Morphological Strategies in Language Acquisition

49
Morphological Strategies in Language Acquisition by L.S. and J.S. Paper submitted in partial fulfillment of the requirements for a graded credit for the course “Psycholinguistics” in Winter Term 2007/2008 Submission Date: 20/03/08 Approved by: Prof. Dr. J.H. Philipps University Marburg

Transcript of Morphological Strategies in Language Acquisition

Page 1: Morphological Strategies in Language Acquisition

Morphological Strategies in

Language Acquisition

by

L.S.

and

J.S.

Paper submitted in partial fulfillment of the

requirements for a graded credit for the course

“Psycholinguistics”

in Winter Term 2007/2008

Submission Date: 20/03/08

Approved by: Prof. Dr. J.H.

Philipps University Marburg

Page 2: Morphological Strategies in Language Acquisition

Contents

1 Introduction......................................................................................................3

2 Language Acquisition......................................................................................4

2.1 The development from first sounds to first words.............................................4

3 Morphological Development...........................................................................6

3.1 The acquisition of morphemes...........................................................................7

3.1.1 Berko’s wug-experiment 7

3.1.2 Brown’s longitudinal study 8

4 Overgeneralization.........................................................................................11

4.1 Plural Overgeneralization................................................................................12

4.2 Past Tense Overgeneralization.........................................................................14

5 Strategies for past tense overgeneralization on irregular verbs................17

6 Conclusion......................................................................................................25

7 References.......................................................................................................27

8 Appendix.........................................................................................................28

8 Appendix 2

Page 3: Morphological Strategies in Language Acquisition

1 Introduction

This term paper deals with morphological strategies in language acquisition

of children. In a first part, it will be shown that children pass through different stages

in the acquisition of their first language. A few examples will be given to clarify how

and at what age they produce their first sounds and then create their first words.

The main focus of this term paper lies on the child’s ability to apply

morphological rules that come along with the newly acquired words. It will be

discussed in how far children possess different morphological strategies in language

acquisition. Indeed, they are able to apply certain morphological techniques to

produce words and their morphological inflections. Furthermore, Berko’s WUG-

experiment will exemplify the child’s ability to productively use morphological

strategies in language acquisition. The longitudinal study by Roger Brown will be be

presented in order to prove that there is a certain order in the acquisition of

morphemes.

The final chapters will concentrate on the strategy of overgeneralization, a

morphological strategy used by children to inflect words, namely nouns, verbs and

adjectives. This strategy will be presented in detail in the following, as it clearly

illustrates the process in a child’s morphological development. Moreover, in a last

part, the acquisition of the plural and the past tense morpheme will be analyzed in

detail and supplemented by examples taken from the CHILDES database, which

provides detailed longitudinal studies on the development of child language. Here,

the focus will lie on strategies of overgeneralization of irregular verbs.

8 Appendix 3

Page 4: Morphological Strategies in Language Acquisition

2 Language Acquisition

This unit deals with the language acquisition of children. Indeed, children

manage to acquire a first language within a few years, without apparent effort and

without the need for formal instruction [LET: 123]. This part will address the

different stages of a child’s language acquisition. However, these different phases

will only be explained shortly, as general language acquisition is not the main topic

of this term paper.

2.1 The development from first sounds to first words

An infant’s speech production abilities are not apparent before the age of six months.

In fact, the production of speech sounds is limited, until the vocal apparatus

undergoes a change beginning at the age of four months which is not completed until

the age of six years (Guasti 2002:47). Nevertheless, children are by and large able to

hear every phonetic contrast used by human languages; besides, research in speech

perception has demonstrated that infants in their first months of life are able to

discriminate between similar sounds, such as /b/ and /p/ (King 2006:211).

Generally, infants of four to six weeks can only produce vowel-like sounds

(referred to as cooing), cries, and vegetative sounds. The phase of infant babbling

starts at around six to eight months. Infants then repeat syllables with no associated

meaning in terms of the so-called reduplicated babbling which results in syllables

such as gigi, panpan, bababa, gugugu, or dadadada. The other form of babbling is

referred to as non-reduplicated babbling, where different syllables are added

together, e.g. bamido (LET 2005:127). Indeed, children discover their capacity to

refer to objects and events in the world by an “autonomic, physiologically based

vocalization”, which is referred to as a protoword [INT1]. Children start to create

protowords when they are about one year old, e.g. a child might use an utterance like

ioioio for anything that can be rotated (LET 2005:138). These protowords do not

8 Appendix 4

Page 5: Morphological Strategies in Language Acquisition

exist in adult language, instead, they are only used by children to name objects and

persons around them, e.g. the protoword wow-wow might refer to a dog.

The first recognizable words at the age of about ten months mark the

beginning of what is known as the holophrastic stage when infants start to use

single-word utterances to communicate, although they do not yet use any

grammatical or morphological inflections. The word mama, for example, might be a

bid for a mother’s attention or a request for food directed towards the mother (King

2006:212). Like the protoword, these one-word utterances mainly consist of content

words which are grounded in everyday experiences, i.e. people, animals, food and

other objects expressed as bed, dog, mama, papa, spoon, milk. At the stage of two-

word utterances, most children are able to express semantic relations, i.e. an agent-

action as in Daddy run or Baby cry. Nevertheless, there still is limited use of

grammatical morphology, which can be illustrated in a child’s formation of the

possessive, where a child forms an utterance such as Miranda bed rather than using

the correct possessive form Miranda’s bed (King 2006:213).

