TA: John Kirn
Date: April 20, 1986
Beginning late in the second year of life, and speeding up in the third through fifth years, children undergo vast changes in their use of language, venturing from single word holophrases and paired words to actual grammatical morphemes and the application of morphological rules. The order of acquisition of these rules seems to vary little within languages, and follows very similar patterns. Often the correct uses of a certain rule will rise slowly and then, in a few months, jump rapidly, and then level off again (Anisfeld, 1984). Once a child develops a rule, he or she almost invariable overgeneralizes it to irregular aspects of the language—using the ’ed rule for past tense to say ‘I goed’ for instance. There is even evidence that bilingual children will overgeneralize from one language to another (Anisfeld), combining morphemes from each language.
In this paper, the development of fourteen different English morphemes will be discussed, in order of their acquisition by children. Where possible, the pattern of acquisition will be discussed, as well as certain variants which may be used by children in place of these rules before fully using the rule. Much of this information comes from Moshe Anisfeld’s book, Language and Development From Birth to Three, and the book Structure and Development in Child Language: The Preschool Years, by Marion Potts, et. al. The latter book reports on a very fine study of morpheme acquisition from ages three to five.
The morphemes to be studied here, in the generally accepted order of acquisition (H. Gardner, 1982, from Roger Brown, 1973) are as follows. The final three (Future Conditional, Passive, and Past Progressive) are placed in order through use of the data found in the Potts, et. al., book. The ages at which any specific child gains morphemes varies greatly. The order of acquisition is much more constant than the age of acquisition.
- Present Progressive
- Past Irregular
- Possessive Inflection
- Uncontractible Copula
- Past Regular
- Third Person Regular
- Third Person Irregular
- Uncontractible Progressive Auxiliary
- Contractible Copula
- Contractible Progressive Auxiliary
- Future Conditional
- Past Auxiliary
The Present Progressive: The present progressive involves the present tense of the auxiliary verb ‘be’ plus a verb plus the stem ‘ing’ (I am coming.) When the child first evidences this, the auxiliary is left out (I coming). In this form, it is one of the first morphemes learned.
Most verbs that take the progressive refer to processes, and most verbs that do not take the progressive refer to states. This is interesting, because despite the existence of verbs which this rule does not apply to, children do not seem to overgeneralize. Brown (Anisfeld, from Brown, 1973) considered two possibilities for this, finally settling for the latter. His first was that children noticed the difference between process verbs and state verbs, and so did not overgeneralize to these verbs. He rejected this, not believing children to be that sensitive, and decided that children must learn specifically which verbs take the progressive. Anisfeld came up with an alternative explanation, and used observations of his own son Shimon. His view is that there are simply not enough ‘state’ verbs “in the child’s realm of interest” for overgeneralization to be seen. Shimon, an English-Hebrew bilingual child, was observed overgeneralizing the progressive rule from English to Hebrew (marbitzing, ‘hitting’ lishkaving, ‘lying down’).
Plural Inflection: An inflection is a morpheme which acts as either a suffix or prefix, and in this case the inflections in question are ‘s’ and ‘es’. Before acquiring the plural inflection, especially the ‘es,’ the child often simply uses the noun alone and leaves interpretation to context. It usually takes longer for the child to use the ‘es’ inflection than to use the ‘s’ inflection. The child will, for example, “say one box-two box and one peach-two peach even when they say one ball-two balls” (Anisfeld). This is generally assumed to be caused by the fact that nouns requiring ‘es’ already sound like plurals. They already have an ‘s’ sound, or something similar, like ‘ch.’ Once the child acquires the ‘es’ ending, however, it is overgeneralized also, even to the point of combining the two pluralization inflections. The child often vacillates between both forms also (footes, footses, foots).
By three years of age the child has already acquired half of the ‘s’ inflection—that pronounced ‘z’ (lives, guitars)—and has acquired the ‘es’ inflection by age five.
Past Irregular: Strangely enough, the past irregular is acquired before the past regular. The reason for this is unknown, but it may be simply that the concept of a past tense arrives before the determination of the past tense rule. It would indeed be odd for the child to be able to determine a past tense rule without any concept of a past tense. Without the rule but with the concept, the child will still be able to use irregular past tenses, since these are still single morphemes—went, broke, told. The appearance of past irregular may be an increase in vocabulary. The child does forsake these irregular forms for a while after forming the past regular rule, saying goed, breaked, and telled, for example.
