaaralan pinagaaralan
study X will study X is/are studying X
ipaglinis ipapaglinis ipinapaglinis
  ipaglilinis ipinaglilinis
clean for will clean for is/are cleaning for
ikatakbo ikakatakbo ikinakatakbo
  ikatatakbo ikinatatakbo
cause to run will cause to run is/are causing to run

(See reference to this example in CONCLUSION of paper.)

One question to be asked concerning reduplication is whether it is an anticipatory or a perseverative gesture, whether for example in the forms above containing tatakbo, the syllable added by the process is the first or the second of the ta's. A tentative answer to this question with respect to Tagalog,[Note 8] based on admittedly skimpy evidence including speaker intuition, is that it is anticipatory. In anticipation of the root takbo and the morphological requirement that it be reduplicated, one first utters ta to fulfill the requirement, then follows through with the full root. If borne out by stronger evidence, this would indicate that "reduplication" is not the most apt term for the process. But what might replace it as being more descriptive? "Presaging?" [back to outline]


Although technically every process constitutes a modification, the term is usually reserved for changes in the phonetic substance that leave one form with no more and no less than the other­­for changes in which the number of segments remains constant.[Note 9] [back to outline] [back to AFFIXATION]

Vowel Modification

A morphological process may consist of the substitution of one vowel quantity for another, or a particular vowel quality for one or more others. In the latter instance, especially when several qualities are substituted for, the substitute vowel can be seen as the "target" of the gesture. Latin perfect verb-stem formation, for example, involved several processes, singly or in combination. Included among them, in addition to suffixation (of -s or -u/v) and initial CV reduplication, were both the lengthening of vowels of every quality, and the substitution of long e: for either a or i.[Note 10] To examine these processes is to develop a feel for the complex of gestures whereby one made a verb perfect in its aspect. [back to outline]

Ablaut and Umlaut

These two terms are of German origin. Ablaut was first used by German linguists to refer to vowel alternations (also called gradations) of the sing-sang-sung variety inherited from Indo-European, which from their origins have been grammatical signaling devices and thus constitute pure examples of vowel modification as a morphological process.[Note 11] Umlaut, on the other hand, although today largely indistinguishable from ablaut, had its origins in Germanic languages as a phonological process , whereby root vowels assimilated to a high-front suffix vowel. When the suffix vowel was later lost, the root vowels became the sole remaining marks of the morphological property originally signalled by the suffix. Thus the mouse-mice alternation, an example of umlaut, can be explained schematically as follows:

  Singular Plural
Germanic [mu:s] [mu:s-i]
assimilation -- [my:s-i]
loss of suffix -- [my:s]
Old English -- [my:s]
Middle English [mu:s] [mi:s]
Great Vowel Shift [maws] [mays]
Modern English mouse mice

The alternations of man-men, tooth-teeth, and foot-feet have similar histories, which although they can be reconstructed, are today lost to the everyday language user. Thus the results of umlaut are largely indistinguishable from those of ablaut. It may be that a contrast is still felt in English between the largely front-back or high-low alternations of ablaut and those of umlaut, which are back-front,[Note 12] but this would be the only legacy of the separate origins. [back to outline]

Vowel Reversal

Whereas ablaut and umlaut associate a given vowel quality with a given grammatical feature, the Romance languages in their formation of the Subjunctive in contrast with the Indicative exhibit a type of vowel modification in which either of two qualities (front (i or e) or low back (a)) may be associated with either the Subjunctive or the Indicative. What is important is that they be reversed one from the other. Which quality is associated with which grammatical category depends on the verb class, and is opposite for the two classes. Here is an example from Spanish:[Note 13]

'buy' Indicative Subjunctive
1S compro compre
2S compras compres
3S compra compre
1P compramos compremos
2P compra'is compre'is
3P compran compren
1S como coma
2S comes comas
3S come coma
1P comemos comamos
2P come'is coma'is
3P comen coman

Here it is impossible to identify a conventional subjunctive morpheme. In answer to the question as to which vowel quality correlates with the subjunctive, the answer must be neither or both, or better yet, that the wrong question is being asked. Comparison with a toggle switch, or with the jump-ball arrow of college basketball, may be in order. The initial state is arbitrary: either quality may occur in the indicative, depending on the verb class; there are occasions when a switch to the other state is called for: the subjunctive is formed by going to the other quality. [back to outline]

