Title: Keyword mnemonic and contextual analysis strategy instruction with at-risk adolescents
Full Citation
Permanent Link: http://ufdc.ufl.edu/UF00102721/00001
 Material Information
Title: Keyword mnemonic and contextual analysis strategy instruction with at-risk adolescents
Physical Description: Book
Language: English
Creator: Fox, Shirley N., 1953-
Copyright Date: 1993
 Record Information
Bibliographic ID: UF00102721
Volume ID: VID00001
Source Institution: University of Florida
Holding Location: University of Florida
Rights Management: All rights reserved by the source institution and holding location.
Resource Identifier: ltuf - AJX6104
oclc - 29891607

Full Text







UN',i7,TY r- r

This work is dedicated to my parents,
Mildred R. and Lyle B. Fox, for their perpetual support
and inspiration for learning.


I would like to take this opportunity to thank the many

individuals who have assisted in the undertaking of this

research. First, many thanks and much appreciation are

extended to Margaret Gentile, who as director of Alternative

Programs for Orange County Schools invested herself greatly

and consistently over the course of the project. Special

thanks are also given to Kathy Roach and Lucille Hall, who

participated as observers and data gatherers. Dr. David

Miller, as research and statistics advisor, provided valuable

consultation and assistance whenever needed. Drs. William

Reid, Cecil Mercer, and Cynthia Griffin offered guidance and

served as resources and support persons. Special thanks are

owed to the other committee members, Dr. Stephen Smith and

Dr. Cary Reichard, whose regard for scholarship has been both

a support and a guide. The special education faculty and

staff, as well as the staff of the many other departments of

the University of Florida, have also been informative and

supportive. I have appreciated their willingness to assist

me during this project and throughout my years at the

University of Florida.

Several other individuals directly assisted in the

project, and each of them deserves special thanks. Marcia

Kirkland and Lisa Ruggeri gave generously of their time to


serve as graders and helpers throughout this research. The

teachers, Becki Kneeland, Nelson Pinder, Bill Van Sickle,

Gayle Mooring, Vicki Lee, and Harvey Smerilson, deserve

special thanks as well. For his artwork in making the cards

for the study, Pawel Nowicki receives special thanks. Cindy

Parks and the Orange County testing department were all very

accommodating and giving of their time.

A special expression of gratitude is extended to Cindy

McCrery for her expertise and speed in the typing of this

manuscript. She has been a friend throughout my graduate

years and has always been more than helpful. She is my

unsung heroine.

I would also like to thank the friends and colleagues,

too numerous to name, that have offered personal and

professional support and encouragement. I appreciate their

patience and tolerance throughout this experience. I

especially would like to extend appreciation to my boss, Gary

Schadow, who has been more than kind and understanding of

these efforts.

Finally, a very special thanks and recognition to my

best friends and loved ones, who have been patient and

tolerant and have shown enduring love and support throughout

the completion of the doctoral requirements. Warmest

acknowledgements are made to my parents, Mildred and Lyle

Fox, my brother and his family, Skip, Nina, Megan, and Erik,

my best friend and colleague, Kathy Shewey, and Dr. Bruce

Suther, who has helped me in every way possible. I would

like to send a heartfelt thanks to each of these persons.

appreciate them all for letting me know that I matter.


ACKNOWLEDGEMENTS .......................................... iii

LIST OF TABLES ............................................ ix

LIST OF FIGURES ........................................... x

ABSTRACT ...................................... .......... xi


1 INTRODUCTION ......................................... 1

Statement of the Problem ................. .......... 6
Purpose and Objective ............................... 7
Rationale .......................................... 8
Definition of Terms ................................... 13
Delimitations ........................................ 15
Limitations ........................................ 15
Summary .............................................. 16

2 REVIEW OF RELATED LITERATURE ....................... 18

Inclusion Criteria .................................... 19
Background Information on Memory Strategies ......... 19
Prior to Twentieth Century ..................... 20
Twentieth Century .................................. 21
Two-System Theory .................................. 21
Multiprocess Memory Theory ..................... 22
Constraints on Short-Term Memory ................ 23
Automatic Processes ................................ 24
Background Information on Mnemonic Strategies ...... 27
Peg-Word Theory ................................... 29
Pair-Associate Tasks ............................... 30
Additional Memory Factors ...................... 31
Keyword Mnemonic Strategies ....................... 33
Keyword Mnemonics with Students with
Mild Disabilities ................................ 38
Research Design and Description of Conditions ... 40
Subject Characteristics ............................ 49
Measurement Methods ................................ 50
Experimental Procedures ............................ 52
Results ........................................... 54
Background Information on Contextual Analysis ...... 56
Factors Affecting Contextual Analysis Research .. 60

Keyword Mnemonics Compared to Contextual Analysis .. 63
Research Design and Description of Conditions ... 64
Subject Characteristics ............................ 74
Measurement Methods ................................ 75
Experimental Procedures ............................ 77
Results .......................................... 79
Summary ................. ............................. 81

3 METHODS .............................................. 83

Sub jects ............................................. 83
Hypotheses ......................................... 86
Instrumentation ................................... .. 87
Written Geography Term Definition Screening
Device ........................................... 88
Teacher Survey to Evaluate Content Validity ..... 88
Daily Measure of Geography Terms' Definition
Device ........................................ 89
Geography Terms' Definition Device .............. 89
Materials ............................................ 90
Procedures ......................................... 91
Survey and Screening Procedure (Phase One) ...... 92
Teacher Training Procedures (Phase Two) ......... 93
Instructional Implementation (Phase Three) ...... 96
Measurement (Phase Four) ........................ 103
Experimental Design and Analysis ................... 105

4 RESULTS ............................................ 107

Teacher Training Interobserver ..................... 107
Interscorer Reliability ............................ 108
Geography Term Definition Screening Device ......... 110
Statistical Analysis of the Data ................... 111
Hypothesis 1 ......................................... 111
Hypothesis 2 ......................................... 115
Hypothesis 3 ......................................... 117
Descriptive Analysis of the Data ................... 119
Daily Acquisition Description ................... 119
Weekly Short-Term Recall Description ........... 120
Two-Week Long-Term Recall Description .......... 121
Summary ................. ............................. 122

5 DISCUSSION ............................................. 123

Review of the Study ................................ 123
Purpose and Objective .......................... 123
Literature Summary Related to Theoretical
Considerations ................................... 123
Hypothesis 1 ...................................... 129
Hypothesis 2 ...................................... 130
Hypothesis 3 ...................................... 130
Summary of Method .............................. 131
Discussion and Implications ........................ 137
Problems and Limitations ........................... 141

Suggestions for Further Research ...................


REFERENCES ..............................................

BIOGRAPHICAL SKETCH .....................................



SCREENING DEVICE ..............................


DEFINITION DEVICE .............................


CONDITION ................... ..................



TEACHER SCRIPT ................................

CONDITION TEACHER SCRIPT ......................



















Table Page

2-1 Review of Keyword Mnemonic Vocabulary Studies
with Mildly Handicapped Subjects .................. 41

2-2 Review of Keyword Mnemonic Vocabulary Studies
with Contextual Analysis Methods .................. 65

3-1 Description of Subjects ........................... 85

4-1 Interobserver Agreement on Teacher Behavior ....... 109

4-2 Results of the Geography Term Definition
Screening Device................................... 110

4-3 Summary Table for the Geography Term Definition
Screening Device................................... 111

4-4 Written Definitions of Daily Quizzes from
4 Weeks of Intervention for Between Subjects
Effects (Acquisition) ............................ 113

4-5 Written Definitions of Daily Quizzes from
4 Weeks of Intervention for Within Subjects
Effects (Acquisition) ............................ 113

4-6 Summary Table for General Linear Models
Procedure for Daily Quizzes........................ 114

4-7 Results of Multiple Comparisons .................... 115

4-8 Written Definitions of Weekly Tests from 4 Weeks
of Intervention for Between Subjects Effects
(Short-Term Retention) ........................... 116

4-9 Written Definitions of Weekly Tests from 4 Weeks
of Intrvention for Within Subjects Effects
(Short-Term Retention) ........................... 117

4-10 Results of Tukey's HSD Comparison ................. 118

4-11 Written Definitions of 48 Targeted Terms from
Week 7 (Long-Term Retention) ...................... 119

5-1 Number of Subjects Available for Dependent
Measures.. ................................. ....... 132


Figure cPae..

1 Recording Component ................................. 34

2 Relating Component ................................ 35

3 Retrieving Component .............................. 36

Abstract of Dissertation Presented to the
Graduate School of the University of Florida
in Partial Fulfillment of the
Requirements for the Degree of
Doctor of Philosophy



Shirley N. Fox

May 1993

Chairman: Dr. Cecil D. Mercer
Cochair: Dr. Cynthia C. Griffin
Major Department: Special Education

The purpose of this study was to compare at-risk

adolescents' recall of geography vocabulary definitions among

three treatments during an acquisition, short-term retention,

and long-term retention phase. The teachers were randomly

assigned to one of three instructional conditions using a

teacher-provided keyword mnemonic strategy, a teacher-

provided contextual analysis strategy, or a teacher-provided

rote free-study strategy. The subjects' recall was measured

with a written assessment of the targeted terms' definitions.

An analysis of variance with repeated measures was computed

to analyze the results.

Significant differences for treatment methods were found

for the written assessments. Subjects in the rote free-study

control condition recalled significantly more than did

subjects in the other two conditions on acquisition and

long-term recall of geography terms' definitions. Subjects

in the keyword mnemonic condition recalled significantly more

definitions than did subjects in the contextual analysis

condition for the first two weeks of vocabulary acquisition.

No statistically significant difference was found between the

mnemonic keyword and rote free-study control conditions on

the short-term recall measures. These results have

educational implications for at-risk adolescents who are

required to learn definitions in their assigned coursework.



With the passage of P.L. 94-142 in 1975, extended

services and programs have been provided for a select group

of students who qualified for Exceptional Student Education

(ESE) programs. These traditional special education services

offer only minimal support towards helping students succeed

in the education system. Not all students in need of extra

services and programs qualify for ESE programs; such students

are considered "at-risk" for school failure and are

identified as those who will encounter difficulty graduating

from high school. These at-risk students are not handicapped

according to the provisions stated in P.L. 94-142, but have

been identified by many other categories in the past; i.e.,

culturally deprived, low-income, dropout, low-achieving, or

disadvantaged (Presseisen, 1988).

Public concern regarding the numbers of students who

drop out before graduating from high school has increased

during the 1980s and 1990s. Calculating dropout rates is

difficult because of definitional and data inconsistencies.

However, high school graduation rates represent a stable

measure of this ever-growing problem. In 1986, slightly less

than three quarters of the nation's 18- and 19-year-olds had

completed high school (Kolstad & Owings, 1986). Cooper

(1990) reported that 28.9% of our nation's high school

students fail to graduate. High school completion rates,

however, vary across districts. The lowest rates are found

in urban areas and in African American and Hispanic

populations (Pallas, 1987). The costs of our national

dropout problem are reflected in higher welfare expenditures,

lost tax revenues, and increased crime and crime prevention

costs (Catterall, 1985). The intangible costs to

individuals, families, and society are also noteworthy.

Our nation's schools are generating 3,600 new high

school dropouts each day (Cooper, 1990). Despite increasing

numbers of students in Florida schools, the number of high

school graduates has decreased from 90,000 in 1981 to 82,000

in 1987. Sixty-two percent of the nation's high school

seniors are graduated (Hodgkinson, 1988a). The dropout

problem has been particularly acute in large urban school

districts such as the Orange County, Florida, Public School

System. In that system, during the 1991-92 school year, 326

middle and 1,415 high school students dropped out of school.

The Orange County public school system is the 18th largest

district out of more than 16,000 school districts in the

nation and is the sixth largest in Florida (Orange County

Public Schools, 1992).

Kaplan and Luck (1977) noted several characteristics

common to at-risk students: academic underachievement,

antisocial behaviors, academic failure, and truancy. Pallas

(1987) also reviewed the data on reasons students drop out of

school and concluded that poor academic performance was the

best predictor of who drops out of school. Presseisen (1988)

further identified reading as the most critical weakness of

at-risk students. She stated that a student who is not

proficient at reading text is headed for school failure.

Today's educators face the challenge of teaching students who

are at-risk for school related problems and must alter the

traditional education program for these students to succeed

in the mainstream. Even though previous research has enabled

the determination to be made that severe academic

deficiencies exist among at-risk students, little is known

regarding optimal instructional strategies for content area

learning with this population. Stringer (1973) suggested

that at-risk students who continue their education in the

mainstream of regular classes without intervention will most

likely experience long-range failure and will eventually

become a school dropout.

Many students experiencing academic problems also

exhibit problems with understanding and using language.