Possessive Inflection: The possessive inflection ‘s,’ as in Frank’s, Gladys’s, and the Bob’s, is very similar to pluralization, having the same sounds, although the spelling remains the same in this case. According to Cazden (Potts, et. al., from Cazden, 1972) the acquisition of this inflection follows a certain pattern. Before acquiring the rule, the child merely juxtaposes the words, with the possessor first (baby toy). While forming the rule, the child will often use the inflection but not the possessed (baby’s), but will otherwise fall back on the uninflected form. Finally, the child will inflect correctly, whether the possessed is mentioned or not (baby’s, or baby’s toy).
Uncontractible Copula: This denotes the use of the verb ‘to be’ as the main verb, without contraction, such as ‘The books are on the floor.’ Before acquisition of this rule, children often simply omit the copula, retaining meaning of the sentence in similar form—‘The books on the floor’, or ‘Books down floor’. Also, younger children may acquire this form before learning the rules by which it specifies number, and might say, for example, ‘The books is on the floor.’
Past Regular: The regular past tense in English is formed by adding ‘ed’ to the end of verbs (gambled, survived). The most common verb form used when the past regular is not but should, is the uninflected verb (gamble, survive), but it is possible that the present tense (gambles, survives) and progressive (is/was gambling/surviving) are used to make up for lack of a past regular. They can even replace it with an irregular form—change ‘She washed it’ to ‘She put it in the water’ (Potts, et. al.)
Third Person Regular: Like the plural and possessive inflections, this utilizes the inflections ‘s’ and ‘es’. Like the former two inflections, the child is least likely to use the rule for words ending in ‘s’ or ‘ch’. When not using this rule, children use the uninflected verb, or even inflect the verb to past tense (Potts, et. al.)
Third Person Irregular: This follows right on the tail of the third person regular. The third person regular and irregular come in the opposite order of the past tense, for which the irregular comes first, and then, much later, the regular. It may be that by now, after learning plural and possessive inflections, the child finds adding the ‘s’ or ‘es’ easier to learn. It could also be that after learning the inflection for past tense the child ‘expects’ an inflection to mark verbs.
Uncontractible Progressive Auxiliary: The progressive auxiliary involves a form of ‘be’ plus a verb plus ‘ing.’ The uncontractible form involves the use of uncontractible forms of ‘be’, such as ‘was.’ Obviously, the uncontractible progressive auxiliary must come after the uncontractible copula, which involves the verb ‘to be,’ and the present progressive, which involves adding ‘ing’ to verbs. If the child does not use the full progressive auxiliary, the auxiliary is most often left out, leaving basically the present progressive (I typing, rather than I am typing).
Contractible Copula: This denotes the use of the verb ‘to be’ with contractions (He’s walking) and is a forerunner of the contractible progressive auxiliary. Otherwise, it is very similar to the uncontractible copula, and the child can easily avoid the contraction by using the uncontractible—‘He is walking.’
Contractible Progressive Auxiliary: This follows immediately the acquisition of the contractible copula, and may not even require a new rule to be formed. Once the child uses contractions with the copula alone, he or she may automatically begin using them in the progressive auxiliary.
Future Conditional: This involves an if-then, with the modal (will, might) plus get or become, plus the verb (If the bird falls, it will get hurt.) Before acquired, children leave out the modal (it get hurt), use the progressive auxiliary plus go (it going to get hurt) or simply leave out the modal or auxiliary altogether (it get hurt).
Passive: The passive seems to be a hard concept for young children. Fraser, Bellugi, and Brown (Potts, et. al., from Fraser, et. al., 1963) found that very few 3 year olds even understand the spoken passive. By age five, however, they can be seen to use it correctly. Before this, the child can use active forms (the clothes have to dry, rather than the clothes have to be dried). Very young children, around three years old, however, seem simply to use an uninflected verb (the clothes have to wash, rather than the clothes have to be washed). The passive involves two previous rules. The child must know the copula, and the child must know the past tense.
Past Progressive: The past progressive is learned quite late, and like the passive and future conditional involves many concepts. The past progressive is a new use of the progressive, with the past tense of the copula—‘He was walking.’ The most common way of getting around using the past progressive was simply using the past tense (he walked).