The so-called stage of telegraphic speech starts at the age of 1;8-2;0 years,

when the child begins to create three- or multi-word combinations with hardly any

functional or inflectional elements, as in Milk all gone and Graham go out. The

following examples illustrate the development from one- to three-or multi-word

utterances:

(1) 1.3 years: More. [reaching for a cookie]

(2) 1.8 years: More read. [holding up book]

(3) 2.1 years: Andrew that off. [wanting to turn off the light]

(4) 2.8 years: He not taking the walls down.

(Clark 1977:295)

By the time the child starts to produce multi-word utterances, as in (4), the

development of morphological rules sets in. Elements such as function words (the

negation not) and inflectional morphemes (progressive –ing, plural –s) can be found

8 Appendix 5

Page 6: Morphological Strategies in Language Acquisition

in the child’s utterances. This aspect of language acquisition will now be dealt with

in more detail.

8 Appendix 6

Page 7: Morphological Strategies in Language Acquisition

3 Morphological Development

Beginning with the topic of morphological development in language

acquistion, it has to be stated that the “two basic functions of morphological

operations are (i) the creation of new words (i.e. new lexemes), and (ii) spelling out

the appropriate form of a lexeme in a particular syntactic context” (Booij 2007: 13).

For the subject of first language acquisition the formation of new words plays a

rather marginal role. Therefore, the main focus in this paper will be on inflectional

morphology. Inflection is the morphological marking on a lexeme resulting in a set

of different grammatical words. The major inflections of the English language are:

with nouns:Plural: boys = boy + sPossessive: Mary’s = Mary + s

with verbs:3rd person singular: works = work + spast tense: worked = work + edprogressive: working = work + ing

with adjectives:comparative: bigger = big + ersuperlative: biggest = big + est

(adapted from Clark, 1977: 23)

Children’s development in inflectional morphology commences at the age of

24-30 months, when they have reached the aforementioned multi-word stage. During

this period, the order of acquisition of function words and inflectional morphemes is

relatively constant among all children, where the present progressive usually appears

first, followed by prepositions, plurals, possessives, determiners, and the past tense

(LET 2005:127). As the grammatical development in the child’s first language

acquisition is of major importance for this term paper, the different morphological

strategies in language acquisition, the mastering of certain function words and

8 Appendix 7

Page 8: Morphological Strategies in Language Acquisition

inflectional morphemes by children will be explained and complemented by

examples in the following part.

3.1 The acquisition of morphemes

When children start forming utterances of several words, they realize that

their language is governed by rules (Aitchison 1989: 119). Therefore, they develop

their own ideas for grammatical rules and, as a result, do not always use grammatical

inflections correctly. At the age of about two years, they increasingly use

grammatical forms which reflect their developing grammatical rule systems. At this

stage, they start using function words and adding word endings to verbs and nouns,

and subsequently pass over to the acquisition of negative sentences and questions

(Clark 1977: 333). Some examples of the function words and inflections children add

to their utterances are prepositions like in, different articles, e.g. the, a, an, different

modals like can and will, the auxiliaries do, be and have, the plural /s/ as in cats and

finally the past tense form /t/ as in worked (Steinberg 2001: 11). However, they still

use incorrect and non-adult-like forms such as *broked or *foots (King 2006: 214).

3.1.1 Berko’s wug-experiment

One of the earliest studies on the child’s ability to productively use

morphological strategies was conducted by Jean Berko in 1958 (Berko 1958). It was

designed to prove that children do have knowledge of morphological rules. The

subjects were 80 children between the age of four and seven years. They were told

English words and nonsense words (nouns, verbs and adjectives) and guided towards

producing a certain grammatical form of these words, namely plural, past tense, third

person singular, possessive, comparative and superlative markings. By using

nonsense words, which were chosen in accordance with phonologically possible

English word formation, it was ensured that the children had never before heard any

inflected form of this word. For examples of tasks from the wug-experiment see

appendix (1).

8 Appendix 8

Page 9: Morphological Strategies in Language Acquisition

All children understood what was requested from them and tried to give the

desired answers. The results were satisfactory: the children knew rules of inflectional

extension which they were able to apply to new words. On some items they did

better than on others and the older children generally had better results than the

younger children.

The children were able to produce various English word forms of the possessive

and plural which required the use of the allomorphs /s/, /z/ and /z/ as well as past

verb forms which required the use of /d/, /t/ and /d/. In all three word forms their

handling was parallel. However, no child was able to productively use the less

common allomorphs /z/ and /d/ on nonsense words. This can be explained with the

low productiveness and rare occurrence of these forms in the language. With real

English words, the children generally had better performance results on the

allomorph /z/ to form possessives, than they had to form noun plurals. Moreover,

they did better on producing possessives and plurals of this form than on applying

the /d/ allomorph for past tense. The best performance could be observed with the

handling of the present progressive morpheme -ing. Since this form has only one

allomorph and is very common it is the easiest of all inflectional morphemes.

Further, the children were also given tasks where they were asked to compound,

to derive or to analyze compound words. The results for these items revealed that

children at that age are not yet able to productively use derivation but rather form

descriptive compounds. (Where many adults would say that the house a *wug lives

in is a *wuggery, the children would either say *wughouse or use real words, such as

birdcage.) Existing compound words are learnt as one lexical item and the children

rarely understand that they consist of two (or more) component parts.