Smith, Goodman, and Meredith (1970) explained that "language

may be viewed as either an external verbalization about

things or as an integral part of the personal process of

experiencing and knowing . language is the very heart of

the teaching-learning process" (p. 68). Smith et al. (1970)

viewed teaching and learning mainly as language games and

believed that these language games are a basis to being

successful in the education system. In addition, Brown

(1985) compared poor readers with good readers and determined

that poor readers lacked strategies in comprehending the

meaning of what they read. Researchers studying at-risk

students support the belief that if these deficiencies

continue to exist as a student goes through the educational

system, the student burdened with below-average reading

scores is twice as likely to drop out of school as are

students who have achieved average or above average reading

levels (Presseisen, 1988). Similarly, Torgesen and his

associates (Torgesen, 1977; Torgesen & Goldman, 1977;

Torgesen & Licht, 1983) have also compared good and poor

readers and suggested that some students do not achieve to

their potential because either they are unable to develop and

use efficient strategies, or they lack the motivation to do


Much of Torgesen's research comparing good and poor

readers has focused on memory tasks. He suggested that the

differences between good and poor readers may be the result

of differences in general cognitive strategies underlying not

only performance on experimental memory tasks, but also on

the attainment of reading vocabulary and comprehension, as

well as other academic skills (Torgesen & Kail, 1980).

Torgesen and Goldman (1977) studied second graders and found

that, even with familiar stimuli, poor readers were deficient

in the use of simple active strategies like rehearsal.

Differences between good and poor readers in the use of more

complex strategies have also been reported (Graves, 1987;

Torgesen, Murphy, & Ivey, 1979). Researchers support the

need for students to develop strategies for learning

vocabulary words, increasing comprehension, and developing an

attitude towards vocabulary and reading that will encourage

students to apply these strategies (Graves, 1987; Marfo &

Ryan, 1990).

Most schools do not provide specialized instruction for

those students who experience difficulty in learning

vocabulary terms. It is the responsibility of regular

educators to teach at-risk students who have deficits in

vocabulary acquisition. Many different methods have been

recommended for teaching vocabulary (Gipe, 1979; Johnson &

Pearson, 1978; Levin, 1981a). Students have been taught

vocabulary by such traditional approaches as drilling with

flashcards, writing the definition from a dictionary, using

context clues, and free-study. Although, two particular

methods are prevalent in the current literature: the keyword

method, which is a mnemonic method; and learning from

context, which expounds upon generalization skills (Condus,

Marshall, & Miller, 1986; Kroll, 1990; Levin & Levin, 1990;

Scruggs & Mastropieri, 1990); a review of the research

literature reveals that the effectiveness of these and other

methods for teaching students who have been identified as

at-risk for dropping out of school before graduation is


Statement of the Problem

The focus of the present study was to determine the

achievement of at-risk students taught three methods of

vocabulary instruction for learning geography terms (i.e.,

mnemonic keyword strategy, contextual analysis strategy, and

rote free-study strategy). Many middle school teachers in

the mainstream teach vocabulary by the traditional methods of

drill and practice (Chall, 1987). Since the rote free-study

technique represents an approach that is typically used to

teach middle school students vocabulary terms, this group

served as a control group. Specific research questions

regarding vocabulary instruction through keyword mnemonics,

contextual analysis, and rote free-study for at-risk students

answered in this investigation included the following:

1. Are there any significant differences among the

three instructional groups on vocabulary acquisition


2. Are there any significant differences among the

three instructional groups on short-term vocabulary retention

measures administered at the end of the instructional week?

3. Are there any significant differences among the

three instructional groups on long-term vocabulary retention

measures two weeks after vocabulary instruction?

The question of which techniques are effective for

teaching at-risk students skills for vocabulary acquisition

is important to investigate for several reasons. First, the

study has provided data on teaching at-risk students

vocabulary acquisition skills using a mnemonic keyword

method, contextual analysis technique, and rote free-study

approach. Second, this study has provided regular education

teachers with information regarding at-risk students'

vocabulary retention after being taught specific vocabulary

skills. Third, the study has contributed data related to the

relationship between teaching vocabulary acquisition

techniques and skill generalization among students

experiencing academic failure. Finally, the study has

provided information that may be useful in curriculum

planning for promoting additional ways for at-risk students

to achieve academic success.

Purpose and Objective

The purpose of the study was to examine the effects of

keyword mnemonics, contextual analysis, and rote free-study

conditions for learning geography terms taught to middle

school adolescents who are at risk of not completing high

school. The objective of the study was to determine the

relative effectiveness of these three techniques in improving

the vocabulary acquisition and retention of students

identified as at risk for dropping out of school before



Researchers and educators in the field of dropout

prevention have been encouraging administrators and teachers

to become more responsive to the needs of the at-risk

adolescent (Gillespie, 1989; Toepfer, 1990). Legislators are

also supporting the move for basic skills improvement and

dropout prevention programs with the passage of P.L. 101-600

(1990), also known as the "School Dropout Prevention and

Basic Skills Improvement Act of 1990." The intent of P.L.

101-600 is for the federal government to increase successful

federally funded demonstration programs that increase the

high school completion rate, encourage states to adopt

specific plans to increase graduation rates, and enlist the

assistance of community-based organizations in preventing

students from dropping out of school (P.L. 101-600, 1990).

P.L. 101-600 (1990) also includes the mandate that

requirements for obtaining a certificate of graduation from a

school not be lowered. During the 1980s, legislatures pushed

to improve public secondary schools by increasing graduation

requirements. The increase in graduation requirements, as

well as reports such as A Nation At Risk (National Commission

on Excellence in Education, 1983) and Action for Excellence

(Task Force on Education for Economic Growth, 1983), have

increased the need for teachers to train students in specific

strategies for successfully mastering the required skills and

curriculum needed for completion of high school.

In 1989, the nation's Governors and the President

established the National Education Goals Process, a framework

to change the national educational emphasis from process to

performance. The National Goals process has three essential

parts. First, in early 1990, six national goals for

educational improvements were developed to be the basis for

Blueprint 2000. Second, the creation of the National

Education Goals Panel took place later in 1990. The purpose

of this panel was to report annually the state and national

progress toward the six goals. Once a year, the panel

provides a report with the most current information available

on where the nation's progress towards achieving its goals

stands. Third, the development of the National Education

Standards and Assessments Council (NESAC) was promoted in

1992 to coordinate the development of standards of

achievement and a system for assessing their attainment.

Goals two and five both focus on high school completion

rates. Germany and Japan now have higher secondary school

completion rates than the United States (National Education

Goals Panel, 1992). The objectives in goal two include a

dramatic reduction in the nation's dropout rate. Ninety

percent of students who drop out of school are expected to

successfully complete a high school degree or its equivalent

by the year 2000. The second objective states that the gap

in high school graduation rates between American students

from minority backgrounds and their nonminority counterparts

will be eliminated by the year 2000. The high school

completion rate in 1991 was 85% for 19- to 20-year olds.

Rates for African Americans and Anglo students were

substantially higher than the rate for Hispanics.

This goal is important to attain for several reasons. A

student who dropped out of school before graduating in 1990

can expect to earn less than one-half as much as similar

youth in 1973 (National Education Goals Panel, 1992). Over a

lifetime, this dropout will earn about $200,000 less than a

high school graduate. Another consequence of dropping out of

school affects our economic system. One-half of welfare

recipients failed to finish high school. Additionally, among

the 1.1 million persons incarcerated in 1990, at a cost of

$22,500 a year per person, 82% were high school dropouts

(National Education Goals Panel, 1992). Helping students

succeed in school and graduate is more cost effective than

paying for the consequences of dropping out of school.

Goal five focuses on lifelong learning and literacy

skills. In 1991, 97% of young adults had mastered the most

basic functional literacy skills, but few were able to

perform more complex tasks requiring the synthesis of many

pieces of information (National Education Goals Panel, 1992).

Students' failure to achieve academically and socially is a

difficult problem to address due to the numerous and complex

contributing factors.

One aspect of preventing school failure is to increase

academic success. As students progress through the

educational system, the curriculum places increasing demands

on them for acquiring and memorizing information (Nagel,

Schumaker, & Deshler, 1986). Student progress is typically

assessed by written means that reflect factual recall (Blough

& Schwartz, 1990; Mastropieri & Scruggs, 1991). Scruggs and

Mastropieri (1993) examined assessment methods in content

areas and revealed that 99% of all content assessments

required direct recall of factual information. They also

stated that the demands on vocabulary learning are

particularly evident. Yager (1983) reported that more new

vocabulary was introduced in some content textbooks than was

presented in foreign language courses. Increasing student

success on vocabulary learning is a beginning step towards

addressing the complex issues of reducing the dropout rate.

Vocabulary knowledge may be one of the most important

abilities for achieving school success (Scruggs &

Mastropieri, 1993).

Exceptional educators have been providing and

investigating numerous strategies and techniques to increase

student success since the 1970s. Alternative instructional

strategies that are effective with exceptional students may

also be effective in regular education programs (Beech,

1989). Despite the similarities in academic failures between

students who are at risk of dropping out of school and

students who are staffed into exceptional education programs

such as SLD, EMH, and EH, there has been little research on

learning strategies for at-risk adolescents. With a better

understanding of the effectiveness of specific instructional

strategies for teaching at-risk adolescents, school systems

may be able to plan and train regular educators to use these

effective strategies and, thus, increase at-risk students'

academic success.

In this study the researcher compared student

performance of recall of unfamiliar geography terms when the

students were provided instruction in learning the keywords

and interactive illustrations, when the students were

provided instruction in utilizing context clues, and when the

students were instructed to use rote free-study techniques.

Definitions of Terms

An at-risk student is a student who is experiencing

academic underachievement, antisocial behaviors, academic

failure, and/or truancy (Kaplan & Luck, 1977) and is

considered to be at risk of dropping out of school prior to


Contextual analysis is a strategy that refers to a

reader's attempt to understand the intended meaning of a word

by scrutinizing surrounding context (Johnson & Pearson,


A dropout is a student who has left the school system

prior to graduation and is no longer attending school.

Imagery refers to mental images produced by memory or


An imagery link is an illustration that links the

keyword to the definition of the vocabulary word via an

illustration of a picture or an imagining of the keyword

interacting in some way with a picture of the vocabulary

word's definition.

Keyword mnemonic method is a strategy involving the

association of the phonetic and visual imagery components of

a word with its definition (Atkinson & Raugh, 1975).

Learning strategy refers to techniques, principles, or

rules that enable a student to learn to solve problems and

complete tasks independently (Deshler & Lenz, 1989).

Middle school refers to schools established to teach

students in grade levels 6 through 8.

A mnemonic strategy is a device for organizing and/or

encoding information through the creation and use of

cognitive cueing structures (Bellezza, 1981).

Pictorial and graphic context clues are graphic devices

such as pictures, charts, graphs, diagrams, tables,

timelines, and maps that students may "read" to help

comprehend new words, general concepts, and main ideas.

Recoding is the process that transforms unfamiliar,

nonmeaningful stimuli into a more meaningful entity (Levin,


Relating is the process that integrates initially

unrelated elements into a meaningful whole (Levin, 1983).

Rote free-study is the process of memorizing material by

repetition or routine that is carried out mechanically,

unthinkingly, and in any way that one can.

Semantic contextual clues are clues that include direct

definitions or explanations, restatements, substitute words,

figures of speech, comparison or contrasts, summary

statements, inferences, subjective clues, and familiar

expressions. The bulk of contextual analysis is

semantic. There are many kinds of semantic contextual clues

available to the reader.

Syntactic contextual clues do not provide direct

definitions of unfamiliar words, but they do help a reader to

decipher word meaning from word order. Certain grammatical

form classes (parts of speech) "fit" in certain places in a

structure where others do not.

The three "Rs" refers to the process of the keyword

mnemonic strategy of stimulus recoding, semantic relating,

and systematic retrieving of a mnemonic (Levin, 1983).

Typographical context clues are quotation marks,

parenthesis, and definitional footnotes. These clues allow

the readers to tie previous experience to the materials they

are reading.

Word comprehension refers to knowledge of word meaning.


The scope of this study was delimited in several ways.

First, geographically this study was restricted to Orange

County in Central Florida, a large, fast growing school

district with more than 100,000 students. Second, the

subjects for the study were seventh grade middle school

adolescents who have been identified as at risk for dropping

out of school. Third, only subjects attending public middle

schools were considered for the study.


Since this study included only at-risk middle school

students, the findings should not be generalized to

handicapped, nonhandicapped, elementary, or high school

students. Moreover, the results of this study cannot be

generalized to vocabulary terms in other content areas

without replication. Further, caution should be exercised in

extrapolating the results of this study to students who live

outside of Orange County and who attend schools other than

the public sector.


Dropouts are a major national problem confronting urban

school districts. School officials recognize the burden

placed on society by large numbers of students who lack

necessary educational skills. Educators, however, are

seeking a variety of methods to increase self-esteem and

school success in hopes of encouraging at-risk students to

remain in school.

The intent of this study was to contribute information

regarding the effectiveness of teaching geography vocabulary

to at-risk students through three different memory

strategies: keyword mnemonics, contextual analysis, and rote

free-study. Specifically, immediate acquisition and delayed

recall were investigated for these teaching strategies of

vocabulary terms. The results of this study should have

direct and immediate implications for regular education

teachers who teach vocabulary terms to at-risk students.


Chapter 2 contains a review of literature pertinent to

this study. Chapter 3 includes the hypotheses, method and

procedures of the study, and treatment of the data. Results

of the study are presented in Chapter 4, and discussion,

conclusions, and implications for instruction and future

research make up the fifth, and final, chapter.