After age five, many, though not all, aspects of language become relatively constant. The child has “close to adult speech competence” (James W. Kalat, 1984). A study of four to six year olds in prepositional phrase structure found adult-like language by age four (Helen Goodluck, 1986). In a study by Joan S. Klecan-Aker (1984), certain uses of clauses and verbs were studied in stories by sixth graders and ninth graders, and no difference was found. Complexity of nouns and vocabulary were not studied, and it would not be surprising to find quite a massive rise between those grades for those parameters. However, that has little to do with the seeking of rules that goes on before age five.
That children are actively seeking rules is readily apparent. In a Harvard study (from Anisfeld) the number of verbs in 49 hours of recorded speech to a child (Eve) was determined. Eve heard more irregular verbs than regular verbs, and the irregular verbs occured with much more frequency—over two thirds of the verbs she heard were irregular. This is most probably very common. However, like most children, Eve acquired the rules for inflection of verbs (in this case, past tense). “Eve formed a rule on the basis of the regular verbs and ignored the irregular verbs. This is because the regular verbs exhibit a pattern, and the irregular verbs do not; and Eve looked for patterns.” (Anisfeld)
Not only do children actively seek patterns, they seek very general patterns. Almost every, if not every, rule which children acquire is overgeneralized. Children say goed, footses, and even cross-language generalizations. Besides generalizing the English progressive rule to Hebrew, Shimon (Anisfeld) also generalized Hebrew plurals to English, saying boyim (boys), pine conim (pine cones), and wash clothim (wash clothes).
The acquisition of syntax requires that the child approach the language as an independent system, not merely a means of expression (Anisfeld). The acquisition of inflections demonstrates this. The choice of which allomorph (a variant of a morpheme) to use in plurals, present tenses, and possessions (‘s’, ‘z’, or ‘iz’) is determined solely by the form of the stem. It has nothing to do with meaning. In a study of Russian children (Popova, 1973, from Anisfeld) gender mistakes show this also. Russian verbs must agree with the subject nouns in gender. Feminine nouns are generally marked with an ‘a’ suffix. There are, however, biologically masculine nouns which are masculine in the language, yet have the ‘a’ suffix. These include ‘papa’ (father), ‘dyadya’ (uncle), and ‘dedushka’ (grandfather). The verb is marked as feminine with the same ‘a’ suffix. However, bioligically masculine nouns should have a masculine verb form, despite the ‘a’ suffix. What Popova found was that “Children who have basically mastered correct agreement make errors with words such as papa, dyadya, dedushka… often using the feminine form of the past tense verb: ‘Dyadya sidela na loshadke’; ‘Moy papa zabolela’; and so on…
‘This is not merely another example of overregularization, but rather evidence that children really appreciate the formal workings of language. When faced with a referential formulation of a generalization and a formal formulation, the children Popova studied chose the latter.”
However, meaning does have some relevance. In a pluralization study, it took from two to six weeks from acquisition of the plural rule for children to form plurals for analogous made-up nouns. Block/blocks vs. bik/biks, for example. This difficulty with artificial nouns can last up to the age of six. (Bryant & Anisfeld, 1969; Berko, 1958, from Anisfeld).
Since children have such a range of rates of acquisition, what affects this rate? Speech of those around the child is an obvious choice. However, it has been found that only a few properties of outside speech directed towards the child affects the child’s language, and only affects certain aspects of the child’s language. Erika Hoff-Ginsberg (1986) correlated various structural aspects of maternal speech with various aspects of child language growth and found three aspects of maternal speech which correlated highly with child language growth. Only one of these correlations lasted throughout the study. All of these were positive correlations. The lasting effect was that noun phrases per utterance in the mother’s speech correlated .50 with the child’s noun phrases per utterance and words per noun phrase. This correlation was significant at the .01 level.
Hoff-Ginsberg also correlated the functional properties of maternal speech with child language growth, and found another lasting effect. Number of acknowledgments of declaratives from the mother correlated negatively (-.60) with noun phrases per utterance, and negatively again (-.56) with auxiliaries per verb phrase.
So few specific aspects of maternal speech affect the child. However, some general aspects do seem to. Simpler parental language, such as the use of short sentences and concrete nouns, does correlate with greater linguistic advancement in the child (Furrow, Nelso, and Benedict, 1979, from Gardner).