Consonant Modification

In English, voicing of final fricatives is used to convert certain nouns to verbs (sometimes with accompanying vowel modification):

Noun   Verb
bath bathe
wreath wreathe
house house
grief grieve
belief believe
half halve

[back to outline]

Tonal Modification

A number of African languages use tonal modification for verb inflection, according to Matthews.[Note 14] He cites the following example from Lumasaaba (a Bantu language from East Africa), in which "a morphological distinction may regularly be carried by tone alone":

'he saw' 'Near Past' 'Perfect'
  _ ^ ^ [a:Bo:ne] ^ \ _ [a:Bo:ne]

(where ^ = high tone, _ = low tone, \ = falling tone, and B is an implosive bilabial stop)

[back to outline]

Stress Modification

Here again English furnishes an example in disyllabic noun-verb pairs, sometimes with accompanying vowel modification:

  Noun Verb
Primary stress on: First syllable Second syllable
  récord recórd
  cóntrast contrást
  pérmit permít
  pérvert pervért
  tránsport transpórt
  cónvict convíct


[back to outline]


Total modification is included here not as a regular process but in the spirit of exhausting the logical possibilities. Its occurrence is by its nature sporadic and idiosyncratic within inflectional categories established by regular processes, as for example the English past inflection, where it is the regular ablaut and -ed past tense forms that establish the category in which suppletive went occurs as an isolated example:

English: Present Past
  strike struck
  string strung
  fling flung
  go went
  play played
  miss missed
  need needed

As Comrie points out, widespread use of suppletion soon becomes a practical impossibility, since a highly synthetic language whose morphology was totally suppletive would approach having each sentence "totally and unsegmentably distinct from every other sentence."[Note 15] [back to outline]


This type is reserved logically for instances in which shorter forms can be formed regularly from longer ones, but not vice versa. Loss of phonetic substance is a common occurrence in language change. As long as there are longer alternants surviving in other contexts, it may be possible­­depending on the circumstances­­to consider the longer forms as basic and the shorter ones as alternants in a special environment. One supposed example of subtraction as a valid morphological process, the formation of masculine adjectives in French from their longer feminine counterparts, has been challenged on this basis. However, there are unequivocal examples in which subtraction is the sole process operating in the creation of inflected forms, as for example the vocatives found in a number of Western Austronesian languages, where an initial syllable may be dropped, or the initial consonant of a monosyllable.[Note 16] Thus (Robert Blust, personal communication) an anthropologist named Jack Prentice who studied the Muruts in Sabah reports that they would refer to him as "John" but call him by shouting "Ohn." [back to outline]


Our catalogue would now seem to be exhaustive. One might ask, however, about seeming instances of parasynthesis, in which a prefix and a suffix are attached simultaneously, neither occurring without the other­­what have sometimes been called "discontinuous morphemes." Thus, for example, in Italian:

brutto 'ugly' imbruttire 'to make ugly'
rozzo 'crude' durizzare 'to make less crude'
vecchio 'old' invecchiare 'to age'[Note 17]

I will take the position that this is a special instance of affixation, differing from prefixation and suffixation only in that the simultaneity of the separate gestures involved may introduce special effects, being both anticipatory and perseveratory rather than one or the other.

Another supposed special process is that involved in many Afro-Asiatic languages, where roots may be seen as consisting of consonants, with variation of vowels accomplishing both derivation and inflection. Thus in probably the most famous example k-t-b is the root underlying modern Egyptian Arabic forms such as the following: [Note 18]

je-ktub 'he is writing' ma-ktab 'place for writing, study'
katab 'he wrote' ma-ka:tib 'places for writing, studies'
kita:b 'book' ka:tib 'writing (person)'

Our position is that each separate lexeme­­the verb 'to write', the noun 'book', the derivatives 'writing (person)' and 'place for writing, study'­­must have a basic, uninflected form that includes vowels, and that vowel modification is a process that figures (among others such as prefixation) in producing the inflected forms in its paradigm. There is no fundamental difference between roots in languages such as these, and the constancy of s-ng in sing, sang, sung, and song.

Finally, there is a whole area of additional resemblances that have generally been considered to be beyonds the bounds of morphemic analysis. These fall into three main types. [back to outline]

Former resemblances: etymology.

Only someone who probes into the history of words is aware that disease was once dis-ease. The word has changed semantically and phonologically­­though not in spelling­­ to such an extent that its original relation to ease is no longer felt by the typical speaker of the language, for whom it is a morphologically simple, with no partial phonetic-semantic resemblances to any other word.