All of the above findings do not only lead to the conclusion that children indeed

process morphological rules and apply them productively to new words, but also

indicate that there must be a general order of acquisition of different morphological

items. In the following study this aspect was investigated further.

8 Appendix 9

Page 10: Morphological Strategies in Language Acquisition

3.1.2 Brown’s longitudinal study

Roger Brown, who studied the linguistic development of children, investigated

the language acquisition of three unacquainted children named Adam, Sarah and Eve

in a so-called longitudinal study. A longitudinal study is the modern approach to

studying language acquisition data, which measure the linguistics progress of the

child at intermittent but predetermined times (LET 2005: 150). In his study, Brown

focused on the acquisition of different function words and inflections in English as a

first language. He examined at what age and in what order children mastered the use

of certain inflectional morphemes and function words. The results for Brown’s

children under investigation are listed below:

Table 1:

Adam Sarah Eve

age 2.6 age 2.10 age 1.9

present progressive plural present progressive

age 3.6 age 4.0 age 2.3

third person regular uncontractible copula

past regular past regular past irregular

uncontractible auxiliary uncontractible auxiliary articles

contractible copula contractible copula third person regular

contractible auxiliary third person irregular uncontractible auxiliary

contractible auxiliary contractible copula

(Table adapted from Brown: 1973: 271)

From this table, it can be inferred that the order of acquisition was similar

across the three children, with present progressive, plural, and past irregular verb

forms appearing first (King 2006: 216). Brown found that all of the three children

gradually learned to use the different morphological inflections. In fact, all children

8 Appendix 10

Page 11: Morphological Strategies in Language Acquisition

whose first language is English seem to undergo the same developmental sequences

in morphological language acquisition. However, there are variations among the

children in terms of speed in which they learned the morphemes. He summed up his

results by ranking a set of 14 morphemes in their average order of acquisition:

8 Appendix 11

Page 12: Morphological Strategies in Language Acquisition

Table 2:Average order of Acquisition

Morpheme / Function Word Type Example

1 Present progressive: -ing: continuing action Mary play-ing2,3 Prepositions: location: in and on The mouse is in the box.

The book is on the table.4 Plural: -s: number, one and more objects /s/, /z/, /z/5 Past irregular: past time came, went, fell6 Possessives: -’s: possession /s/, /z/, /z/, The girl-‘s dog

is big.7 Uncontractible copula be: are, was Are they boys or girls?

Was that a dog?

8 Determiners / articles: definite, indefinite the, a9 Past regular: past time, earlier in time: -ed He walk-ed the dog.10 Third person regular: 3rd person present singular She run-s fast.11 Third person irregular does, has12 Uncontractible auxiliary be: is, were Is Mary happy?13 Contractible copula be: -‘s, -‘re Mary-’s hungry.

They-’re running very slowly.

14 Auxiliary be Contractible: tense carrier Mary-’s playing.(Table adapted from LET 2005: 128, and Brown 1973: 274)

As we have also seen in the example of the wug-experiment, children usually

learn those morphemes that do not follow an exceptional rule more easily. Thus, they

start to learn the first suffix –ing, which can be added to verbs relatively simple.

Nevertheless, there are three different inflections that share the same phonetic

realizations and which are quite difficult to master for children.

This is the case in the regular plural marking and the genitive marking of

nouns and finally the third person singular marking. These endings are all realized in

/s/, which is added as a suffix at the end of a word.

Concluding from Brown’s longitudinal study, the regular plural from /s/ is

acquired fairly early before the acquisition of possessives and the third person

regular, which might be due to the fact that the plural form /s/ can be applied to

nouns relatively easy and is not as semantically complex as /s/, indicating belonging

or possession.

8 Appendix 12

Page 13: Morphological Strategies in Language Acquisition

4 Overgeneralization

The acquisition of morphology is accompanied by problems of morphological

alternation. One strategy children apply very often is referred to as

overgeneralization. As it has already been stated, children go through several stages

in the acquisition of inflections and function words. During this process, they

construct rules for the use of different grammatical morphemes, and often

overgeneralize these in their early phases of acquisition. (Clark 1977: 342 and LET

2005: 138).

These overgeneralizations result, for example, in incorrect word endings such

as adding the past tense –ed to irregular verbs or the plural –s to all nouns, which

leads to incorrect forms in the speech of young children, e.g. *goed and *fishes.

Therefore, overgeneralization implies the replacement of a correct form in adult

language with one that is incorrect.

The development of acquisition of rules for inflection can roughly be divided

into three (or more) phases (Booij 2007: 237). In a first phase the child will learn the

correctly inflected forms for some words by rote and thus use correct regular and

irregular forms such as the plural feet or the past tense form went. In a next step

however, the child will have internalized the rule for regular plural or past formation

and is very likely to produce forms such as *feets or *foots and *goed or *wented.

Only in the last phase, will the child have gradually acquired the rule and learned

which words are formed irregularly and will thus be able to apply the correct forms

for each word.

It has been found, that overgeneralization describes a normal and natural state

in the child’s development of language acquisition. Indeed, “Overgeneralization in

certain stages of language acquisition nicely reveals that discovering the regularities

is part and parcel of language acquisition” (Booij 2007: 237).