The purpose of this chapter is to investigate, evaluate,

and summarize existing relevant information related to memory

techniques that pertain to keyword mnemonics, contextual

analysis, rote free-study strategies, and the use of visual

imagery. The problem addressed by the present research, the

effects of memory strategy training on the performance of

at-risk adolescents' vocabulary acquisition and retention,

has not appeared in the literature, although several studies

concerning the effect of memory strategy training on other

populations have been reported. The available studies most

closely related to the present research have been chosen for

review and discussion. The inclusion criteria for research

reviewed are stated first. Second, a historical summary of

memory and, specifically, of the literature supporting the

use of both the keyword mnemonic strategy with mildly

disabled adolescents and the contextual analysis strategy

with various populations is presented. Third, a description

and a critique of research that experimentally compares the

keyword mnemonic strategy and the contextual analysis

strategy with other memory strategies is provided.

Inclusion Criteria

The following criteria were used to select the

information reviewed.

1. The experimental questions) of each study must

include either the effectiveness of keyword mnemonic,

contextual analysis, and/or rote free memory strategies.

2. The dependent variable of each study must include

recall of factual information.

3. Each study must include a complete research report

of data, including a description of subjects, methodology,

and results.

Sources used for locating applicable information

included card catalogs and computer systems at the University

of Florida, University of Central Florida, and Rollins

College libraries; Current Index of Journals in Education;

Educational Resources Information Center; the interlibrary

loan services; the reference sections of related articles,

chapters, and books; and personal communication with other

investigators who have conducted research using memory

strategies or studied learners who are at-risk.

Background Information on Memory Strategies

The teaching of memory strategies has evolved since the

days of ancient Greece and Rome (Minninger, 1984; Wittrock,

1986). Learning how to remember information was devoted to

rhetoric, the art of public speaking. Learning how to

memorize the sequence of points in speeches was an important

part of rhetoric because paper and pencil were not readily

available. Teachers and orators used memory strategies

derived from Aristotle's model of memory, which emphasized

putting points in a linear sequence, with each point

represented by an image involving an interaction with an

easily retrievable familiar object.

Prior to Twentieth Century

Many of Plato's and Aristotle's ideas are still relevant

today (Anderson & Bower, 1973; Weimer, 1973). Both

philosophers formulated sophisticated theories about the

acquisition of knowledge (Hett, 1964; Lamb, 1967). Aristotle

divided memory into two parts: memory, a type of remembering

common to all animals; and recall, a superior human action,

involving a sense of time, deliberation, and the relation of

a chain of experiences to a present experience (Minninger,

1984). Plato, Aristotle's teacher, described mechanics of

memory by calling up a detailed picture of events and

associations from a single item (e.g., a lyre making one

think of its owner, a special town, and then friends).

Later in the 1800s, Hermann Ebbinghaus was one of the

first people to systematically research memory processes. He

studied how fast information could be acquired and how long

information could be kept in memory. He invented strings of

nonsense syllables and recited sequences of them aloud to

test how many repeats were necessary before he could memorize

them (Ebbinghaus, 1885/1913). He tested himself at the same

time every day under the same conditions, but varied all

other factors such as the length of the lists or the time

between memorization and recall. He discovered that he could

remember much better first thing in the morning and concluded

that fatigue affected memory.

Twentieth Century

The next important group of studies of memory strategies

were conducted in the early part of the twentieth century by

Sir Frederick Bartlett in England (Bartlett, 1932). He felt

that Ebbinghaus' approach was artificial and designed

experiments to investigate subjects' ability to use known

information to facilitate recall by integrating new thoughts

and patterns into the previously learned information. His

subjects looked at pictures and read stories. They used

personal associations and prior experiences to help recall

specific attributes and facts about the material. Bartlett

(1932) concluded from these studies that memory occurred by

laying a new pattern of information over a similar old one.

His studies lacked strict controls necessary to ensure a

uniform basis of comparison, and at the time, many felt his

work was too vague and complex (Minninger, 1984).

Two-System Theory

Before the 1950s, memory was thought of as one system.

Then, a two-system theory became popular: a short-term

memory for our immediate environment and a long-term memory

for "facts" (e.g., language, history, math). During this

period, researchers discovered that even the ability to

remember information for fifteen or twenty seconds depended

on activities performed by the learner (Bransford, 1979).

Peterson and Peterson (1959) investigated the rate of

forgetting strings composed of three consonants; for example

JHB on trial 1, TSR on trial 2. Students were required to

count backwards by threes from a number they were given in

order to prevent the use of a rehearsal strategy. The

intervals between presentation of a particular letter string

and the request for recall included three, six, nine, twelve,

fifteen, and eighteen seconds. The results from Peterson and

Peterson's strategy indicated that without rehearsal recall

dropped as the retention interval increased from three to

eighteen seconds.

Waugh and Norman (1965) also studied the effects of

short-term memory while preventing rehearsal strategies.

Waugh and Norman, however, presented people with additional

items and told them not to rehearse. Their results are

similar to those of Peterson and Peterson (1959). When

rehearsal is prevented, short-term memory is influenced by

the number of intervening items.

Multiprocess Memory Theory

Ellis (1970) adopted the multiprocess memory theory

based on the theoretical conceptions proposed by Waugh and

Norman (1965). This theory supports the belief that

information or external stimulation is taken in by the

learner through the attention process. Next, this

information goes directly to the primary memory, which is

considered a limited storage system. This level of memory is

only capable of holding small bits of information

momentarily. Without further memory processing or with the

passage of more than a few seconds, primary memory will be

lost. Therefore, rehearsal strategies are necessary and are

the next process in Ellis' model. The learner uses these

rehearsal strategies to try and retain information that is in

the process of being lost from any of the three memory

systems--the primary memory, the secondary memory, or the

tertiary memory. In Ellis' multiprocess memory theory, these

rehearsal strategies are considered as the means for which

information is transferred from one memory system to the

next. This transfer of information makes retention of

information possible.

Constraints on Short-Term Memory

In addition to Peterson and Peterson (1959), Waugh and

Norman (1965), and Ellis (1970), other investigators have

emphasized the constraints on memory strategies for short-

term retention (Chase & Simon, 1973; DeGroot, 1965; Miller,

1956). These researchers all agreed that effectiveness of a

memory strategy for short-term retention depended on the

degree to which one used previous knowledge to encode

materials into meaningful units. Other theorists (Atkinson &

Shiffrin, 1968) postulated that there is a limit to the

amount of information that can be held in short-term memory

and that a transfer of information from short-term storage to

long-term storage is an important aspect of the memory and

learning process. Atkinson and Shiffrin (1968) further

classified the two dimensions of the memory system. The

first categorization distinguished permanent, structural

features of the memory system that included both the physical

system and the processes that are invariant from one

situation to another. The other dimension referred to the

control processes that are modifiable by the individual and

may vary from one task to the next. The use of a particular

control process for remembering information in a given

situation depends upon the type of instructions, the

meaningfulness of the material, and the individual's learning

history. In other words, the structural features include the

basic memory stores and the control processes involved in any

coding procedures, rehearsal operations, and categorization

strategies. Other theorists and researchers have argued that

automatic processes in cognition have some basic operating

characteristics of the information processing system. These

theorists support the belief that memory encoding includes

fundamental aspects of experience.

Automatic Processes

Hasher and Zacks (1979) conducted studies to support the

theory of the use of automatic processes in cognition and to

specify the details of their operation. These authors

hypothesized that some automatic processes are inborn and

therefore require no attentional resources. They claimed

that frequency of occurrence, spatial location, and the order

of events are automatically encoded into long-term memory and

continually register in memory at an optimal level. Hasher

and Zacks (1979) stated that six criteria must be jointly

satisfied in order to make certain that a process, aspect, or

attribute of experience is automatically encoded. According

to these criteria, encoding of frequency, temporal order, and

spatial location information should not be affected by

intention, age, and simultaneously processing demands, or by

practice and individual differences.

The frequency of occurrence of events has been the most

studied aspect of the environment for its automaticity (Ellis

& Palmer, 1988; Hasher & Zacks, 1979, 1984). Two other

aspects of the environment, spatial location and temporal

order information, were also thought to be encoded

automatically, although they were not as thoroughly

investigated as frequency of occurrence. Although Hasher and

Zacks (1979) relied on several studies (Attneave, 1953;

Howell, 1973) to support their automaticity theory for

spatial location information, other researchers (Ellis, 1990,

1991; Naveh-Benjamin, 1987, 1988) revealed that the results

of some tests of criteria for automaticity are ambiguous,

whereas other criteria have not been studied at all.

Naveh-Benjamin (1987) conducted four experiments that

examined five criteria for the automaticity of cognitive

processes. Results showed that memory for spatial location

information was influenced by intention, age of subjects,

competing task loads, practice, strategy manipulations, and

individual differences. This author's results contrasted

Hasher and Zacks' (1979) theory that spatial location

information is automatically encoded.

Ellis (1990, 1991) conducted research examining the

effects of memory for spatial location and concluded that

memory included both effortfully and automatically processed

subtasks. Ellis' research did not support all components of

Hasher and Zacks' theory, but the results of his experiments

uphold the main idea of their position. Ellis' (1991) data

suggested that memory for location is not consistent in tasks

requiring memory for long series of location. Ellis (1991)

stated that memory for a limited number of locations in such

a series can be improved strategically and is facilitated by

instruction, and that this augments the automatic memory

component. He concluded that the remembering of locations of

objects in the environment is predominantly an automatic

process even though the Hasher and Zacks (1979) criteria for

defining memory for spatial location as an automatic process

were not met in his experiments.

Memory theories have evolved throughout the years.

Early theorists divided memory into two parts based on the

capability of the brain, but a two-system theory did not

emerge until the 1950s. Later, the memory system was further

classified and included the transfer of information into

different storage systems. Next, memory theorists focused on

the use of automatic processes in cognition. Research that

evolved from these theories based on the transfer of

information from short-term storage to long-term storage and

the existence or nonexistence of automatic processes has lead

to numerous studies on devices, procedures, and operations

that may improve memory.

Background Information on Mnemonic Strategies

One strategy that effectively incorporates what a

student has already learned is mnemonic instruction. A

mnemonic device can be defined as a strategy for organizing

and/or encoding information through the creation and use of

cognitive cueing structures (Bellezza, 1981). Mnemonic

devices have been used for more than 2,000 years (Higbee,

1979), however, controlled studies of mnemonic effects were

limited until 1975, when Atkinson published an experimental

study of the "keyword" method for teaching Russian

vocabulary. Atkinson's research initiated a renewed interest

in mnemonic strategies. Organization for many mnemonic

strategies usually involves a strong imagery component.

Typically, imagery represents the to-be-remembered

information by implying a sequence of symbolic

transformations, which go from words to images, and then back

to words (Paivio, 1971). These mnemonic strategies have not

only helped students to acquire unknown information, but have

also given students a strategy for retrieving information.

Mnemonic instructional procedures are taught to provide a

retrieval link between stimulus and response information,

thus, facilitating later recall (Mastropieri, Emerick, &

Scruggs, 1988).

Researchers have established that pictures make more

substantive events than do words (Bower, 1972; Pressley,

1977; Rohwer & Harris, 1975) because of the greater

concreteness of pictures. Levin, Divine-Hawkins, Kerst, and

Guttmann (1974) studied fourth graders' acquisition of paired

associates. They found that for 20% of the students the

nature of the materials presented was critical on acquisition

rates. That is, students functioned poorly when words were

presented, but excelled when pictures were presented. Levin

(1983) described a trio of assumptions (and associated

corollaries) underlying the use of pictorial imagery

strategies for school learning:

1. Pictures can substantially improve students'
learning of school content.
la. Pictures should be used as school learning
2. The degree of picture facilitation expected depends
on the relationship between the particular learning
task and the kind of pictures provided or generated.
2a. Pictures that are directly related to the task
content and component processes will be more
effective than those that are not.
2b. Pictures that transform task content into a
more meaningfully coded form will be more
effective than those that do not.

3. Picture effects can be expected to vary as a
function of relevant student characteristics.
(p. 214)

Levin (1983) further explained that the potentially

valuable result of picture use and instruction was one of

training learners to transfer from a reliance on externally

provided cues to the skill of producing internally generated

cues. Levin claimed that pictures are more concrete than

words in that they provide learners with a closer

approximation to their environments. For children's and

adult's learning of unconnected materials, pictures have been

superior to words, in tasks involving both recognition and

recall memory (Levin, 1976). Various types of illustrations

have been incorporated into mnemonic methods. One

alternative mnemonic procedure that incorporates mental as

well as pictorial images for recall is the peg-word mnemonic


Peg-Word Theory

The peg-word theory is one of the oldest and most

influential mnemonic processes described in the Roman books

on rhetoric. Based on this theory, Paivio (1971) proposed

that the stimulus member of a pair functions as a "peg" to

which its associate is connected during learning, and from

which it can be retrieved during recall. According to

Paivio, the more the concrete the stimulus, the more "solid"

it is as a conceptual peg, and the better the recall.

Concreteness can be defined in terms of its "image-arousing

value," the speed and ease with which the word arouses some

mental image (Bower, 1972). On recall trials, then, the

image-arousing value of the stimulus is critical, because it

reintegratess the compound image from which the response

component can be retrieved and recorded as a word" (Paivio,

1971, p. 248). Paivio (1971) developed the "conceptual peg

hypothesis" based on studies of paired-associate tasks.