No doubt the environment does affect language acquisition in some way, but how does the child go about gaining language, and more basically, morphemes? There are many theories attempting to explain the acquisition of morphemes, especially the order of acquisition. The major ones follow.
Cumulative Complexity: Roger Brown (1973, from Potts, et. al.) developed the theory of cumulative complexity, in which the cumulative semantic features of a morphological structure are analyzed. For example, the plural includes the feature of number. The third person regular consists of number and earlierness. The uncontractible progressive auxiliary consists of temporal duration, number, and earlierness. So going by the cumulative semantic features, the plural should be acquired before the third person regular, which should be acquired before the uncontractible progressive auxiliary. (Frank Keil, 1986) Brown does this for transformational complexity also. The same rationale holds, using transformations rather than features. Brown is unsure about which of the two, transformational or semantic complexity, is more useful, and indeed both often predict the same things. There is a case of two languages where bilingualism occurs and grammatical complexity is more important than semantic complexity, in Northern Yugoslavia. The locative (in, in, under, etc.) is much simpler grammatically in Hungarian than in Serbo-Croatian, and the bilingual child uses the locative much earlier in Hungarian (Keil). That it should be the grammatical complexity which counts goes along with Anisfeld’s theory that the child approaches language as a system independent of its use for expression.
Transformational Scope: McNeill (1970, from Potts, et. al.) proferred that it is the scope of a syntactic structure that determines the order of acquisition. For example, the inflection suffixes ‘s’ and ‘es’ are used for the plural, the possessive, and the third person. The plural only relates to the single word. The possessive relates two words within a phrase, and the third person relates two phrases—the noun phrase and the verb phrase. Therefore, the plural should be acquired before the possessive, which should be acquired before the third person.
Upon a close look, it seems that these two theories are quite similar. Any morpheme which has a high grammatical complexity almost by definition has a large transformational scope. Any morpheme with a large transformational scope is probably going to be grammatically complex. However, there are differences. Transformational scope applies to rules more than to morphemes. These rules happen to apply to morphemes, but also apply to basic sentence structure. The more general the rule (the fewer exceptions across sentences) and the fewer structures the rule covers within sentences, the earlier it will be learned. Cumulative transformational complexity, however, thinks in terms of the grammatical complexity of the morpheme—what the morpheme affects within the sentence.
Since the amount of speech, above a minimum for language development, does not seem to correlate with morphological development except in very specific cases, it might seem that generality across sentences shouldn’t make much difference. However, generality combined with low complexity could do so. Recall that simple and concrete language does correlate with language development. An experiment is required to break apart morphology and generality.
To determine if transformational scope is a more general form of cumulative transformational complexity, we would have to study a morphology where complexity can be held constant, and vary the generality. It would be best if complexity could be held to a minimum and generalization to a high, although not a maximum. At first glance, pluralization seems to fill the bill. There are three different forms. The allomorph pronounced ‘iz’ is much less general than the other two, and does indeed seem to be acquired last. However, there is a catch. That allomorph is associated with nouns which have ending phonemes similar to the pluralization ending.
What else is there? Actually, the pluralization paradigm might work after all, but using the inflection spelled ‘s’. It has two pronunciations, ‘s’, and ‘z’. Each must almost surely have the same grammatical complexity. If, as is probably the case, one is more general than the other, a large-scale study involving many chances of applying each, given to children around eighteen to twenty-five months, might uncover a difference in acquisition time. This difference will probably be small, so a large pool, both of children and of pluralization questions, would be needed in order to gain statistical significance.
The pretest should consist of randomly sampling parental speech to eighteen to twenty-five month old children to determine which of the two inflections is most common. It should also be determined if the same inflection is the most common for each of the children in the pretest. If this is not the case, the main experiment will have to control for that also. I will assume, however, that one is generally most common. Taking a random sample of twenty-five nouns from the Random House dictionary, I found four which use the ‘s’ inflection, and twenty-one which use the ‘z’ inflection. I will assume that the ‘z’ inflection also would win out in the pretest, and probably by a very good margin.