In the following, the children’s productive knowledge of English

morphology will be investigated on examples of plural and past tense

8 Appendix 13

Page 14: Morphological Strategies in Language Acquisition

overgeneralization. Or rather, it will be explored how children’s knowledge about

morphological rules evolves.

4.1 Plural Overgeneralization

An example taken from the CHILDES database, illustrates a case of plural

overgeneralization, where a child of 1.5 years uses an incorrect plural form of cars.

(For means of simplification, the extra analyses that the CHILDES database

provides, have been left out.)

@Location: Living room@Activities: Free play*BEN: brrmo.%par: car noise*BEN: carses.

[INT2: > English UK > Wells-zipped > benjamin02.cha]

Children use nouns in combination with different words at their first stages of

language acquisition, e.g. more or a numeral like two as in more cookie and two kitty

(Clark 1977: 338). At this stage they rarely use the adult version of plurals with the

suffix –s added to nouns. Indeed, the plural form in English has three different

phonological realizations (allomorphs), depending on the final consonant in the

word: /s/ (if the stem ends in a voiceless consonant) as in cats, /-z/ (if the stem ends

in a voiced consonant) as in dogs and /z/ (if the stem ends in /s/ /z/ or alveolar

fricatives) as in houses. Moreover, the rule for English plurals specifies that the

voiced segment requires the voiced /-z/, while voiceless segments require /-s/ which

is learned more easily than /z/ (Clark 1993: 105).

However, children normally learn and use the correct plural form, even of

irregular nouns, e.g. men or sheep, until they realize that there are different rules for

plural formation, when they learn the use of the regular plural inflectional suffix {-s}.

8 Appendix 14

Page 15: Morphological Strategies in Language Acquisition

As a result, they often abandon the correct irregular words and use incorrect noun

plurals, such as *foots or even *feets.

This suggests, as already mentioned above, that the three main phases in the

development of a child’s acquisition of morphological rules also apply for the

acquisition of plural formation. These different phases can be further subdivided as

shown in the following table:

Table 3:

Phase Noun 1 Noun 2 Noun3 Noun 4 Noun 5

1 boy cat man house foot, feet

2 men

3 boys cats mans house foots, feets

4 boyses catses, catsez mansez, menez housez footsez, feetsez

5 boys cats mans houses feets

6 boys cats men houses feet

(LET 2005:139)

This table nicely illustrates how children process the rules for plural

formation in English. During phases 1 and 2 the child is probably still forming two-

word utterances and considers those plural forms he uses as different lexical items. In

the example of noun 5, feet is treated as a different item and not seen as the plural

form of foot. In the next phase, the strategy of overregularization is first applied: The

inflectional morpheme –s is consequently used for plural marking, regardless of

whether or not the item’s plural is formed this way or not. At this stage their rule is

to add an s-sound to the noun. Thus, children often think that forms like noun 4:

house or rose already are a plural form and do not add a further morpheme to build

the plural. In phase 4, the child overregularizes the rule which applies to stems

ending in /s/, /z/ or alveolar fricatives and creates double marked plural forms, such

8 Appendix 15

Page 16: Morphological Strategies in Language Acquisition

as *boyses or *footsez and *feetsez, which are still treated as different lexical items.

(This is also the case in the above example from the CHILDES database.)

Moreover, children also use different and sometimes incorrect allomorphs to

create the plural form of a noun, as in *housez when they should use the correct

voiceless /s/ plural ending as in houses. Only after children have passed these

different stages, they begin to use the correct plural forms, as represented in phase 6,

and they will have learned the rules for plural formations and the exceptions to them.

4.2 Past Tense Overgeneralization

In English, the past tense form of verbs is generally formed by the addition of

the suffix –ed as in watch-ed. It has three different phonetic realizations: /d/ occurs

after voiced consonants other than /d/, /t/ occurs after voiceless consonants other than

/t/ and /d/ after verbs ending in /d/ or /t/.

However, there are irregular verb forms where the addition of the past tense

morpheme may involve a change in the verb stem as well as in the suffix, as in bring

which turns into brought (Clark 1977: 342). At the so-called three-word stage,

children start to use the correct irregular past tense forms, such as went, ran, etc.

Nevertheless, the children do not relate the past tense form went to the present tense

form go. Therefore, they often add the ending –ed, *goed, or *buyed and *breaked,

to irregular verbs in the past. Furthermore, children sometimes create a double

marking of a past form such as *wented and *broked. From this it can be inferred

that children create their own ideas when talking about past events, thereby applying

their own rule to regular verb forms as well as irregular verb forms. The following

rule exemplifies that children relate the –ed ending to something which happened in

the past (Clark 1977: 343):

Verb stem + “earlier in time” → Verb stem + -ed.