Pair-Associate Tasks

Bower (1972), Levin (1976), and Rohwer (1966) have

concluded that when two unrelated nouns are paired, both the

verbal and imaginary elaborations that relate the two nouns

in a meaningful experience are more powerful than

elaborations that maintain the separate characters of the two

items. Pictures, as applied to most paired-associate tasks,

can either be improved, provided by an experimenter or

instructor, or induced, generated by a subject. Bower (1972)

attempted to create a high memory load for his college-aged

subjects by presenting them with five 20-pair lists. Imagery

subjects were instructed to generate interactions of the

associated noun items, whereas control subjects were given an

equivalent amount of time to study the lists. Results, based

on the recall of the 100 pairs, showed that imagery subjects

remembered twice the amount of information than did control

subjects on both short-term and long-term assessments.

Additional Memory Factors

Some researchers have identified specific factors that

affect a subject's ability to benefit from both visual and

verbal memory strategies. Levin (1976) addressed the

variable of age. He suggested that students at all age

levels possess a wide range of cognitive-developmental

abilities. He maintained that this results in great

performance variability. Age, therefore, interacts with

other variables such as socioeconomic status, intelligence,

and learning styles in determining performance for the

learning of paired associates for words.

As previously mentioned, an additional factor in the

success of remembering a word is its concreteness.

Therefore, abstract information must be encoded into a

concrete word or picture symbol, so that it can fit into some

organizational scheme (Bellezza, 1981; Bugelski, 1970;

Paivio, 1968). According to Bower (1972), abstract words

presumably evoke little or no imagery directly, but may do so

indirectly through associated words, such as heart for

"love," or church for "religion."

Semantic encoding procedures have made abstract words

and concepts more memorable in paired associate learning

tasks using unconnected elements. For example, Paivio (1968)

studied paired associates using 2 ten-item lists of concrete

nouns. The purpose of his investigation was to observe the

effects of imagery or no-imagery conditions. For List 1,

subjects were presented the to-be-recalled nouns preceded by

the numbers 1 to 10. They were not taught a mnemonic

strategy. The numbers were then presented in a random order,

and subjects were directed to recall the corresponding items.

For List 2, subjects were assigned to an imagery instruction

or no-imagery instruction condition and were taught either a

concrete (e.g., one-bun; two-shoe; etc.) or an abstract

(e.g., one-fun; two-true; etc.) pegword mnemonic. Subjects

in the imagery group were instructed to use mental images to

relate the pegword rhymes with the to-be-remembered items.

For example, if the to-be-associated item with one was

pencil, students in the concrete pegword group would be asked

to construct an interactive mental picture involving the

pegword, bun, and the to-be-associated item, pencil. A

likely image would be that of a pencil inside a hot dog bun.

Students in the abstract pegword group, on the other hand,

would need to form a mental image of the pegword, fun,

interacting with the pencil. Because the abstract term, fun,

cannot be directly represented in a picture, the students had

to encode fun into a concrete referent in order to put it in

a scene with pencil. Subjects who were not given imagery

instruction were told to recall the list by saying to

themselves the rhyming words, along with the to-be-remembered

item (e.g., "one-bun-pencil" or "one-fun-pencil").

Paivio found the recall was better for the subjects in

the mnemonic condition (List 2) than for subjects studying

the control list (List 1). Subjects who received imagery

instructions also had better recall than those who received

no-imagery instructions. The method appeared to be equally

facilitative for both concrete and abstract pegwords. Paivio

concluded that when the subject is presented with abstract

items, semantic encoding operations transform this

information into concrete, familiar illustrations or images.

Another mnemonic strategy that has been successfully utilized

for rendering abstract and unfamiliar terms to more concrete

has been the mnemonic keyword method (Levin & Pressley,

1985), which is a phonetically based operation.

Keyword Mnemonic Strategies

In 1975, Richard Atkinson of Stanford University was one of

the first researchers to study the use of mnemonic devices in

educational settings (Higbee, 1979). Atkinson and Raugh (1975)

used college students as subjects to demonstrate the benefits of

a mnemonic imagery strategy for learning foreign language

vocabulary. This mnemonic technique, the keyword method, is a

two-stage process connecting an auditory-perceptual link to an

imagery link and has been beneficial as a strategy to enhance

the acquisition of native-language vocabulary (Levin & Pressley,

1985). Basically, the keyword method attempts to enhance

learning and memory by facilitating the encoding of information

so that the information can be easily retrieved. The vocabulary

term to be taught is first reconstructed to a concrete,

previously learned, acoustically similar proxy, or "key word."

Next, the reconstructed keyword is shown in an interactive

picture with its response. Effective conversions assimilate the

"three Rs" of associative mnemonic techniques: (stimulus)

recoding, (semantic) relating, and (systematic) retrieving

(Levin, 1983).

First, during the recoding component a learner transforms

an unfamiliar term into a concrete, familiar word that sounds

like a salient part of the new vocabulary word. For example, a

new geography term "fjord" could be transformed into the keyword

"board" (see Figure 1)



"fjord" transformed into the keyword "board"

Figure 1. Recording Component

Second, the learner applies a relating component, where

the keyword becomes linked to a desired response through an

interactive scene. For "fjord," a narrow arm of the sea

bordered by steep hills, a board would be drawn connecting

two steep hills on each side of a narrow strip of water (see

Figure 2).


a narrow
arm of the
sea bordered
by steep hills .-

"fjord," a narrow arm of the sea bordered by steep hills; a
board connects two steep hills on each side of a narrow strip
of water.
Figure 2. Relating Component.

Lastly, the third component of the method provides the

student with a systematic means of retrieving the meaning of

the geography term. In the present example, to retrieve the

meaning of "fjord," the student would be led directly from

the stimulus (fjord) to the keyword (board) to the

interactive picture (a board connecting two steep hills over

a narrow body of water) to the correct response (a narrow arm

of the sea bordered by steep hills) (see Figure 3).

Levin (1981b) described several psychological principles

of the keyword method: (a) meaningful stimuli are more

reliably encoded than are nonmeaningful stimuli;

(b) interacting items are more reliably associated than

noninteracting items; (c) the greater the similarity between

the two stimuli, the more reliably one will evoke the other;


"fjord"--student was led from the stimulus (fjord) to the
keyword (board) to the interactive picture (a board
connecting two steep hills over a narrow body of water) to
the correct response (a narrow arm of the sea bordered by
steep hills).

Figure 3. Retrieving Component.

and (d) thematic interactions are reliably retrieved from

appropriate cues.

Some researchers who have examined the effect of keyword

mnemonic instruction have compared mnemonic keyword picture

conditions with no-picture control conditions (Pressley &

Levin, 1978), making it difficult to determine whether it was

the specific use of the mnemonic keyword picture or the

utilization of pictures in general that was responsible for the

increased acquisition of terms to the mnemonic keyword

conditions. In one study, however, Levin, McCormick, Miller,

Berry, and Pressley (1982) presented subjects with mnemonic

keyword pictures and compared these subjects to subjects who

were provided with nonmnemonic pictures. The subjects in the

mnemonic keyword picture condition outperformed the subjects

shown nonkeyword pictures. The subjects in this study were

fourth grade students who were presented difficult English

vocabulary words. The subjects who were provided with

nonkeyword pictures relating each word with its contextual

definition performed at the same level as the no-picture

control condition. In contrast, the subjects provided with

mnemonic keyword pictures, interacting the keyword to the

vocabulary word's definition, substantially outperformed

subjects in the control condition. These mnemonic keyword

pictures displayed objects or events that were not mentioned in

the text and that had little or no relation to the conceptual

content of the material being learned (Bellezza, 1981).

According to Levin (1983), the representation,

organization, and interpretation functions of pictures merely

reiterate, consolidate, and symbolize, respectively, the

information presented. A truly mnemonic picture represents a

physical recoding of the textual information and must implicate

the transformation function. Mnemonic pictures, therefore,

must contain different information from that in the text,

because the illustrations or images have provided a physical

recoding of the to-be-remembered information. Mnemonic

instruction has been successfully used with both educationally

disabled and nonhandicapped populations in a variety of content

areas (Levin, 1983; Levin, McCormick, Miller, Berry, &

Pressley, 1982; Levin, Pressley, McCormick, Miller, & Shriberg,

1979; Mastropieri, Scruggs, & Levin, 1985a; Mastropieri,

Scruggs, & Levin, 1985b). This instruction has been used in

classes such as foreign language, English, science, and

geography. Mnemonic instruction has been particularly

successful in helping mildly disabled students learn an

effective and efficient way to remember unfamiliar definitions

(Scruggs & Mastropieri, 1992). If mildly disabled students

have been successful at acquiring specific strategies to learn

unfamiliar definitions, then at-risk students may benefit from

similar strategy training to help them remember definitions of

unfamiliar terms.

Keyword Mnemonics with Students with Mild Disabilities

The effects of keyword mnemonic instruction for students

with mild disabilities, as represented in the literature

since 1986, are reported and comprise a total of eight

studies. The majority of the studies were conducted with

adolescents who had learning disabilities (LD). Exceptions

to this population included seven students that attended a

varying exceptionalities (VE) class who were classified as

emotionally handicapped (EH) (King-Sears, Mercer, &

Sindelar, 1992) and one mildly developmentally delayed

student who was included in a resource class with learning

disabled students (Scruggs & Mastropieri, 1992). Another

study was a pilot study conducted with at-risk adolescents

(Fox, King, & Evans, 1987). One of the studies contained an

additional experiment (McLoone, Scruggs, Mastropieri, &

Zucker, 1986); thus, a total of nine studies were analyzed.

Other keyword mnemonic research exists with the learning

disabled population, but these eight studies are

representative of the research evaluating the effectiveness

of mnemonic instruction individually administered, taught to

small groups of 2 to 4 students, and taught to larger groups

of 10 to 18 students per class. Most of the research

concerning keyword mnemonic instruction with mildly disabled

students has focused on the acquisition and retention of

vocabulary terms or specific concepts, particularly those in

social studies and science curriculum.

The evolution of the research documenting the

effectiveness of the keyword mnemonic strategy for learning

disabled students has followed a progression of replications

and extensions of previous findings: one-to-one teaching and

testing situations using trained experimenters (Mastropieri,

Scruggs, Bakken, & Brigham, 1992; Mastropieri, Scruggs, &

Fulk, 1990; McLoone, Scruggs, Mastropieri, & Zucker, 1986;

Scruggs & Mastropieri, 1992; Veit, Scruggs, & Mastropieri,

1986); randomized subjects to treatment conditions

(Mastropieri, Scruggs, & Fulk, 1990; McLoone, Scruggs,

Mastropieri, & Zucker, 1986); small groups randomized to

treatment conditions (Veit, Scruggs, & Mastropieri, 1986);

brief intensive experimental sessions (Mastropieri, Scruggs,

& Fulk, 1990; McLoone, Scruggs, Mastropieri, & Zucker, 1986);

highly unfamiliar concrete materials to classroom mnemonic

instruction (Fox, King, & Evans, 1987; King-Sears, Mercer, &

Sindelar, 1992; Mastropieri, Scruggs, Bakken, & Brigham,

1992; Scruggs & Mastropieri, 1989; Scruggs & Mastropieri,

1992; Veit, Scruggs, & Mastropieri, 1986); existing

classrooms counterbalanced across treatment conditions

(Mastropieri, Scruggs, Bakken, & Brigham, 1992; Scruggs &

Mastropieri, 1989; Scruggs & Mastropieri, 1992); situations

involving existing classroom teachers (King-Sears, Mercer, &

Sindelar, 1992; Scruggs & Mastropieri, 1989); and using

existing curriculum for training purposes (Mastropieri,

Scruggs, Bakken, & Brigham, 1992; Scruggs & Mastropieri,

1989; Scruggs & Mastropieri, 1992; Veit, Scruggs, &

Mastropieri, 1986). A description of each of these

experiments is included in the following review and these

experiments are critiqued in terms of their research designs,

subject characteristics, experimental procedures, materials,

instructional medium, content taught, and measurement

methods. Table 2-1 lists the investigators, subjects,

interventions, and content taught to the subjects with mild

disabilities. Study results are reported and summarized.

Research Design and Description of Conditions

A group research design was used in all reviewed studies

to determine the effectiveness of the keyword mnemonic

strategy on mildly disabled students' recall of information.

Subjects were either randomly assigned to treatment

conditions, in groups that were randomly assigned to a

Table 2-1
Review of Keyword Mnemonic Vocabulary Studies with Mildly Handicanped Subiects

King-Sears, Mercer,
& Sindelar (1992)

37 sixth, seventh,
and eighth graders
-30 SLD
-7 EH

Imposed Keyword
-Teacher provided
-Systematic teaching
Induced Keyword
-Student provided
-Systematic teaching
Systematic Teaching
(direct instruction)

Four vocabulary

12 science terms
per week for 4
weeks (48 terms)

Subjects in the imposed keyword mnemonic groups acquired and remembered more
definitions than subjects in the systematic teaching group. Subjects in the
imposed keyword mnemonic group consistently, and sometimes significantly,
outperformed subjects in the other two groups. There were no significant
treatment effects 3 weeks following instructions.

