Go to Chapter 2

POPULATION

The target population is primary age children acquiring English as their second language in an English-speaking environment while simultaneously being introduced to English literacy. The available population consists of second and third grade LEP students in the ESL triad program (an integrated program containing both mainstream and ESL students) at an elementary school in Western Washington in which the majority of the students speaks English as a second language. Over 85% of the school's students qualify for free lunch and many of the student body live in Federal Public housing. The target population includes students who have been exited from the ESL program but who are still limited English proficient.

SUBJECTS

The subjects included in the study were ten students, ages 8 to 10, all of whom were in the researcher's second/third grade classroom. All of the subjects were Southeast Asians acquiring English as their second language and were in the ESL program for at least one year before entering the classroom. Half were in second grade at the beginning of the study and continued in the classroom the following year. The rest were in third grade and remained in the study the following year in other classrooms.

More than ten students returned permission slips. Only twelve, however, remained for the duration of the data-gathering. Of these, two were dropped from the study. They both had learning disabilities and language difficulties beyond second language issues and were diagnosed for special education. Their low skills made it difficult to acquire reading and writing samples and their disabilities were a confounding factor in their language development.

At the beginning of the data-gathering, the students represented a range of English and literacy skills, varying in English proficiency from Level 2 to Level 4 on the Student Oral Language Observation Matrix (SOLOM) used by the school district for assessing English proficiency of LEP students (see Appendix A). Level 1 of the SOLOM is completely non-English-speaking and Level 5 is native-like proficiency. Reading levels ranged from pre-primer to third grade. The following are profiles of the subjects upon entering the study.

Subject 1: Cambodian, M, 2nd grade, academically strong, Level 4 SOLOM, 2.1 Reader, birth place: refugee camp in Thailand

Subject 2: Vietnamese, F, 3rd grade, academically strong, Level 4 SOLOM, 3.1 Reader, birth place: U. S.

Subject 3: Vietnamese, M, 3rd grade, academically weak, Level 2.5 SOLOM, 2.1 Reader, birth place: Vietnam

Subject 4: Vietnamese, F, 3rd grade, academically average, Level 2.5 SOLOM, 2.1 Reader, birth place: Vietnam

Subject 5: Cambodian, F, 2nd grade, academically average, Level 3 SOLOM, 1.2 Reader, birth place: U. S.

Subject 6: Cambodian, M, 2nd grade, academically average, Level 2.5 SOLOM, Primer Reader, U. S.

Subject 7: Cambodian, M, 2nd grade, academically weak, Level 2.5 SOLOM, Pre-primer Reader, birth place: refugee camp in the Philippines

Subject 8: Cambodian, F, 2nd grade, academically average, Level 2 SOLOM, 2.1 Reader, birth place: U. S.

Subject 9: Cambodian, M, 3rd grade, academically average, Level 3 SOLOM, 1.2 Reader, birth place: U. S.

Subject 10: Cambodian, F, 3rd grade, academically strong, Level 4 SOLOM, 3.1 Reader, birth place: U. S.

Of the ten subjects, only Subject 4 had had any schooling in the native language, having attended first grade in Vietnam. The other students began school in the U. S.

RESEARCH DESIGN

This study is an error analysis of language development in LEP children. Error analysis is neither a quantitative nor qualitative method although it can use statistical techniques in analyzing data. Primarily descriptive, it is used in many fields to determine the cause of errors by analyzing the pattern of mistakes. For example, a math teacher might find that her students were consistently carrying the ones rather than the tens in addition and conclude that a refresher on place value was required. In language acquisition, however, errors are not regarded as mistakes per se, but rather as marking developmental stages in the process of acquiring a language. The current theory of language development posits that children acquire language by making a series of hypotheses about the structure of the language. As they interact with the language of their environment, the hypotheses are retained or discarded and new hypotheses formed until the structure of the child's language matches that of the adult or native speaker. Error analysis provides a means of observing and describing this developmental process.