Just as other overgeneralization processes, the development of

overgeneralization of past tense forms has often been divided into several phases:

8 Appendix 16

Page 17: Morphological Strategies in Language Acquisition

Stage 1: Little or no use of past tense forms

Stage 2: Sporadic use of some irregular forms (went, broke)

Stage 3: Use of regular suffix {-ed} for all past tense forms (jumped,

goed, breaked)

Stage 4: Adultlike use of regular and irregular forms (jumped, went)

(Clark 1977: 343)

The following table gives an even more detailed impression, as it illustrates

the stages Clark has listed in more detail and lists steps in between:

Table 4:

Phase Noun1 Noun2 Noun3 Noun4 Noun5

1 walk play need come go

2 came went

3 walked played needed comed goed

4 walkeded playeded neededed comeded, cameded goed, wented

5 walked played needed comed goed

6 walked played needed came went

(LET 2005: 139)

The acquisition of the correct past formation is a relatively long process,

which usually begins after the acquisition of other inflectional morphemes such as

plural, progressive and possessive morphemes. This is due to a more difficult

structure and far more exceptions. In fact, a very large fraction of the verbs a child

8 Appendix 17

Page 18: Morphological Strategies in Language Acquisition

first uses form their past irregularly, being part of the most common verbs in the

English language. (In effect, the irregular patterns could only persist over time

because of their high frequent use.) The formal complexity resulting from these

irregularities presents a challenge to the child, who will only gradually acquire the

correct past tense forms of the various irregular English verbs. The wug- experiment

has shown that no child was able to overgeneralize on supposedly irregular nonsense

words. Where all adults from the control group would form the past forms *bang and

*glang for the nonsense verbs *bing and *gling, children employed the regular past

inflection –ed. It is not surprising, that this pattern of irregular past formation is

productive for adults but not for children, since the children have not yet mastered

the correct irregular forms belonging to this group. Only a few of the older children

were able to supply the correct past tense form rang. (Berko, 1958)

In order to give a deeper insight into how children tackle the challenge of

correct regular and irregular past formation, the next chapter will investigate selected

language samples from a longitudinal gathered set of data.

8 Appendix 18

Page 19: Morphological Strategies in Language Acquisition

5 Strategies for past tense overgeneralization

on irregular verbs

To further investigate the strategy of morphological overgeneralization in

language acquisition, this chapter will discuss data from one child of whom language

samples have been recorded over a certain period of time. It will be explored in how

far this child applies the strategies which have been described in major studies, as

presented above. In order to give an extensive impression of these strategies, we will

concentrate on one aspect: The focus will be on overgeneralization in the

development of irregular verb formation.

The language samples which served as a basis for this investigation are seven

different recordings of interviews held with the child Courtney over a time span of

nine months, during which the child was between 3;4 and 4;0 years old. (INT2: >

English UK > Belfast-zipped > court01.cha – court07.cha) The research was mainly

conducted with the CLAN Program. The main search options which have been

entered to retrieve the information were:

search for word patterns (in context):

kwal +t*CHI +s"*ed" @ court*.cha

kwal +t*CHI +s"[verb stem]*" @ court*.cha

search for verb frequencies (with verb count):

freq +t*CHI +t%mor -t* +s@"|+v" @ court*.cha

search for the MLUs :

mlu +t*CHI @ court*.cha

During the research process several instances have been encountered where

the search output did not match the actual data which could be found in the file. For

example, some verbs which could be found in the search for regularly formed past

8 Appendix 19

Page 20: Morphological Strategies in Language Acquisition

endings (through the option kwal +t*CHI +s”*ed”) were not listed in the

morphological search for verbs (freq +t*CHI +t%mor -t* +s@"|+v"). Therefore

the results presented in this paper may in some cases not exactly match the actual

data.

Before we will turn to have a close look at the development of Courtney’s

past tense formation rules, a few samples from the data will illustrate at what point in

language acquisition Courtney can be put down at age 3;4, where our recordings

start. A useful indication is the MLU rate. The mean length of utterances, which is

calculated dividing the number of morphemes by the number of utterances, has

proven to be quite telling about the child’s language proficiency in stages of early

language development (Berko/ Bernstein 1998: 368). At the time under investigation

the range of Courtney’s MLU lies between 3.9 and 5.9, the average MLU is 4.8.

Correspondingly, Courtney has already reached a stage where she is able to form

longer sentences with the basic syntactic rules.

For example, all regular plurals such as horses, babies or elephants are

employed correctly in the first recorded file:

File "court01.cha": line 992. Keyword: n|elephant *CHI: oh # this is the big elephant xxx .

File "court01.cha": line 1077. Keywords: n|elephant-pl, n|door *CHI: I go outside and see some elephants banging the door .

This indicates that she has internalized the rule about regular plural formation

in English. Conversely, at the one incident where she uses a noun which forms an

irregular plural she overregularizes the plural -s:

File "court01.cha": line 369. Keyword: n|sheep-pl *CHI: and sheeps .

We will begin our investigations on past tense overgeneralization looking at

the accuracy of the use of correct irregular verb forms. For better results the primary

8 Appendix 20

Page 21: Morphological Strategies in Language Acquisition

verbs do, have and be have been left out of the entire investigation. To start with, we

will analyze an early sample of Courtney’s language production.

At her first interview, at age 3;4, Courtney uses four different full verbs in the

past tense. Of these four verbs, three form their past irregularly. These are tell, get

and hurt. The child produces each past form correctly, namely told, got, hurt and

banged for the regular verb. She does not yet overextend the rule of regular past

formation to all verbs, it seems that she has learnt the past forms for the verbs she

employs by rote and does not yet apply a general rule for past formation. Speaking in

terms of her rate of accuracy for irregular past tense formation, she would reach 100

percent.

At the second interview, one month later, Courtney employs a total of 16

different verbs in the past tense. Among these are 14 verbs which are usually formed

in an irregular pattern. Of these 14 verbs she forms 19 past forms, out of which she

uses 13 correctly and overgeneralizes 6. This calculates to an accuracy of 68.4%.