McLoone, Scruggs,
Mastropieri, & Zucker

Experiment 1

60 SLD seventh
and eighth graders
from three junior
high schools

Keyword Mnemonics
-Teacher provided
Direct Rehearsal
Individually Taught

Two vocabulary
1. 16 low
English terms
2. 16 Italian

Results: The teacher provided keyword mnemonic strategy condition subjects produced
greater recall of vocabulary definitions than the directed rehearsal strategy
condition subjects.


Table 2-1--continued
Investigators Subiects Interventions Content
McLoone, Scruggs, 60 SLD seventh Keyword Mnemonics 13 vocabulary
Mastropieri, & Zucker and eighth graders -Student generated terms
(1986) from three junior -transfer task
high schools -Direct rehearsal 13 vocabulary
Experiment 2 -rote free-study terms
-transfer task
Results: In the transfer, student generated conditions, subjects trained in the mnemonic
condition performed significantly better than subjects trained in the directed
rehearsal condition.

Veit, Scruggs, &
Mastropieri (1986)

64 seventh, eighth,
and ninth graders
-"Reading Disabled"
-one junior high

Pegword Mnemonic
-Picture -Teacher
Direct Questioning
with pictures
Small group instruction
2-4 students

3 facts of 16
1. name (Greek root)
2. attribute
3. reason extinct

Results: Mnemonically instructed subjects scored statistically higher than direct
questioning condition subjects on all immediate and one day delayed recall tests,
except the vocabulary test (recall of meaning for word parts taken from dinosaur

Fox, King, & Evans

16 at-risk sixth,
seventh, and eighth
grade students
One middle school

Keyword Mnemonic
Direct Instruction

35 geography terms
-from 7th grade
Geography text

Results: A t-test for matched pairs revealed no significant difference between the two
treatment groups in recall and matching of terms and their definitions. However,
both treatment conditions did produce significant differences between the pretest
and posttest.

Table 2-1--continued
Investigators Subiects Interventions Content
Scruggs & Mastropieri 26 seventh and eighth Keyword Mnemonics 4 U.S. History
(1989) graders -Teacher provided 1-World War I
-deficient in study Traditional Instruction 2-Great Depression
skills -Teacher provided 3-Roaring Twenties
-inner city 4-Seeds of Conflict
-one junior high
Results: Mnemonically instructed subjects scored statistically higher than traditional
condition subjects. Overall, students earned an average grade of "B" under
mnemonic instruction, and an average grade of "D+" under traditional instruction.

Mastropieri, Scruggs
& Fulk (1990)

25 SLD sixth, seventh,
and eighth graders
-resource room in one
middle school

Keyword Mnemonics
-Teacher provided
Direct Rehearsal
(individually taught)

16 English
vocabulary terms

Results: SLD subjects in the keyword mnemonic condition performed better than SLD subjects
in the direct rehearsal condition on two assessment measures. The first assessment
required subjects to verbally state the definitions of the terms. The second
assessment required subjects to state the term when given a novel instance of the
word. Subjects performed better recalling concrete terms than abstract terms.

Scruggs & Mastropieri 20 sixth, seventh,
(1992) and eighth graders
-19 SLD
-1 EMH
2 self-contained
classes at one
middle school

Keyword Mnemonic
-Picture-Tchr Provided
Traditional Instruction
Mnemonic Transfer
-Student generated
(Group Instruction -
10 students)

4 -Science chapters'
a. vertebrate
b. invertebrate
c. earth history
d. geology

Results: In each classroom, subjects performance was substantially higher under mnemonic
instruction compared with the traditional free-study conditions.

Table 2-1--continued


Mastropieri, Scruggs,
Bakken, & Brigham

29 SLD low SES
seventh and eighth
-26 self-contained
-3 resource room

Keyword Mnemonic
Traditional Pictorial
-teacher provided
-free study
(group taught)

40 states and their

Results: The SLD students receiving keyword mnemonic instructional methods produced
statistically significant differences in both forward and backward recall when
compared to the same instructional procedures excluding the keyword mnemonic

treatment condition, or treatment order was counterbalanced

across classrooms.

All nine of the experiments compared the keyword

mnemonic strategy to traditional teaching control groups such

as directed rehearsal. Veit, Scruggs, and Mastropieri (1986)

compared the keyword mnemonic to a control group taught the

same information using the principles of direct instruction,

including teacher-directed questioning, choral group

responding, fast instructional pacing, and cumulative review.

The mean number of correct definitions for each group was

compared and analyzed using a t-test.

McLoone, Scruggs, Mastropieri, and Zucker (1986) also

compared keyword mnemonic instruction to a direct instruction

control group. This direct rehearsal condition incorporated

techniques from direct instruction procedures (e.g., teacher

states the definitions and the students repeat the definition

back at a rapid pace). Immediately following Experiment 1,

comparing learning disabled adolescents' recall of unfamiliar

vocabulary words using either keyword mnemonic strategies or

directed rehearsal strategies, the same subjects participated

in an extension study, Experiment 2. In the second

experiment, the subjects were trained to transfer the keyword

method or rehearsal method to another, similar list of

vocabulary words using their respective strategies learned in

Experiment 1 independently. This generalization skill was

called a transfer task. The researchers were interested in

examining if subjects, under some circumstances, could learn

to generate their own keywords and interactive images.

Again, mean percentage of correct definitions was compared

using t-tests.

Another study examining aspects of keyword mnemonic

instruction with mildly disabled students was conducted by

Mastropieri, Scruggs, and Fulk (1990). Three additional

research questions were examined concerning the keyword

mnemonic strategy: (a) could LD students benefit from

keyword vocabulary instruction when vocabulary was not

selected for keyword "obviousness?" (b) could LD students

learn abstract as well as concrete terms with the keyword

strategy? and (c) does mnemonic instruction inhibit

comprehension? A direct rehearsal condition was used as a

control that included stating the definitions, having the

students repeat the definitions, and providing for drill and

practice of the definitions. A two-condition (keyword vs.

directed rehearsal) by two-item (abstract vs. concrete)

analysis of variance (2 X 2 ANOVA), with repeated measures on

the item type variable (production recall task and

comprehension task) was employed.

King-Sears, Mercer, and Sindelar (1992) examined the

independent use of the keyword mnemonic method by students

with mild disabilities. These researchers compared a

systematic teaching method, an imposed keyword method, and

an induced keyword method. The systematic teaching method

included a combination of effective teaching techniques

derived from direct instruction procedures. The imposed

keyword method included the provision of a keyword and an

interactive illustration. The induced keyword condition

required students to devise their own keywords and

interactive images. This third condition was important to

examine due to the time and cost spent in developing

materials for teacher-provided keyword mnemonic instruction.

Students' existing varying exceptionalities (VE) teachers

randomly assigned to one of the three treatment conditions.

A repeated measures ANCOVA was used to analyze the data with

IQ as the covariate.

Fox, King, and Evans (1987) investigated the keyword

mnemonic strategy with at-risk adolescents to determine

whether or not similar benefits would occur with this

population as did for LD adolescents while learning

unfamiliar vocabulary terms. The keyword mnemonic strategy

condition was compared to a direct instruction condition.

Students were matched based on the results of two protests

and randomly assigned to one of the treatment groups. A

t-test for matched pairs was calculated to determine whether

there was a significant difference in posttest results

between the two treatment conditions.

Scruggs and Mastropieri (1989) conducted one of the

first studies evaluating learning disabled students' use of

keyword mnemonic strategies to learn existing material in an

assigned U.S. history textbook by their assigned teacher.

Experimenters employed a within-subjects design across

classrooms. Each classroom received instruction on four

chapters: one chapter of mnemonic instruction; one chapter

of traditional, textbook-based instruction; another chapter

of mnemonic instruction; and a final chapter of traditional

instruction. The means of the two groups' test scores were

compared by a t-test for correlated samples.

An additional study that extended the application of the

keyword mnemonic strategy with students with mild

disabilities was conducted by Scruggs and Mastropieri (1992).

The authors of this study also evaluated actual classroom

curricular materials, four chapters from a science text, with

respect to the keyword mnemonic strategy. One purpose of

this investigation was to investigate the extent to which

mildly disabled students could transfer complex mnemonic

strategies to science context and evaluate any negative

effects. A within-subjects design, in which treatment order

was counterbalanced across classrooms, was used to compare

the keyword mnemonic strategy to a traditional teaching

strategy condition. Test score data were analyzed by a 2 X 2

(Classroom X Instructional Unit) analysis of variance with

repeated measures on the instructional unit variable.

Mastropieri, Scruggs, Bakken, and Brigham (1992) also

used a counterbalance design across classrooms to evaluate

the effectiveness of keyword mnemonic strategy for mildly

disabled students' acquisition of 40 states and their

capitals. Subjects in each of two self-contained LD

classrooms received mnemonic and traditional instruction on

different alternating weeks. Another issue these researchers

examined was forward versus backward recall of mnemonically

instructed information. A two condition (mnemonic vs.

traditional) by two response (state vs. capital) analysis of

variance procedure was used to analyze subjects' performance.

Subject Characteristics

In eight of the reviewed studies, all except for one

student had been classified as Specific Learning Disabilities

(SLD) according to federal, state, and local criteria. One

student had been classified as Educable Mentally Handicapped

(EMH), but attended an SLD resource class (Scruggs &

Mastropieri, 1992). Another study included seven students

classified as Emotionally Handicapped (EH) that attended VE

classrooms with other SLD students (King-Sears et al., 1992).

Only one study included the use of keyword mnemonics with

at-risk adolescents (Fox, King, & Evans, 1987). The number

of subjects in the reviewed studies ranged from 16 (Fox,

King, & Evans) to 64 (Veit, Scruggs, & Mastropieri, 1986).

All subjects attended public middle or junior high school.

Sixth, seventh, and eighth graders participated in the

studies conducted by Veit et al. (1986), Mastropieri et al.

(1990), and Scruggs & Mastropieri (1992). The subjects in

the Fox et al. (1987) study attended a special summer program

and were incoming seventh graders, incoming eighth graders,

and incoming ninth graders. Seventh and eighth graders

participated in the remaining five studies. Schools ranged

from inner-city to suburban to rural.

Measurement Methods

All studies reviewed used posttest data to determine the

effectiveness of the keyword mnemonic strategy with mildly

disabled adolescents. Many scores and percentages of correct

responses were reported in all investigations. An

independent t-test was used when two interventions were

compared to determine the statistical significance between

the scores. Two studies used a 2 X 2 ANOVA to determine main

effects. One study used an ANCOVA with IQ as the covariate.

Additional tests were also used such as Mann-Whitney U, a

nonparametric test and the Pearson r to help examine the

correlation between strategy usage and student performance.

All studies reviewed included dependent measure designed

to examine acquisition of factual information. Two

investigations also measured strategy usage (Mastropieri,

Scruggs, Bakken, & Brigham, 1992; Scruggs & Mastropieri,

1992). These experimenters also surveyed students asking

them to rate the different instructional conditions with

respect to (a) how much they enjoyed the instruction, (b) how

much they had learned, (c) how hard they had tried, and

(d) how much they would like to use the method again.

Mastropieri, Scruggs, and Fulk (1990) assessed subjects'

knowledge of vocabulary definitions using a matching format

as well as an orally produced definition of each targeted

vocabulary word. Scruggs and Mastropieri (1989) and King-

Sears, Mercer, and Sindelar (1992) were the only researchers

to evaluate the teacher's perception of the appropriateness

of each type of instructional material in each condition.

Most measurement methods utilized only verbal responses.

Experimenters read the questions and wrote student responses

verbatim. Only three studies included written responses

(Fox, King, & Evans, 1987; King-Sears et al., 1992; Veit et

al., 1986) with the exception of the use of a multiple-choice

test by Scruggs and Mastropieri (1989), and Fox, King, and

Evans (1987), and King-Sears, Mercer, and Sindelar (1992).

The at-risk subjects in the Fox et al. (1987) study both read

and wrote their own answers. Veit et al. (1986) had the

experimenters read the questions, but the students wrote

their responses. King-Sears et al. (1992) required subjects

to read the terms and write the definitions from memory

(production) first, then required subjects to match

definitions to terms (recall). These experimenters also

incorporated two means of scoring, strict and loose.

Mastropieri et al. (1990) also used strict scoring for

complete responses and gave one-half point for partial

answers. Scorers using the strict scoring procedure gave one

point to a response if it was clear and independently

correct. When using the loose scoring procedure, scorers

allowed credit for items that had any part of the correct

response given. These two scores yielded a conditional

probability score that was computed on the probability of the

student correctly applying a vocabulary word part that had

not been correctly answered. Two independent scorers unaware

of instructional conditions scored responses in all of the

studies to establish interscorer reliability.

Experimental Procedures

The treatment conditions ranged from 1 day (Mastropieri,

Scruggs, & Fulk, 1990; McLoone, Scruggs, Mastropieri, &

Zucker, 1986) to 8 weeks (Scruggs & Mastropieri, 1989). The

total instructional time for one session ranged from 15

minutes (King-Sears et al., 1992; Mastropieri, Scruggs, &

Fulk, 1990; Veit, Scruggs, & Mastropieri, 1986) to 50 minutes

(Mastropieri et al., 1992; Scruggs & Mastropieri, 1989,

1992). McLoone et al. (1986) conducted her study in one 25

minute session. All studies kept the amount of time in

treatment conditions equivalent except for Mastropieri et al.