The current study attempts to compare language development in three of the four language modalities - speaking, reading and writing. Of the two receptive modalities, listening and reading, the use of miscue analysis - a miscue being any deviation from the text - allows the researcher to examine a child's developing cognizance of specific morphemes in reading. The same is not true of listening so this modality has been omitted. The child's use of the morphemes can be observed in both of the productive modalities, speaking and writing.

In analyzing the subjects' errors, the researcher is attempting to demonstrate that the structure of the language is developing across the modalities simultaneously, that engaging in reading and writing during the language acquisition process provides additional opportunities for a child to develop and test hypotheses about the structure of the language. During approximately 18 months, 4 samples were taken for each of the target modalities, 4 to 6 months apart. These were edited for all errors in order to see which structures could best be analyzed. A number of morphological structures showed development across all three modalities. These included:

1. present tense
2. regular past tense
3. irregular past tense
4. pronouns
5. regular plural
6. irregular plural
7. present to be form
8. past to be form
9. contractions
10. articles,
11. possessive s

Of these, the majority did not have sufficient occurrences in which the morpheme was required to confidently analyze. Not every sample contained opportunities for every morpheme, and some morphemes such as possessive s rarely had more than one or two occurrences if any. This was especially true of the writing samples in which the nature of the assignment often dictated the types of structure used, e.g., an informational piece required few past tense morphemes while stories rarely needed present tense. Five morphemes were chosen as occurring in sufficient numbers to provide fairly robust data. They averaged several occurrences per sample and had few missing cases. These were:

1. Pronouns
2. Regular past tense
3. Irregular past tense
4. Regular plural
5. Past to be verb

The data were analyzed statistically to determine acquisition order in the individual modalities and whether specific morphemes were acquired first in any one modality. In addition, each student's errors were scanned to observe the specific nature of the errors.

DATA-GATHERING

Reading samples:

IRI passages were taken from the Qualitative Reading Inventory (QRI). The specific passage used depended on the subject's reading level. Passages were read aloud. No assistance was provided except in maintaining place in the passage and the provision of unfamiliar names which were not included in the analysis. The passages were marked for miscues, any deviation from the text, according to the directions provided by the QRI (1990, p. 46-47). The readings were taped in order to check the original marking. See Appendix B for an example.

Speech samples:

The speech samples were taken at the same time as the reading samples. They consist of two parts. The first is the retelling of the QRI passage which the subject read. The second is a story generated by the student from looking at a picture or a picture book. Students were asked to pick the picture about which they could best tell a story. Picture books included The Three Billy Goats Gruff, Peter Rabbit and Little Red Riding Hood. The pictures were an assortment taken from old calendars. Among them were: 1. Christopher Robin and an assortment of animals pulling Winnie the Pooh from Rabbit's hole; 2. several photographs of cats in various poses - in dresses hanging up clothes in the park, with surfboards at the beach, by motorcycles at a restaurant; 3. reproductions of Norman Rockwell pictures - three children painting a fire hydrant while a policeman observes, two children asleep in a big chair with Santa Claus tiptoeing by, a boy leaping to catch a ball which goes over his head to hit an irate woman's window. All speech samples were taped and checked to verify the transcripts taken at the time. After the speech samples were typed, they were edited and marked with correct grammatical usage. See Appendix B for an example.

Writing samples:

Since writing is a time-consuming process, especially at the primary level, writing samples were taken from classroom projects written over a period of time and thus do not correspond to the exact day of the reading and speech samples. The writing samples were produced within the month in which the other samples were taken. The pieces of writing included were produced entirely within the classroom, taking more than one session, and had undergone the full process of writing, revision and final draft. No assistance in proper grammar or form was given. Assistance was limited to requested aid in spelling and to questions designed to prompt additional ideas, e.g., What did the dinosaurs do when they reached the sun? Since the classroom used peer editing, only rough drafts were included to ensure that the writing sample reflected only the subject's language. This is limiting since the samples thus do not reflect one of the primary differences between speech and writing, the opportunity which writing provides the child to revise and experiment with word order and form. Once the samples were typed, they were first edited for spelling and punctuation and then marked with correct grammatical usage. See Appendix B for an example.