The accuracy rate already gives a rough indication that Courtney has, within

the last weeks, progressed in her development of past tense formation. In order to

make further statements about the stage of her language development, we will have a

closer look at the verb forms she employed in the interview:

*CHI: I slided [: slid] [*] # my Granpa Robs fell in the snow and hurthis bum # <he was xx> [>] .

("court02.cha": line 2159)

Table 5: Courtney, age 3;5: Irregular verbs employed with past tense inflection:

# verb form employed

Correct / Incorrect

1 breaked x2,3 broke (2) xx4 built x5 brought x

8 Appendix 21

Page 22: Morphological Strategies in Language Acquisition

67 drinked x8 et [: ate] x9-10 fell (2) xx11 got x12 hit x13 hurt x14 keeped x15 put x16-18 saw (3) xxx19 slided x

accuracy: 13 correct forms

19 forms total = 68.4%

(adapted from INT2: > English UK > Belfast-zipped > court02.cha)

The data clearly shows that Courtney begins to add the –ed past form to both,

regular and irregular verbs. Courtney’s strategy is to add the –ed ending to the base

form of the verb, as in *comed, *drinked, *slided. In other cases she uses the correct

irregular form. It seems that the child uses, alongside with the regular past,

overgeneralized and irregular past forms at a fairly equal rate. There are 6 different

irregular verbs which are formed via overgeneralization and 7 which are formed

correctly. However, we have to take into consideration that 4 verbs which are formed

correctly (/t/ suffix, i.e. no phonological change from present to past) are build, hit,

hurt and put, verbs which already end on a t-sound, indicating past tense to the child.

This leaves us with three forms which do not follow the rule: verb stem + “earlier in

time” → verb stem + t-sound. These are broke, fell and saw.

The child uses the verb see 10 times in the present and three times in the

(correct) past. It is to assume that she has understood this exception to the rule, since

she easily alternates between the past and the present form and never (in all files)

overgeneralizes to forms such as *seed or *sawed. Additionally, at age 3;4 she said:

I haven’t seen that (“court01.cha”: line 488). The correct usage of the past and past

8 Appendix 22

Page 23: Morphological Strategies in Language Acquisition

participle of see can probably be traced back to the high frequency of this word in the

child’s every day life.

The example of break indicates that Courtney begins to cast away the correct

irregular forms and replaces them with regularly built forms. Within the same

sample, break is used in both forms, twice in its irregular past form and once with an

–ed ending. We will later have a closer view at the past tense use of this verb across

all of the files under investigation (see table 6, Verb 3). For another example of a

verb form where two past forms coexist see appendix (2). The excerpt has been

chosen because here, Courtney is extensively occupied with the use of one verb.

All of above observations lead to the conclusion that, at the time of this

interview, Courtney’s rule seems to be: verb stem + “earlier in time” → verb stem +

t-sound. She is already overgeneralizing the grammatical aspect of regular past tense

formation to most verbs, although she has some knowledge about the irregular forms.

However, it is very likely that she regards the remaining irregular past forms which

she applies correctly, namely saw and fell, and the corresponding present form as two

different lexical items. Courtney never fully abandons irregular forms. Even at the

point of a very high overregularization rate (see Table 7, age 3;9) she still applies

correct rules for past formation to 56,5% of the verb forms. This corresponds to the

findings from other studies, such as Kuczaj 1977, who found that the proportion of

overregularizations always stayed below 46% (Harley, 2001: 125).

Unfortunately, the data which has been recorded for Courtney covers only a

time span of nine months and does therefore not reflect the whole process of

acquisition of past tense inflection. Nevertheless, it offers several examples of

overgeneralization as it might be expected from children at that age and gives

evidence for a continuous development. Selected verbs reoccurring in several

interviews demonstrate this development:

Table 6: Occurrence of past forms of selected irregular verbs*number behind the verb indicates the number of occurrencesfile Age Verb 1 Verb 2 Verb 3 Verb 4 Verb 5 Verb 601 3;4 come (1) hurt (1)

8 Appendix 23

Page 24: Morphological Strategies in Language Acquisition

02 3;5 pp has fallen (1)*pp has fell (1)fell (2)

broke (2)

03 3;604 3;7 broke (1) ate (1) found (1)05 3;9 falled (6)

fell (1)felled (2)

came (1) broked (1) ate (1) found (2) hurted (1)

06 3;11 falled (1)07 4;0 came (1) et (1) finded (2)

found (1)hurt (1)

(adapted from INT2: > English UK > Belfast-zipped > court01.cha-court07.cha)

Comparing the verb forms from the file which has been investigated above

(file 02) to later data collections, the most apparent difference lies between those

forms where she incorrectly attaches an –ed ending to the verb. At age 3;5 these are

*breaked,* comed, *drinked, *slided (see Table 5) and at age 3;9 these are

*broked,*falled, *felled,* likeded and *hurted. It seems as if Courtney has changed

her rule on past tense formation. She consequently no longer attaches the –ed ending

to the base form of the verb but to the irregular or even to the regular past form. This

offers different explanations. One solution would be to say that she only considers

the irregular past forms as separate lexical items. On account of the above examples,

though, we suggest that she has a fairly clear idea about these forms belonging to the

same verb as the present forms and being a variant which suggests “earlier in time”

but considers it necessary to add a further –ed marking to the verb, in regularity with

all other past forms.