(1992). These experimenters utilized a counterbalance design

across two classrooms. One classroom had three mnemonic and

two traditional weeks of instruction, while the other had two

mnemonic and three traditional weeks of instruction. Scores

of all subjects were converted to percentages due to the

number of test items not being equivalent across classrooms.

Materials. The materials used in five of the keyword

conditions were cards (e.g., 5" X 5", 8-1/2" X 11") that

contained the vocabulary term, its keyword, definition, and

an interactive illustration of the keyword and the term's

definition. The remaining three studies included similar

information, but used overhead transparencies. The materials

in all the control conditions were identical materials to

that of the keyword condition with the exception of the

keyword strategic information.

Instructional medium. Experimenters conducting the

studies were trained in the specific strategies and most were

certified teachers but only two studies utilized the

subjects' assigned classroom teacher to provide the

instruction (King-Sears et al., 1992; Scruggs & Mastropieri,

1989). Subjects were taught and assessed on an individual

basis in two of the studies (Mastropieri, Scruggs, & Fulk,

1990; McLoone et al., 1986). Veit, Scruggs, and Mastropieri

(1986) delivered instruction to small groups of two to four

students by one of two female experimenters. Mastropieri et

al. (1992) and Fox et al. (1987) conducted training sessions

by one male or one female graduate student who were certified

teachers. The regularly assigned special education teachers

remained present during instruction in all the studies

incorporating classroom implementation. The instructors in

the study by Scruggs and Mastropieri (1992) were project

staff from Purdue University who were also certified special

education teachers.

Experimental constants. Attempts to keep the

instructional conditions constant throughout the

investigation phases were reported in all reviewed studies.

All studies included prepared scripts and activities that

were either timed with a stopwatch or pacing cassette tape

(Fox, King, & Evans, 1987; King-Sears et al., 1992;

Mastropieri, Scruggs, & Fulk, 1990; McLoone et al., 1986) or

videotaped (Mastropieri et al. 1992; Scruggs & Mastropieri,

1992; Veit et al., 1986) to ensure standardization of

experimental procedures. The instructors in all of the

studies employed the principles of effective instruction

(e.g., daily review, statement of instructional objective,

guided practice, rapid paced questioning, corrective

feedback). Assessment procedures for all conditions within a

study were also consistent.


Experimenters throughout the evolution of keyword

mnemonic techniques with mildly disabled students from

individual-based investigations with experimenter-developed

tasks, into larger-scaled, classroom-based investigations

using existing curricular materials and teachers have

reported that mnemonically instructed students outperformed

control students on recall of factual information.

Investigators in eight of the reported studies found that

mnemonic instruction resulted in substantial immediate and

sometimes delayed recall gains over more traditional teacher-

led procedures. Only one investigation did not reveal

significant recall differences in favor of the keyword

strategy (Fox, King, & Evans, 1987). Even though no

statistically significant differences were found between the

keyword mnemonic condition and the direct instruction

condition, students in both conditions did significantly

improve their recall of targeted definitions on the pretest

and posttest measures. Six studies (Fox, King, & Evans,

1987; King-Sears et al., 1992; Mastropieri, Scruggs, & Fulk,

1990; McLoone et al., 1986; Veit, Scruggs, & Mastropieri,

1986) compared keyword mnemonics to direct rehearsal

components of instruction. Results from this research are

consistently in favor of the keyword mnemonic condition with

the exception of Fox et al. (1987) who found no significant

differences between the treatment conditions. The remaining

three studies (Mastropieri et al., 1992; Scruggs &

Mastropieri, 1989; Scruggs & Mastropieri, 1992) compared

keyword mnemonics to traditional instructional components

that also included direct rehearsal strategies. Again, these

researchers reported keyword mnemonic conditions

significantly more effective for facilitating the factual

recall of mildly disabled adolescents than traditional

teaching conditions.

The results presented in these studies are conclusive of

studies conducted with mildly disabled adolescents examining

the effectiveness of keyword mnemonics. Although the keyword

mnemonic strategy has been proven effective for mildly

disabled adolescents, studies on the effectiveness of this

strategy with at-risk adolescents are limited. Further

research is needed on the use of keyword mnemonic strategies

instructed by the regularly assigned teacher with mildly

disabled students. Furthermore, research is needed to

determine the effectiveness of memory strategy training for

at risk-adolescents.

The keyword method has been compared both to control

conditions in which no specific instruction was given and to

variations of the learning from context method. Sternberg,

Powell, and Kaye (1983) found the outcome of these studies

supported the keyword method as being "at least as effective

as, and usually more effective than, the alternative methods

against which it. .(was) compared (to)." (p. 126). Other

researchers, however, have found the keyword mnemonic

strategy to produce nearly equivalent retention scores when

compared to conditions such as the contextual analysis.

Background Information on Contextual Analysis

Johnson and Pearson (1978) defined contextual analysis

and the use of context clues as terms that refer to a

reader's attempt to understand the intended meaning of a word

by scrutinizing surrounding context. In essence, using

context clues means educated guessing. With contextual

analysis the focus shifts from the individual word to other

words, the structure of the sentence and other features that

can help pinpoint the meaning of the unfamiliar word. A

critical distinction between mnemonic approaches to

vocabulary learning and contextual approaches is that

mnemonic approaches focus on remembering strategies while

contextual approaches focus on inferring the meaning from

internal and external contextual cues (Jenkins & Dixon,

1984). Although there is some debate about the extent to

which students acquire word meanings from encountering words

in context (Jenkins, Stein, & Wysocki, 1984; McDaniel &

Pressley, 1989; Pressley, Levin, & Miller, 1982), many

reading theorists purport that vocabulary words should be

learned in contexts (Gipe, 1979; Johnson & Pearson, 1978;

Otto & Smith, 1978) that can be synthesized to prior

knowledge and incorporated by a schema (Carr & Wixson, 1986;

Johnson & Johnson, 1986; Nelson-Herber, 1986).

Sternberg and Powell (1983) have proposed that learning

from context involves three basic processes: knowledge-

acquisition processes, contextual cues, and moderating

variables. The authors proposed three processes of knowledge

acquisition. The first process, selective encoding, involved

separating relevant from irrelevant information for the

purposes of formulating a definition. The second process,

selective combination, involved combining relevant cues into

a workable definition. The third process, selective

comparison, involved taking new information about a word and

relating it to old information already stored in memory.

According to Sternberg and Powell's theory (1983) the three

processes of selective encoding, selective combination, and

selective comparison operated together in a stable set of

cues provided by the context in which the new words occur.

Sternberg and Powell (1983) described eight context cues:

1. Temporal cues: cues regarding the duration or

frequency of X (the unknown word), or when X can occur.

2. Spatial cues: cues regarding the location of X, or

possible locations in which X can sometimes be found.

3. Value cues: cues regarding the worth or

desirability of X, or the kinds of effects X arouses.

4. Stative descriptive cues: cues regarding

properties of X (such as size, shape, color, etc.).

5. Functional descriptive cues: cues regarding

possible purposes of X, actions performed, or potential uses

of X.

6. Casual/Enablement cues: cues regarding possible

causes of or enabling conditions for X.

7. Class membership cues: cues regarding one or more

classes to which X belongs, or other members of one or more

classes of which X is a member.

8. Equivalence cues: cues regarding the meaning of X,

or contrasts to the meaning of X.

The moderating variables interact with knowledge-

acquisition processes and these context cues. These

variables make it either easier or harder to apply the

processes to the cues according to Sternberg and Powell

(1983). The moderating variables include the number of

occurrences of the unknown word, the variability of contexts

in which multiple occurrences of the unknown word appear, the

importance of the unknown word to understanding the context

in which it is embedded, the helpfulness of surrounding

context in understanding the meaning of the unknown word, the

density of unknown words, and the usefulness of previously

known information in cue utilization.

Sternberg and Powell (1983) first tested their theory by

asking 123 high school students to read 32 passages of

approximately 125 words in length that contained embedded

within them from one to four extremely low-frequency words.

Thirty-seven of the words (all nouns) were used in the

passages; each target word appeared from one to four times,

resulting in 71 different presentations altogether. The

students' task was to define each of the low-frequency words

within each passage. The qualities of definitions were rated

independently by three trained raters. An average of the

three ratings was used as a definition-goodness score for

each word defined for each subject. These averages were then

averaged over subjects to obtain a mean goodness-of-

definition rating for each word. The main independent

variables were ratings of the number or strength of the

occurrences of contextual cues and moderating variables with

respect to their roles in helping in the understanding of the

meaning of each low-frequency word in the passage. Sternberg

and Powell (1983) concluded that the contextual cues and

moderating variables proposed by their subtheories provided

good prediction of the goodness-of-definition data. All

correlations between predicted and observed goodness ratings

differed significantly from zero (.92 to .77). The data,

although limited, were consistent with the notion of the

proposed theory.

Factors Affecting Contextual Analysis Research

Other researchers have provided evidence that

inconsiderate contexts also prevent context clue instruction

from being as effective as it might be (Graves, 1987).

Carroll and Drum (1983) examined five content area texts

(American literature, English, government, biology, and

chemistry) and found that definitions of target words

generated by students after reading the text varied as the

contexts in which the words were embedded varied. For

example, "The setting of a story- the specific time and place

of the events is rarely a dominant element" generated high

scoring responses for the target word setting. On the other

hand, "setting is stressed in . that fiction that

concentrates on a specific era" produced a wide variety of

definitions for the target word era.

Konopak (1986) found similar results when secondary

students read passages on the same topics from different

physics texts. While one text provided reasonably clear

definitional information, the second text presented the

information in an informed style, such as analogies to

everyday events. For example, in the informed text, kinetic

and potential energy were described as players and reserves;

the players "tear around the field" while the reserves "wait

to be given a chance to move off." While this is an

interesting comparison, a precise definition was not given.

Vacca and Vacca (1986) emphasized the teacher's role in

helping students use context effectively in deriving word

meanings. Furthermore, Sternberg (1987) elaborated that in

typical contextual training programs, including those used in

experimental comparisons to other methods of vocabulary

teaching, the learning-from-context method consisted of

presenting words embedded in a series of sentences, either

with or without prior definitions. But if subjects do not

know the processes, cues, and moderating variables that can

be used for contextual learning, they will not benefit

optimally from such instruction. In other words, subjects

have to learn how to learn from context before they actually

can learn from it. A vocabulary training program that uses

learning from context is incomplete if it fails to provide

instruction in how to use the context (Sternberg et al.,

1983). In the keyword mnemonic method, one is specifically

taught how to form and use keywords. Fair comparison of the

keyword mnemonic method to the contextual analysis method

requires comparable instruction in the context method

regarding how context can be used to infer word meanings.

Gipe (1979) compared four methods of learning new

English vocabulary words. These methods included

association, categorization, contextual analysis, and

dictionary approaches. His subjects were third and fifth

grade students from four intact classrooms at each grade

level. The dependent variable for this study was the total

number of correct words written in the blank space of

provided sentences. The results of this study indicated that

a contextual approach (in which subjects were provided with

sentence contexts and then asked to relate the vocabulary

word to their own experiences) was significantly better than

the other three methods on the sentence recognition task

(p < .001). However, the results of this study should be

interpreted with caution given that the amount of study time

allocated to each condition was uncontrolled. In addition, a

within-subjects design was employed that may have produced

unknown carryover effects from one method to the next.

Gipe (1981) conducted two additional experiments to lend

further support to the contextual analysis method. In the

first experiment, however, Gipe was unable to replicate her

previous findings. In the second experiment, she concluded

that subjects assigned to the contextual analysis strategy

condition performed significantly better than subjects in the

no-strategy control condition. It should be recognized that

the no-strategy control subjects were not given any

instruction. However, they were exposed to the targeted

words and their definitions. All subjects were given a

pretest and a posttest on the definition of targeted words.

These findings must be interpreted with caution given the

methodological flaws. First, when given some exposure

compared to no exposure, the words will more likely be

remembered. As emphasized by Stahl and Fairbanks (1986), a

strategy that provides multiple exposures to word-definition

pairs has a more profound effect on vocabulary learning than

one that gives a single or double exposure to the items to be

remembered. Furthermore, a no-exposure control group seems

to be more appropriately utilized when examining the effects

of vocabulary instruction on an outcome such as reading

comprehension, rather than when contrasting the effects

produced by various vocabulary-learning strategies (Stahl &

Fairbanks, 1986). Stahl and Fairbanks (1986) suggested for

analysis of this latter type of comparison, a no-instruction

control group in which subjects are allowed to study the

targeted words and definitions any way they desire seems to

be the best baseline for establishing the effects of other,

more novel, strategies.