DATA-CODING

Once all samples had been edited and the morphemes to be analyzed had been selected, all occasions in which a given morpheme was required were marked. The subject's usage and the correct usage were both recorded on a data sheet. See Appendix B for an example. Errors and the nature of the error, i.e., decoding or grammatical, were marked. Only morphological errors, errors resulting from the stage of the subject's language development, were included in the analysis. If the nature of the error was unclear or the correct usage was ambiguous, that specific occurrence was omitted from the analysis. For example, in the following writing sample - Coastal Indians live in longhouse. Several family live in the longhouse. - it is clear that families is the correct usage rather than family. However, in the first occurrence of longhouse, it could be edited to say "Coastal Indians live in longhouses." or "Coastal Indians live in a longhouse." Since it is not possible to ascertain whether the subject was thinking of one or more than one, the occurrence is not included.

Reading samples:

On the data sheet, all occurrences in the text of a given morpheme, e.g., irregular past tense, were recorded along with what the subject said. Errors were marked either grammatical or decoding. For example, when the sentence "They paid $5.00 for a black puppy and brought it home with them." is read as "They paid five dollar for a black puppy and bought it home with them.", dollar is a grammatical error and shows the absence of the s-morpheme while bought is a decoding error. A grammatical error does not significantly affect meaning but reflects the subject's current grammar. A decoding error reflects the subject's word attack skills. Once utterances were recorded and marked, the percentage of correct utterances out of the total possible was computed. Decoding errors were not included in the computation.

Speech samples:

On the data sheet, all occurrences in which a given morpheme, e.g., irregular past, were required were noted with the correct form and the utterance given by the subject. Errors were marked either grammatical or meaning. The latter refers to errors resulting from a vocabulary lack rather than knowledge of the language structure. Since the analysis is dealing with specific morphemes, meaning errors were not included in computing the percent of correct usage. For example, in retelling a story about the zoo, a subject referred to chumps. The correct word should have been chimps. However, in recording occurrences in which the regular plural was required, while the usage was in error, it could not be considered to reflect the subject's acquisition of the plural form. Such errors were rare. As with the reading samples, the percentage of a subject's correct usage were noted for each morpheme.

Writing samples:

Writing samples were coded exactly like the speech samples.

Reliability:

Samples from each modality were given to five native speakers, four teachers and one teacher intern, to mark as the researcher had done. The directions for the marking are provided in Appendix C. The five were asked to edit samples from the speaking and writing modalities, giving the native speaker form as the researcher had done. On the reading sample, in which the correct form is provided by the passage, they were asked to mark errors as decoding errors or grammatical errors. For all the errors included in the study, the researcher's markings were tabulated against those of the five native speakers as shown in the tables on the following pages. Only one of the reading errors showed much disagreement, the substitution of the for his. This was one which had been omitted as ambiguous. Of the others used, the markings were unanimous with the exception of the substitution of say for said which one person marked as a decoding error. On the speech and writing, the five native speakers agreed unanimously with the researcher as to the correct form required.

RELIABILITY MEASURE

TABLE I: READING

Given/Correct usage	Researcher	NS1	NS2	NS3	NS4	NS5
say/said	G	G	D	G	G	G
you/you'll	G	G	G	G	G	G
nots/notes	D	D	D	D	D	D
price/pieces	D	D	D	D	D	D
give/gave	G	G	G	G	G	G
noted/noticed	D	D	D	D	D	D
pick/picked	G	G	G	G	G	G
read/read	G	G	G	G	G	G
girl/girls	G	G	G	G	G	G
walk/walked	G	G	G	G	G	G
look/looked	G	G	G	G	G	G
that/that's	G	G	G	G	G	G
tap/taped	D	D	D	D	D	D
girl/girls	G	G	G	G	G	G
find/found	G	G	G	G	G	G
fix/fixed	G	G	G	G	G	G
read/read	G	G	G	G	G	G
let/let's	G	G	G	G	G	G
look/looked	G	G	G	G	G	G
the/his	D	D	G	G	D	D
want/wanted	G	G	G	G	G	G
want/went	D	D	D	D	D	D
look/looked	G	G	G	G	G	G
want/went	D	D	D	D	D	D
look/looked	G	G	G	G	G	G
play/played	G	G	G	G	G	G
went/wanted	D	D	D	D	D	D
Chris/Chris's	G	G	G	G	G	G
want/wanted	G	G	G	G	G	G