In all case, it is quite evident that the child has progressed to a further stage

which implies more overregularized past forms and a larger variety of past tense

formation. The accuracy of correct past tense formation has continuously dropped

from the very first sample (age 3;4) to the fifth sample (age 3;9). At this stage she

seems to have reached the highest level of overgeneralization of the past morpheme

-ed. This goes in accordance with other studies, such as Brown, 1973. Although our

study does not cover a very long time span and the acquisition of past inflection is

not completed at the last recording, a “u-shaped” development can be observed for

8 Appendix 24

Page 25: Morphological Strategies in Language Acquisition

the accuracy of irregular verbs in this data (Table 7; see also: Harley, 2001: 125).

Table 7 is designed to exhibit the development of Courtney’s performance over the

whole time span of the gathered data. It displays the accuracy of past forms

employed by Courtney at three selected stages, which has been calculated by

dividing the number of correct past forms by the total number of past forms among

(i) irregular and (ii) both, regular and irregular, verbs.

8 Appendix 25

Page 26: Morphological Strategies in Language Acquisition

Table 7:

Accuracy of verb forms

68,40%

50%

83%90,50%

56,50%

71,40%

0,00%

20,00%

40,00%

60,00%

80,00%

100,00%

3;5 3;9 4;0

Age

Perc

enta

ge verb forms pastirregularverb forms past (all)

(adapted from INT2: > English UK > Belfast-zipped > court01.cha-court07.cha)

Table 8:

Accuracy of verbs

57,10%

88%

75%

90,50%73,30%

62,50%

0,00%

20,00%

40,00%

60,00%

80,00%

100,00%

3;5 3;9 4;0

Age

Perc

enta

ge

verbs past irregularverbs past (all)

(adapted from INT2: > English UK > Belfast-zipped > court01.cha-court07.cha)

Table 7 shows that the child’s usage of incorrect (overgeneralized) verb forms

increases before it decreases again. In contrast, table 8 suggests that the number of

different verbs which are formed correctly increases steadily. (At age 3;9, Courtney

employs 10 incorrect past forms but 6 of these are for the verb fall. Earlier, at age

3;5, she only used 6 incorrect past forms but on 6 different verbs (compare also table

8 Appendix 26

Page 27: Morphological Strategies in Language Acquisition

5).) This raises the question if these results give an indication towards a different

interpretation of Courtney’s development on past tense formation. However, these

results do not have to be contradictory to table 7. It might only indicate that Courtney

has maybe already stored some irregular forms under the same lexical item as the

verb stem. Thus, the number of individual irregular verbs that is formed correctly has

increased. Yet, the variations between and coexistences of overgeneralized and

correct past forms of other verbs has increased as well resulting in a temporarily

poorer overall performance – if all verb forms and not only the performance on

separate items are taken into account (table 7).

Concluding from the investigations on Courtney’s language samples, it can be

said that the child employs the strategies of overgeneralization in accordance with the

results from other studies on this subject. It can be inferred that she starts off with a

rather good performance on past tense verb forms, at a stage where she has not yet

developed any rules for past tense formation. In a next phase, this good performance

continuously drops down to poor performance, at a stage where she has acquired the

regular past tense ending –ed and increasingly overgeneralizes it to irregular verbs.

Finally, although she has not yet fully mastered all irregular verbs, many irregular

forms are no longer overgeneralized but have been understood as exceptions to the

rule.

Additionally, the findings of this chapter suggest that after a phase of

overregularization which is applied not only to the present stem of irregular verbs but

also to the past form of both, regular and irregular verbs, children gradually begin to

except that some verbs form their past along different patterns. They finally begin to

associate and list the irregular past forms as belonging to the same lexical item as the

base form. However, this is not a sudden development. Children usually do not have

acquired all irregular past forms until their first school years. Only then will they

eventually be able to, hypothetically (see p. 15), overextend irregular past formation

patterns productively on unknown verbs. Based on our interpretations, Courtney

might during the course of the recordings have progressed through the stages 3-5 of

table 4 (p. 15) in her language development.

8 Appendix 27

Page 28: Morphological Strategies in Language Acquisition

6 Conclusion

Language is central to our lives. The first cry of a new born baby signals its

successful arrival in the world. From the day we are born, we are able to

communicate our needs. In order to refine our utterances, however, we have to learn

to speak in the language that surrounds us. Over a period of less than a year the baby

will be able to utter words and soon utterances which eventually take the form of

complete sentences. However, it is essential for children to be exposed to a language

in order to learn it. The constant contact with language enables them to daily learn

new words and structures at a tremendous rate.

It has been of central interest for this paper that children do not only copy

words and phrases which they pick up but that they develop rules, eventually

enabling them to form grammatically correct and structurally complex sentences.

Additionally, this paper has illustrated that children follow a certain order in their

acquisition of these grammatical aspects. It takes several years for children to learn

the rules of a language and refine their overgeneralized rules.