Probably fifty children would suffice for the main test. This may seem like a lot, but I expect a lot will be required to gain the statistical significance necessary to pull out a small difference. The children, at the start, should all be about eighteen months old. The children should be tested once a week (possibly 10 children a day), and their progress monitored, until all have mastered plurals of both inflections. It is very unlikely that this would take over six months. Then, each child’s progress should be studied to see which allomorph was gained first. Preferably, this allomorph should have a higher percentage at every or almost every test.
Transformational scope would predict that the more general pluralization allomorph is acquired sooner. Cumulative transformational complexity would predict that the two allomorphs are acquired at the same time. I doubt that the more general allomorph is acquired second, since the least general allomorph (iz) is acquired last, and I don’t know what that would prove, if anything. However, if the more general allomorph is acquired first, then generality, even small variations, do affect the order of acquisition. If no difference is found, then generality, above a certain required generality for rule learning, doesn’t affect order of acquisition.
Both of these results are quite plausible from the evidence. The past irregular is not very general at all, and it is gained before the past regular. However, others do seem to move in order of lowering generality. This could be simply that the less general a rule is, the more complex it is. But it could also be that the more complex a rule is the less general it is.
- Abrahamsen, Adele A., Child Language: An Interdisciplinary Guide to Theory and Research, University Park Press, Baltimore, Maryland, 1977;
- Anisfeld, Moshe, Language Development From Birth to Three, Lawrence Erlbaum Associates, Hillsdale, New Jersey, 1984;
- Brown, Roger, Development of the First Language in the Human Species, from Readings in Developmental Psychology, 2nd Edition, Judith K. Gardner, editor, Little, Brown & Company, Boston, Massachusetts, 1982;
- Donaldson, Margaret, Children’s Minds, W. W. Norton & Company, New York, New York, 1979;
- Gardner, Howard, Developmental Psychology, 2nd Edition, Little, Brown & Company, Boston, Massachusetts, 1982;
- Ginsberg, H. & Opper, S., Piaget’s Theory of Intellectual Development, 2nd Edition, Prentice-Hall, Inc., Englewood Cliffs, New Jersey, 1979;
- Gleitman, Henry, Basic Psychology, W. W. Norton & Company, New York, New York, 1983;
- Golinkoff, Roberta M., editor, The Transition from Prelinguistic to Linguistic Communication, Lawrence Erlbaum Associates, Hillsdale, New Jersey, 1983;
- Goodluck, Helen, Children’s Knowledge of Prepositional Phrase Structure: An Experimental Test, from Journal of Psycholinguistic Research, Vol. 15, No. 2, 1986;
- Gould, James L., Ethology: The Mechanisms and Evolution of Behavior, W. W. Norton & Company, New York, New York, 1982;
- Hoff-Ginsberg, Erika, Function and Structure in Maternal Speech: Their Relation to the Child’s Development of Syntax, from Developmental Psychology, Vol. 22, No. 2, 1986;
- Kalat, James W., Biological Psychology, 2nd Edition, Wadsworth Publishing, Belmont, California, 1984;
- Keil, Frank C., Developmental Psychology: Lecture Notes for Psychology 209, Ithaca, New York, 1986;
- Klecan-Aker, Joan S., A Study of the Syntactic Skills of Normal Middle-School Males and Females, from Language and Speech, Vol. 27, Part 3, 1984;
- Nelson, Katherine, Individual Differences in Language Development: Implications for Development and Language, from Readings in Developmental Psychology, 2nd Edition, Judith K. Gardner, Eidtor, Little, Brown & Company, Boston, Massachusetts, 1982;
- Nelson, Keith E., Theories of the Child’s Acquisition of Syntax: A Look at Rare Events and at Necessary, Catalytic, and Irrevelant Components of Mother-Child Conversation, from Studies in Child Language and Multilingualism, Virginia Teller, & Sheila White, eidtors, New York Academy of Sciences, 1980;
- Potts, M.; Carlson, P.; Cocking, R.; Copple, C., Structure and Development in Child Language: The Preschool Years, Cornell University Press, Ithaca, New York, 1979;
- Segall, Marshall H., Cross-Cultural Psychology: Human Behavior in Global Perspective, Brooks/Cole Publishing Company, Monterey, California, 1979;
- Yeni-Komshian, G.; Kavanagh, J.; Ferguson, C., editors, Child Phonology, Volume 1: Production, Academic Press, New York, New York, 1980;