Keyword Mnemonics Compared to Contextual Analysis

Researchers have been examining the effects of keyword

mnemonic strategies compared to contextual analysis

strategies for learning and retaining specific vocabulary

terms (McDaniel, Pressley, & Dunay, 1987; Pressley, Levin, &

Miller, 1982; Sweeny & Bellezza, 1982). In general, the

keyword mnemonic method has produced greater definition

recall than when vocabulary words were presented in context

as represented in the literature throughout the 1980s. Seven

studies comparing keyword mnemonic strategies to contextual

analysis strategies conducted in this period were located and

are reported. Four of the studies contained two additional

experiments (Levin, McCormick, Miller, Berry, & Pressley,

1982; McDaniel & Pressley, 1984, 1989; McDaniel, Pressley, &

Dunay, 1987) comprising a total of eleven studies. These

studies are summarized and critiqued in Table 2-2 and in the

following sections. The research designs, subject

characteristics, measurement methods, experimental

procedures, and results are discussed.

Research Design and Description of Conditions

All experimenters used a group research design to

evaluate the effectiveness of keyword mnemonic strategies

when compared to contextual analysis strategies on the

acquisition and recall of vocabulary meanings. Random

assignment of subjects to treatment conditions occurred in

all reviewed studies.

Ten of the investigations compared a keyword mnemonic

strategy to a contextual analysis strategy incorporating a

three-sentence context. Levin et al. (1982) used a one-

sentence context for a contextual comparison condition in his

Table 2-2
Review of Keyword Mnemonic Vocabulary Studies with Contextual Analysis Methods

Pressley, Levin,
& Miller (1982)

120 introductory
-University of
Western Ontario

Imagery Keyword
Sentence Keyword
Sentence Judgment
Sentence Generate
Sentence Provided
Free-Study Control

32 English vocabulary
-low frequency
-easily identifiable
-concrete imageablee)
-no greater than
3 syllables

The students' recall of word meanings who were in the imagery keyword condition
was statistically better than students' recall who where in the control condition
and all the sentence-contextual conditions, regardless of whether the data were
scored according to strict or more lenient criteria. Subjects in the sentence
keyword condition did not produce significant differences on recall of definitions,
but the performance of sentence keyword subjects was greater than the control
subjects when essence definitions were accepted, and to subjects in the free-study
control group and to subjects in two of the sentence-contextual variations when
meaning fragments were allowed. Learning in none of the sentence context
conditions was significantly greater than that in the free-study control condition.

Levin, McCormick,
Miller, Berry, &
Pressley (1982)

30 fourth graders Mnemonic Keyword
2 elementary schools Free-Study Verbal

Experiment 1

13 vocabulary terms

-thought to be
difficult to grasp

Results: Subjects in the keyword mnemonic condition had a greater mean performance (82.8%
correct) than that of subjects in the control condition (55% correct). These
researchers concluded that fourth grade elementary students could adapt the
keyword mnemonic strategy to learn vocabulary terms that were not selected on the
basis of their being well suited to the keyword method.


Table 2-2--continued

Levin, McCormick,
Miller, Berry, &
Pressley (1982)

Experiment 2

64 fourth graders
-3 classrooms from
one elementary

T$-^rryirv+- i r*M' a

Keyword Context
Experimental Context
Picture Context

14 vocabulary terms

Statistical analysis revealed that students in the keyword context condition
substantially outperformed those in the control condition, the picture context
condition, and subjects in the experimental context differed significantly from

McDaniel &


Experiment 1

69 introductory
psychology students
-University of Notre
-received extra
credit points for

Keyword Mnemonic
Contextual Analysis
Combined Keyword and
Contextual Analysis
Free-Study Control

61 English terms

Subjects in the keyword mnemonic condition had greater totat recall when strict
scoring procedures were used than either the context condition or combined
condition, although not more effective than students in the free-study control
condition. Even when utilizing liberal scoring methods context subjects did not
recall as much as subjects in other conditions. There was no evidence in the
overall analysis to support the position that the context method was a more
effective method of vocabulary learning relative to the free-study control
procedures or to the keyword method. The keyword mnemonic method was not
significantly better than the free-study control method. A supplementary analysis
revealed that students with low-ability in the keyword condition performed
significantly better than the other treatment conditions, but not the high-ability



" I.&_IL.I." V" hd V" I."~~I

n +_T *+

Table 2-2--continued

McDaniel & Pressley

Experiment 2

42 introductory
psychology students
-Indiana University
South Bend
-received extra credit
points for

Keyword Mnemonic
Contextual Analysis

61 English terms

Results: Students in the keyword mnemonic condition outperformed students in the Contextual
analysis condition on recall in both strict and liberal scoring procedures. The
number of correct sentences generated also was significantly greater for students
in the keyword mnemonic condition than students in the contextual analysis
condition. There were no significant differences between conditions if a word's
meaning was recalled then, at least one or two sentences were generated correctly.

Condus, Marshall,
& Miller (1986)

64 12-year-olds
-LD students
-"poor readers"
-4 public elementary

-Teacher provided
Picture Context
Sentence Experience
Control Free Study

50 vocabulary terms
-5 sets of 10 words

Results: All subjects in the three treatment conditions significantly performed better than
subjects in the control condition. Subjects in the keyword-image condition
recalled more definitions in all other treatments across four levels of time
(immediate, end of week, 2 weeks later, 8 weeks later).

Y UU1~~Cc i ~IL~L V ~I C VI1 ~ V~ C~~ r

Conl"^ ve

f r\ *n

Table 2-2--continued
Tnv-t- i tn-r

McDaniel, Pressley,

McDaniel, Pressley,
& Dunay (1987)

Experiment 1

42 introductory
psychology college
-medium sized
Midwestern University

Keyword Mnemonic
Contextual Analysis
-no definition
-3 sentence context

30 English vocabulary

All subjects required at least two study-test trials before they could provide
correct definitions for all the words when provided with the vocabulary terms.
Subjects in the keyword mnemonic condition recalled significantly more definitions
than did subjects in the contextual analysis condition on the first two learning
trials. Despite differences in the acquisition of terms, the different
instructional methods produced nearly equivalent retention scores 1 week after

McDaniel, Pressley,
& Dunay (1987)

Experiment 2

40 college students
-same university as
Experiment 1

Keyword Mnemonic
Contextual Analysis
-3 sentence context

30 English vocabulary

(Similar to Experiment 1) All subjects required at least two study-test trials
before meeting the learning criterion. Subjects in the keyword mnemonic condition
recalled significantly more definitions than subjects in contextual analysis
condition on both learning trials and required significantly fewer trials to meet
criterion. Both groups performed equivalently on the 1 week delayed-cue recall test.

McDaniel & Tillman

45 introductory
psychology college

Keyword Mnemonic
Contextual Analysis
Free-Study Control

60 English vocabulary
-uncommon terms

Subjects in the keyword mnemonic condition produced significantly greater
definition recall than subjects in the context condition when recall was cued with
the vocabulary term. No significant differences were found between subjects
receiving the keyword mnemonic method and subjects receiving the contextual analysis
method for free recall of definitions or free recall of the vocabulary terms.




Tnt- oT i'n-t i nn

Table 2-2--continued

McDaniel & Pressley

Sub ects
75 introductory
psychology college
-University of Notre
-participated for
extra credit course

Keyword Mnemonic
-interactive image
on own
Semantic Context
-3 sentence passage
Free-Study Control

45 vocabulary terms
-Old English

Subjects in the keyword mnemonic condition produced significantly better recall of
definitions than subjects in both the context and free-study conditions. No
differences between the instructional conditions were found when looking at
reading times of embellished and unembellished texts. The meanings of vocabulary
words that were embedded in the embellished texts were better recalled than those
embedded in unembellished texts. For words viewed once, the keyword method
produced better definition recall than either the control condition, or the
context condition. Similar results were found for twice presented words.

McDaniel &


Experiment 2

72 undergraduate
college students
-Purdue University
-part of requirements
in introductory
psychology course

Keyword Mnemonic (e)
Keyword Mnemonic (u)
Semantic Context (e)
Semantic Context (u)
Free-Study Control (e)
Free-Study Control (u)

45 vocabulary terms
-Old English

Key: e = embellished
u = unembellished

Vocabulary instructional techniques did not significantly influence how fast
subjects could access and integrate vocabulary meanings when encountering those
vocabulary terms in text. For vocabulary recall, the context group tended to
recall less than either the keyword group or free-study group. No significant
differences were found between conditions. A subsidiary finding was that the test
text embellishments increased comprehension.



T t- t- r i- UV t-~IL

Tn't" r iront i nn Q

Cr~rni~ n +

first experiment. The percentage of correct responses for

each condition was compared and examined using t-tests. In

Experiment 1, Levin et al. (1982) compared a teacher-provided

keyword mnemonic context condition to a control free-study

condition. In Experiment 2, a picture context condition

without keywords and experiential context condition were

compared in addition to the two conditions of Experiment 1.

McDaniel and Pressley (1984) and Pressley, Levin, and

Miller (1982) also used t statistics to compare instructional

conditions. McDaniel and Pressley (1984) compared a keyword

mnemonic strategy to three other conditions in Experiment 1:

(1) Contextual analysis--three-sentence context, no

definition provided; (2) combined keyword mnemonic and

contextual analysis; and (3) free-study control. To further

convince their colleagues in the field of vocabulary

research, these authors replicated the original study in

Experiment 1. In Experiment 2, the keyword mnemonic method

was compared to the contextual analysis condition of

Experiment 1. An additional measurement procedure was added

in Experiment 2 that required subjects to produce meaningful

sentences with the targeted vocabulary terms.

Pressley, Levin, and Miller (1982) compared two keyword

conditions, one based on the construction of visual images

and the other based on the construction of sentences

(Atkinson, 1975), to four other vocabulary strategy

conditions. Both an imagery keyword and sentence keyword

were compared to each other as well as three different

verbal-contextual approaches to vocabulary instruction and a

control group. The three verbal-contextual conditions

included a sentence-judgment condition, a sentence-generate

condition, and a sentence-provided condition. All

comparisons involving the control condition were conducted as

one-tail tests, whereas all other comparisons were conducted

as two-tailed tests.

Four investigations utilized ANOVAs to examine effects

of the strategy conditions. Condus, Marshall, and Miller

(1986) investigated the effectiveness of the keyword mnemonic

method related to two contextual methods and a control group:

picture context, sentence-experience context, and a free-

study control condition. Experimenters in the picture-

context condition taught subjects vocabulary definitions by

presenting illustrations of the definitions of the targeted

terms. Subjects in the sentence-experience context condition

learned the meanings of words by following a two-step

procedure. In the first step, students listened to and then

reread a three-sentence passage that was printed on a sheet

of paper. No definition was provided. Instead, students

were expected to understand the word's meaning through the

context of the three sentences. In the second step, students

had to relate the meaning of the word to a personal

experience. Finally, students in the free-study control

condition were provided with a list of the targeted terms and

their definitions, pencils, and additional paper.

Experimenters instructed these subjects to choose their own

method of studying to learn the definitions. A two-step

analysis procedure--three-way ANOVA--was used to analyze the

four-treatment condition by the two receptive-language

abilities by the four-time measures, a 4 X 2 X 4 factorial

design with repeated measures. A Scheffe post hoc comparison

was used to analyze significant interactions.

McDaniel and Tillman (1987) also compared keyword

mnemonic condition to a contextual analysis condition and

control condition using ANOVA procedures. Keyword mnemonic

subjects were provided a keyword and told to form an image of

the keyword interacting in some way with the definition of

the vocabulary term. Experimenters instructed subjects in

the context condition to interfere each vocabulary word's

meaning from three sentences provided on a card. These

subjects were also taught internal and external cues

(Sternberg et al., 1983) that may be found in a text to aid

in the discovery of a word's meaning. Control subjects were

told to learn each word and its definition any way they

wanted. A two-factor mixed analysis of variance of the

number of definitions correctly recorded indicated no

difference between treatment conditions.

McDaniel and Pressley (1989) also compared keyword

mnemonics to contextual analysis instruction in two

experiments. In Experiment 1, three conditions were

compared: (1) keyword mnemonic; (2) semantic context; and

(3) free-study control; whereas, in Experiment 2, six

conditions were compared: the three conditions in Experiment

1 and two text conditions, embellished and unembellished

text. Again, the subjects in the keyword condition were

provided a keyword for each targeted term and told to imagine

an interactive image. The subjects in the semantic context

condition were provided both a three-sentence passage and a

one- or two-word definition. For the control condition,

students received each word and its one- or two-word

definition. A three (instructional condition) by two

(embellished or unembellished story) between-subjects ANOVA

of average reading time was used to show differences between

conditions. Comprehension was analyzed with a two-factor

between-subjects ANOVA with instructional condition and text

type (embellished vs. unembellished) as factors.

All subjects were instructed to read two short stories

presented on the computer screen. Fifteen of the targeted

terms were embedded in each story. The particular story

presented to a subject was counterbalanced across

instructional conditions. Two formats were devised for each

story. One format, embellished, provided context words to

help cue the meaning of a targeted term. For the second

format, unembellished, the word cues were deleted.

Additionally, in Experiment 2, the authors further analyzed

subjects' overall comprehension of the texts. An ANOVA was

used to determine any differences between the treatment


Vocabulary definition recall was analyzed with a three-

factor mixed ANOVA, with the within-subjects being the number

of times that the word was viewed prior to testing (one or

two, depending on whether the word was encountered in the

text) and the between-subjects variables being instructional

condition and text type (embellished vs. unembellished).