Overgeneralization is a strategy which is employed by all children. It is

employed with different syntactic and morphological structures and seems to be

essential in the course of language development and rule formation. That children

indeed possess morphological rules which they can productively use on new words,

has been shown in the wug-experiment where children applied their rules to

nonsense material. The child first acquires the most simple grammatical rules and

gradually develops the more complex. In the course of this development, children

pass from a stage of rather good performance to poor performance at a stage of high

overgeneralization of this aspect. When they have internalized the rule plus

exceptions they pass over to good performance again. The investigation of

CHILDES database material has underlined these findings. In addition, new

questions have come up. Due to slight irregularities by the CLAN program and the

8 Appendix 28

Page 29: Morphological Strategies in Language Acquisition

limited quantity of material we cannot formulate new rules on the development of

irregular past tense formation. However, the child under investigation clearly uses

overgeneralization as a strategy of language acquisition and has applied this strategy

on numerous verbs and in different degrees at certain stages of the time span covered

by the analysis.

8 Appendix 29

Page 30: Morphological Strategies in Language Acquisition

7 References

Aitchison, Jean. 1989. The Articulate Mammal. An Introduction to Psycholinguistics. London: Routledge.

Berko, Jean. 1958. ”The Child’s Learning of English Morphology”. In: Word, 14: 150- 177.

Berko Gleason, Jean/ Bernstein Ratner, Nan (eds.). 1998. Psycholinguistics. Fort Worth: Harcourt Brace College Publ.

Bloom, Paul. 2001. How Children Learn the Meaning of Words. Cambridge, Mass.: MIT Press.

Booij, Geert. 2007. Morphology. Oxford: Oxford University Press.

Brown, Roger. 1973. A First Language. The Early Stages. London: George Allen & Unwin Ltd.

Clark, Eve. 1993. The Lexicon in Acquisition. Cambridge: Cambridge University Press.

Clark, Herbert / Clark, Eve V. 1977. Psychology and Language. An Introduction to Psycholinguistics. New York et al: Harcourt Brace Jovanovich.

Fletcher, Paul. 1985. A Child’s Learning of English. Oxford: Blackwell.

Guasti, Maria Teresia. 2002. Language Acquisition and the Growth of Grammar. London et.al.: The MIT Press.

Harley, Trevor. 2001. The Psychology of Language. From Data to Theory. Hove: Psychology Press.

King, Kendall A. 2006. “Child language acquisition.” In Fasold, R. W., Connor-Linton, J. (eds.). An Introduction to Language and Linguistics. Cambridge: Cambridge University Press. pp. 205-234.

Linguistic Engineering Team (LET). 2005. Psycholinguistics. Marburg: The Virtual Linguistics Campus.

Mac Whinney, Brian. 1991. The CHILDES project. Hillsdale et al: Lawrence Erlbaum Ass.

Steinberg, Danny D. / Nagata, H. / Aline, D. P. 2001. Psycholinguistics: Language, Mind and World. Second Edition. London: Longman.

Internet Sources

8 Appendix 30

Page 31: Morphological Strategies in Language Acquisition

[INT1]: http://www.oup.com/uk/catalogue/?ci=9780195177879, accessed 4 March 2008.

[INT 2]: CHILDES DATABASE. http://childes.psy.cmu.edu/data/, accessed 4 March 2008.

8 Appendix 31

Page 32: Morphological Strategies in Language Acquisition

8 Appendix

(1) Wug-Experiment, Berko 1958: Examples (INT2):

Plural Singular and Plural Possessive

Plural Past Tense

8 Appendix 32

Page 33: Morphological Strategies in Language Acquisition

(2)

Situation: The Child is playing a fishing game with the investigator:

*INV: you see if you can catch a fish .*INV: oh it fell off !*INV: O [=! laughs] .*MOT: no you've to do it with the rod Courtney .[…]*CHI: <I caught a yellow one> [>] .[…]*CHI: one two three four five once I caught a fish alive # six seven eight nine ten # when I let it go again # why did you let it go # cause it bit my finger xxx # which finger did it bite # this little finger on the right [sings] .[…]*CHI: I got one !*CHI: have to catch a red one .*CHI: xxx red one .*INV: ah well done # that's good # ah very good !*CHI: look what I catched Mummy !*MOT: uhhuh very good .[…]*CHI: Dad I catched a yellow one !*FAT: very good .[…]*CHI: I catched two ![…]*CHI: I think he doesn't want caught .*INV: I think he doesn't # no .*CHI: do you want caught or what ?*INV: O [=! laughs] .*CHI: doesn't .[…]

*CHI: it doesn't want caught .[…]*CHI: xxx doesn't want caught .*INV: he doesn't want caught does he ?*CHI: he won't caught # he doesn't want caught ![…]*CHI: that fish doesn't want # caught .*INV: oh # it doesn't .*INV: it's trying to swim away !*CHI: it's trying to fall in .*CHI: Mum that fish doesn't want caught !*MOT: does it not ?[…]*CHI: fish do you want caught ?[…]*CHI: xxx I caught one .[…]*CHI: I &w that one doesn't want caught .[…]*CHI: <I can't catch it # can't catch it>[<] !*CHI: <yeah # I caught it>[<] ![…]*CHI: xxx caught .[…]*CHI: caught it # xxx catch it .[…]*CHI: xxx caught .[…]*CHI: I caught them ![…]

*CHI: <I catched it>[<] !

8 Appendix 33

Page 34: Morphological Strategies in Language Acquisition

(INT2: > English UK > Belfast-zipped > court02.cha": lines 1883-2795.)

8 Appendix 34