NS = Native Speaker
G = Grammatical Error
D = Decoding Error

TABLE II: SPEECH

Usage	Researcher	NS1	NS2	NS3	NS4	NS5
Chris	Chris'	Chris'	Chris'	Chris'	Chris'	Chris'
ask:	asked	asked	asked	asked	asked	asked
Chris	Chris'	Chris'	Chris'	Chris'	Chris'	Chris'
askses	asked	asked	asked	asked	asked	asked
shoe	shoes	shoes	shoes	shoes	shoes	shoes
hide	hid	hid	hid	hid	hid	hid
jump	jumped	jumped	jumped	jumped	jumped	jumped
scare	scared	scared	scared	scared	scared	scared
can:	could	could	could	could	could	could
clear	cleared	cleared	cleared	cleared	cleared	cleared
make	made	made	made	made	made	made
rabbit	rabbit's	rabbit's	rabbit's	rabbit's	rabbit's	rabbit's
like	liked	liked	liked	liked	liked	liked
like	liked	liked	liked	liked	liked	liked
he's	he was	he was	he was	he was	he was	he was
will try	tried	tried	tried	tried	tried	tried

NS = Native Speaker

TABLE III: WRITING

Usage	Researcher	NS1	NS2	NS3	NS4	NS5
walk	walked	walked	walked	walked	walked	walked
ask	asked	asked	asked	asked	asked	asked
snowman	snowman's	snowman's	snowman's	snowman's	snowman's	snowman's
live	lived	lived	lived	lived	lived	lived
walk	walked	walked	walked	walked	walked	walked
snowman	snowman's	snowman's	snowman's	snowman's	snowman's	snowman's
thing	things	things	things	things	things	things
dynamite	sticks of dynamite	sticks of dynamite	sticks of dynamite	sticks of dynamite	sticks of dynamite	sticks of dynamite
snowman	snowman's	snowman's	snowman's	snowman's	snowman's	snowman's
explode	exploded	exploded	exploded	exploded	exploded	exploded
piece	pieces	pieces	pieces	pieces	pieces	pieces
dollar	dollars	dollars	dollars	dollars	dollars	dollars
give	gave	gave	gave	gave	gave	gave
dollar	dollars	dollars	dollars	dollars	dollars	dollars
clean	cleaned	cleaned	cleaned	cleaned	cleaned	cleaned
time	times	times	times	times	times	times
hour	hours	hours	hours	hours	hours	hours

NS = Native Speaker

DATA ANALYSIS

The data for each morpheme across the three target modalities were graphed. Then a simple categorical ANOVA was run for each morpheme, comparing the modalities against each other. ANOVA is a test of significance which determines whether a difference in the means of two or more groups demonstrates a real difference between the groups or simply a chance difference in score. Thus, for the irregular past, Reading was compared with Writing and Speech to see if the differences in mean value were significant. Writing and Speech were also compared.

Within modalities, specific morphemes were compared with each other using a simple categorical ANOVA to see if the differences in the means in the acquisition order were significant.

Since several samples were taken from each student, the original intent was to use a repeated measures design to eliminate variation resulting from differences in student levels from the total variation. However, for each morpheme, an occasional sample lacked occurrences, leading to missing values. In the ANOVA, that sample was eliminated from the analysis. However, in the repeated measures design, all samples from a given subject were eliminated if one contained a missing value, resulting in considerable data loss. Consequently, the decision was made to use the categorical ANOVA even though it might make significant differences more difficult to achieve. Since with such few subjects, numbers randomly selected to fill in the missing values could skew the results, that option was not considered.

In addition to the statistical analysis, the developmental process was examined at the individual level by comparing ordinal placement of individual subject scores across modalities and morphemes and by analyzing specific errors in individual samples.

Go to Chapter 4