For a word like cupboard, the spelling does serve to keep us aware of its earlier morphological complexity, that it was once a compound. On the other hand, current spelling fails to reveal to us the relation of buxom to handsome, winsome, etc. [back to outline]

Limited resemblances: coincidences?

What should be made of the fact that two of the directions in English end in -st and that the other two end in -th?[Note 19] In looking at Latin ama-t (3S) and ama-nt (3P) 'love', shall we segment the -n- as a pluralizer, even though it recurs nowhere else? Are data like the following sufficient evidence to posit a Turkish a possessive plural person morpheme -iz?

evim 'my house' evimiz 'our house'
evin 'your (sg.) house' eviniz 'your (pl.) house'

Such questions come from the gray area some include within morphological analysis, but which others consider to be "subanalysis."

Note that even if such resemblances (and the former resemblances of etymology) are included, no new processes are introduced. The questions are generally ones having to do with the segmentation of affixes (or members of compounds). However, one final type is not so easily disposed of. [back to outline]

Vague resemblances: phonesthemes.

Following is a sampling of English phonesthemes, gleaned primarily from the writings of Dwight Bolinger.[Note 20] Affix-like in appearance, it may be asked how they differ from conventional derivational morphemes. For one thing, with the exception of -ician (which may better qualify as a true derivational affix), it can be noted that the remainder of the words containing them do not otherwise occur.[Note 21] If examples like the following were considered true morphemes, the remainders of their words would be cranberry morphs (sometimes referred to as hapax logomena). For this reason, and because some linguists feel their resemblances in sound are too vague and hazy "to carve out a definite morpheme," they are identified as a separate category to which has been given the name phonestheme.

-ag lag, flag (v.), sag, bag, drag 'droopy' and 'flabby'
-amble amble, ramble, scramble, shamble 'locomotion'
-ash gnash, crash, trash, smash, bash 'breaking' or 'fragments'
-ee shivaree, jamboree, jubilee, spree, whoopee, whee, free, yippee, glee 'absence of restraint, carefree abandon'
-eazy sleazy, greasy, queasy 'the sleaze factor'
-icious delicious, luscious, scrumptious, licentious, nutritious, voluptuous 'sensual indulgence, appealing to the senses'
-ician logician, beautician, mortician, magician, dietician, syntactician 'practitioner'
ob- objectionable, obnoxious, obtrusive 'objectionable'
-ump hump, rump, bump, crump, lump, stump, mump, chump, thump, slump 'heavy masses',
'heaviness and bluntness'
-ush lush, slush, gush, flush 'moist' and 'oozy'
-utter mutter, stutter, sputter, splutter 'utter'
-urry scurry, worry, flurry, blurry, hurry 'haste' and 'confusion'
-rl twirl, whirl, birl, furl, gnarl 'circular motion'
gl- glitter, glow, glare, gleam 'visual phenomena'
fl- flitter, flow, flicker, flurry 'phenomena of movement'
cr- crawl, cringe, creep, crumple 'bent'
sl- slip, slide, sled, slime, sledge, sleigh, sluice, slush 'horizontal movement,
lack of traction'
sw- swoop, swell, swish, swoon, swagger 'flourish', 'sweep'
tor- torture, torment, tortuous 'violence'
tur- turbulent, turment, turmoil 'agitation'
tw- twist, twirl, tweak, twine, twinge 'twisting motion'
vi- vituperative, vitriolic, vindictive, vicious 'intense ill-temper'

Bolinger notes that:

Given a particular word for a particular thing, if other words for similar things come to resemble that word in sound, then no matter how arbitrary the relationship between sound and sense was to begin with, the sense is now obviously tied to the sound. The relationship between sound and sense is still arbitrary as far as the outside world is concerned (and would appear that way absolutely to a foreigner), but within the system it is no longer so. [Note 22]

There can be little doubt that a process is involved, and that it is ultimately a morphological process, one that can figure in the production of new words. Bolinger, for example, attributes hassle to tussle, bustle, and wrestle. How the process relates to prefixation and suffixation we will leave for the subject of another paper. [back to outline]


We have noted in this paper a variety of ways in which new words or inflections can be formed that do not involve either prefixation or suffixation. Each deserves further study as a process in its own right, unrelated to the concept of morpheme and other bits apparatus built upon the assumptions of affixation. As a result, some of these latter will need to be abandoned. For example, Tagalog uses infixation and reduplication to form the contemplated aspects of verbs. (See the example above.) There is currently a debate among linguists as to whether these aspects are inflectional or derivational, but if they prove to be inflectional, it would make no sense to say concerning these word-internal processes that "Inflection is always peripheral with respect to derivation." [Note 23][back to outline] [back to list of essays]


  1. I am indebted to Bob Blust, Vanna Condax, Greg Lee, Ken Rehg, Laurie Reid, and Stan Starosta for helpful comments on an earlier version of this paper, which was written for use with my class in morphology. Responsibility for errors and infelicities that remain is mine alone, however. [back up]
  2. Compounding is be excluded from the purview of this paper. [back up]
  3. Charles F. Hockett, "Two models of grammatical description," Word 10:210-31 (1954), included in Martin Joos, ed., Readings in linguistics, American Council of Learned Societies, 1957, pages 386-99 (see pages 393-394 in the latter). [back up]
  4. Greenberg notes, with reference to phonological processes, that some such as the dissimilatory "sporadic changes of sonants and sibilants involve associative interference at a distance and constitute a 'higher-level' psychological process than the more directly physiological adjustment of neighboring sounds that underlies regular conditioned sound changes." (Joseph H. Greenberg, Anthropological linguistics: An introduction, Random House, New York, 1968, pages 107-109.) We will consider the levels involved in the gestures of morphological processes to be an open question. [back up]
  5. Our choice here is the one that seems less clumsy, and no less descriptive than the others, at least at the moment. [back up]
  6. The equivalent within inflection is known as syncretism, the partial suspension of a paradigmatic contrast. [back up]
  7. Our source for Tagalog is Paul Schachter and Fe T. Otanes, Philippine reference grammar, University of California Press, 1972, pages 283-409. [back up]
  8. It may well be that reduplication in some languages, especially where it comes late in the root, as with the final CVC reduplication of Marshallese referred to above, is perseverative. This may correlate with the directionality of linking proposed by A. Marantz, On the nature of grammatical relations, Cambridge, MA: MIT Press, 1984. [back up]
  9. With the possible exception of the effects of lengthening, which is sometimes viewed as gemination, or the creation of a second identical segment. [back up]
  10. Both the lengthening and change in quality have their origin in the Indo-European ablaut discussed below. [back up]
  11. That their reflexes in daughter languages have been complicated by phonological interaction with laryngeals that were later lost as such should not obscure the fact that ablaut was originally purely a morphological process. For a simple and clear account of de Saussure's reconstruction of the laryngeals, see Bruce L. Pearson, An Introduction to linguistic concepts, Knopf, 1977, pages 71-74. [back up]
  12. If one starts in each case from the unmarked grammatical property. [back up]
  13. I am indebted to P. H. Matthews, Morphology: An introduction to the theory of word structure, Cambridge University Press, 1974, pages 139-40 not only for insight into this process, but for the inspiration for writing of this paper gained from Chapter VII, which bears the same title. [back up]
  14. Same work, page 133. [back up]
  15. Bernard Comrie, Language universals and linguistic typology, University of Chicago Press, 1981, page 45. [back up]
  16. Robert Blust, "Proto-Western Malayo-Polynesian Vocatives," Bijdragen tot de taal-, land- en volkenkunde 135:205-251, 1979 (Koninklijk Instituut voor taal, land- en volkenkunde, Leiden, The Netherlands). [back up]
  17. These examples are from Sergio Scalise, Generative morphology, Foris, Dordrecht, 1986, page 147. [back up]
  18. Data from Leonard Bloomfield, Language, Holt, New York, 1933, pages 243-244. [back up]
  19. Posed by P.H. Matthews, Morphology: An introduction to the theory of word structure, Cambridge University Press, 1974, page 15. [back up]
  20. A number of which are included in Dwight L. Bolinger, Forms of English: Accent, Morpheme, Order, Harvard University Press, Cambridge Mass., 1965. [back up]
  21. Or in some cases, other phonesthemes, as for example slush and sleazy, which seem to be composed of combinations of phonesthemes. [back up]
  22. Dwight Bolinger, Aspects of language (2d ed.), Hacourt Brace Jovanovich, Inc., New York, 1975, pages 218-219. [back up]
  23. Sergio Scalise, Generative morphology, Foris, Dordrecht, 1986, page 103. [back up]