McDaniel, Pressley, and Dunay (1987) were the only

experimenters who did not specifically describe their

statistical procedures for both of their experiments.

Results were only reported comparing the keyword mnemonic

condition to the contextual analysis condition in each of

these experiments. These two treatment conditions were

similar to previously described keyword and three-sentence

passage contextual conditions. The difference between

Experiment 1 and Experiment 2 was the inclusion of the term's

definition along with a three-sentence passage in the

contextual analysis condition in Experiment 2.

Subject Characteristics

Eight of the reviewed studies used subjects who were

college students enrolled in introductory psychology courses

(McDaniel & Pressley, 1984, 1989; McDaniel, Pressley, &

Dunay, 1987; McDaniel & Tillman, 1987; Pressley, Levin, &

Miller, 1982) from at least four different universities. Two

of these experimental descriptions did not state the name of

the university subjects were selected from (McDaniel,

Pressley, & Dunay, 1987; McDaniel & Tillman, 1987). Condus,

Marshall, and Miller (1986) conducted their investigation

with 12-year-old subjects who had been identified as Learning

Disabled (LD) and further classified as having "low" or

"high" receptive vocabularies based on the subject's

performance on the Peabody Picture Vocabulary Test-Revised.

Fourth grade subjects from two elementary schools

participated in the two Levin et al. (1982) studies.

Measurement Methods

All reviewed studies used posttest measures to ascertain

the effectiveness of the compared conditions. Two

experiments also included a pretest (Condus et al., 1986;

Pressley et al., 1982). Pressley et al. (1982) gave subjects

a short definition recall test on the sample items after

demonstration of the strategy. The subjects were reminded of

their strategy and were presented with the targeted terms.

After instruction, subjects were given a self-paced test in

which they had to write down the definition of each

vocabulary word or as much of the definition as could be


All studies reviewed included definition recall among

the dependent measures. Levin et al. (1982) presented

targeted terms individually to students on cards and asked

the students to state the meaning of the word. Students

received a prompt by the experimenter if a subject's response

did not provide a definition of the word. All responses and

the use of prompts were recorded by the experimenter.

McDaniel and Pressley (1984) evaluated subject recall

after the training session by administering a sheet of paper

with a listing of the words. Students were allowed as much

time as needed to write the definitions. Subjects in both

the context condition and combined keyword-context condition

were required to complete an additional assessment after the

recall test. These subjects were asked to reread the texts

in which each vocabulary word was embedded and write a short


McDaniel, Pressley, and Dunay (1987) examined delayed

recall of terms one week after instruction. Students

repeated treatment acquisition procedures until they could

provide the correct definitions for all targeted words. One

week later, students attempted to recall the definition for

each target term. Experimenters did not describe whether

subjects were required to write or state the definitions.

McDaniel and Tillman (1987) incorporated a free recall

test, as well as a cued recall test. The free recall measure

required students to write as many words and definitions as

possible after instruction. After completing this task,

students were given a sheet with a list of the words and

asked to write the definitions.

McDaniel and Pressley (1989) assessed comprehension, as

well as recall. In Experiment 1, subjects were given a true-

false test to measure comprehension. In the second

experiment, the experimenters asked subjects to recall as

much of the exact story as they could. Afterward, they

received a 15-item cued-recall test containing one question

for each of the targeted vocabulary terms.

Condus et al. (1986) were the only researchers that

assessed subjects' knowledge of vocabulary definitions using

a multiple-choice test format. These authors also included a

pretest two weeks prior to treatment so subjects could be

grouped according to either high-language or low-language

abilities. These experimenters repeated the multiple-choice

measure in four phases: immediately after instruction, at

the end of the week, 2 weeks later, and 10 weeks after


Experimental Procedures

Most of the reviewed studies were conducted within one

session (Levin et al., 1982; McDaniel & Pressley, 1984, 1989;

McDaniel & Tillman, 1987; Pressley, Levin, & Miller, 1982).

McDaniel, Pressley and Dunay (1987) had a one-day session,

but measured recall after 1 week. Condus et al. (1986)

conducted their study over a 17 week period. Five weeks were

devoted to intervention, 3 days of intervention per week.

Assessment measures occurred in weeks both prior to and after

intervention. The total instructional time for each session

ranged from 15 minutes (McDaniel, Pressley, & Dunay, 1983) to

25 minutes (McDaniel & Pressley, 1984; McDaniel & Tillman,

1987). The total instructional time for daily sessions was

not stated in two of the studies (Levin et al., 1982;

Pressley, Levin, & Miller, 1982), but experimenters stated

that time spent in treatment conditions was equal.

Materials. A variety of media was used to present

vocabulary information in the reviewed studies. Five of the

experiments used cards (e.g., 4" X 6", 5" X 8") that

contained critical information of the vocabulary term for

each treatment condition. McDaniel and Pressley (1984)

presented vocabulary information in eight-page booklets

containing eight entries per page except for the last page,

which had only five entries. A cassette tape was made to go

with each booklet for each condition. McDaniel and Pressley

(1989) utilized computers to present vocabulary information.

McDaniel, Pressley, and Dunay (1987) presented vocabulary

information on a cathode-ray tube.

Instructional medium. All but one study used trained

experimenters who were not the subjects' teachers to conduct

the study. However, Condus et al. (1986) trained seven

teachers of students with learning disabilities. Each

teacher was trained proficiently in experimental and control

conditions. Small group instruction was conducted in the

teachers' resource classroom. Teachers were blind as to the

experimental status of the condition they were teaching.

Testing was conducted by independent examiners. The testing

in all other reviewed studies was also conducted by


Experimental constants. All investigators strived to

keep instructional techniques and materials constant

throughout the investigation periods. Several authors kept

instruction paced by mechanical means (e.g., tape recorders,

computer presentation, timers) (McDaniel & Pressley, 1984,

1989; McDaniel & Tillman, 1987). However, instruction

relative to each condition differed even though time spent

learning the words remained constant. The keyword mnemonic

subjects received explicit instructions on that strategy.

Only one of the reviewed studies provided strategy training

for subjects in the contextual analysis condition (McDaniel &

Tillman, 1987). These subjects were typically presented with

information about the targeted term, not how to use the



Six studies (Condus et al., 1986; Levin et al., 1982;

McDaniel & Tillman, 1987; Pressley, Levin, & Miller, 1982)

found the keyword mnemonic strategy significantly better than

either the contextual analysis or control free-study

condition. McDaniel, Pressley, & Dunay (1987), however,

found that despite significant differences in the acquisition

of vocabulary terms, the different instructional methods

produced nearly equivalent retention scores one week after


McDaniel and Pressley (1989) also did not produce

significant results in Experiment 2, even though the context

group tended to recall less than either the keyword group or

control free-study group. McDaniel and Pressley (1984) found

significant differences between treatment conditions, but the

keyword mnemonic method was not significantly better than the

free-study control method. A supplementary analysis revealed

that students with low-ability in the keyword condition

performed significantly better than other treatment

conditions, but not the high-ability subjects. Condus et al.

(1986) also found results from the 10-week follow-up

assessment produced data that indicated that students with

low receptive vocabularies in the keyword mnemonic condition

learned more definitions on the average than students with

high and low receptive vocabularies assigned to all

alternative conditions.

Even though experimenters such as Condus et al. (1986)

demonstrated that the keyword mnemonic strategy was effective

when implemented in small group classroom settings, one must

view these findings cautiously. First, the vocabulary words

chosen for this and the other reviewed studies were words

that contained concrete keywords that were easily identified.

If words were specifically chosen to favor one of the

treatment conditions, then all conditions are not equal from

the start.

Researchers studying the effects of the keyword

mnemonic strategy compared to contextual approaches may find

less dramatic results when all conditions are controlled to

have equal strategy training. Although trends can be

detected, the results presented in the studies reviewed are

inconclusive. It is apparent that further research is needed

to help determine the most effective strategies for

vocabulary instruction.


The strategy instruction examined in these research

studies is of great importance to educators who are teaching

students experiencing little academic success. Mildly

disabled adolescents who have been trained with the keyword

mnemonic strategy to help them learn new vocabulary terms

have made impressive learning gains, when compared to other

strategy interventions. Mnemonic instructional methods have

also produced positive results in learning textbook content

with mildly disabled students. Although the keyword mnemonic

strategy appears to be more effective for the acquisition and

retention of information by mildly disabled adolescents, an

examination of the literature reveals a lack of research to

support the effectiveness of this strategy with at-risk


The results of existing studies comparing keyword

mnemonic strategies to contextual strategies and free-study

control conditions are inconsistent. Therefore, further


exploration is evident. The purpose of the present study was

to extend the research comparing keyword mnemonic strategies

to other contextual and control conditions and to determine

whether at-risk adolescents could benefit from the keyword

mnemonic strategy.


The research methods in this investigation were designed

to assist the investigator in comparing subjects' recall of

the definitions of geography terms among three treatment

conditions: keyword mnemonics, contextual analysis, and rote

free study. Recall effects on initial acquisition, short-

term retention, and long-term retention of definitions in

each condition were measured. The research methodology

including descriptions of subjects, null hypotheses,

instrumentation materials, procedures, and experimental

design and analysis is presented in this chapter.


The subjects for this study were seventh grade students

who have been identified as students at risk for dropping out

of school. All students were placed in alternative education

programs according to the criteria required by the Orange

County school system. These criteria include students having

two or more of the following characteristics: failure of two

or more subjects, failure of one or more grades and currently

unsuccessful, frequent unexcused absences, frequent

discipline problems, evidence of low self-esteem, poor social

skills, loneliness, or stressful loss (Student Alternative

Programs Department of Special Services, 1992). Ninety-eight

7th grade students participated in this investigation. All

subjects received treatment instruction from their

alternative education teacher in their geography classroom

located in Orange County, Florida. Each teacher was randomly

assigned to one of the three treatment conditions.

Additional subject characteristics are presented in

Table 3-1 adapted from the recommendations of Smith, Deshler,

Hallahan, Lovitt, Robinson, Voress, and Ysseldyke (1984).

These authors suggested including gender, age, ethnicity,

socioeconomic status (SES), intelligence quotient (IQ)

scores, and achievement scores. Since many students who have

been identified as at risk for dropping out of school have

never been recommended for an exceptional student evaluation,

IQ scores were unavailable. Since IQ scores were

unavailable, standard scores from the Comprehensive Test of

Basic Skills (CTBS) are included in describing subject

characteristics. Additional information on previous

retentions of subjects are also included. All subjects

participated in a screening procedure to determine both

subject eligibility for participation in this study and to

target unfamiliar geography terms from the state adopted

seventh grade geography texts used in the Orange County

School System. The purpose of this screening was to

ascertain that the students were unfamiliar with the

definitions of the geography terms identified for the study.

The screening consisted of 90 geography terms. The students

were required to write the definition of each term. Critical

components of the definitions described in the texts'

glossary or written text were used to determine correct and

incorrect responses. An overall score of less than 40% of

terms defined correctly determined subject eligibility for

participation. Results of the screening are reported.

Table 3-1
Description of Subjects

Mnemonic Keyword Context Clue Rote Memory

Male 13
Female 15
Total 28

Mean 12.5
Range 12-14

Anglo 04
African A.24
Hispanic 00
Other 00

CTBS Rd. Stan. Score:
Mean 637.61
Range 516-755
SD 54.17

Num. of Retentions:
Mean 0.14
Range 0-1

Male 21
Female 13
Total 34

Mean 12.8
Range 11-15

Anglo 12
African A. 19
Hispanic 03
Other 00

CTBS Rd. Stan. Score:
Mean 643.85
Range 476-789
SD 71.36

Num. of Retentions:
Mean 0.76
Range 0-3


Mean 12.2
Range 11-13

Anglo 16
African A.16
Hispanic 02
Other 02

CTBS Rd. Stan. Score
Mean 645.59
Range 476-723
SD 60.99

Num. of Retentions:
Mean 0.19
Range 0-1


The dependent variables examined in this study were the

recall of geography terms' definitions for subjects taught in

one of three ways. The three interventions were teacher-

provided keyword mnemonics, teacher-provided contextual

analysis, and rote free study. The subjects' recall of

definitions was measured for acquisition during the

instructional intervention, short-term retention at the end

of the week after instruction, and long-term retention two

weeks after instruction had ended. Recall was measured by

using a written (production from memory) format. In

addition, the subjects' rate of acquisition of definitions

was analyzed during each instructional week. The following

null hypotheses addressed the effects of the three treatment

conditions regarding student performance related to

acquisition and retention.

HI: There will be no statistically significant

differences among the three experimental groups on

acquisition of geography terms' definitions recalled on the

day instruction has occurred.

H2: There will be no statistically significant

differences among the three experimental groups on short-term

retention of geography terms' definitions recalled at the end

of the week following instruction.

H3: There will be no statistically significant

differences among the three experimental groups on long-term

retention of geography terms' definitions recalled two weeks

following instruction.

The .05 level of significance was used as the basis for

rejection of a null hypothesis.


Four teacher-made research instruments were used in this

study. The first instrument, the Written Geography Term

Definition Screening Device, was used to determine if

students could write the definitions of less than 40% of the

geography terms. The second instrument, the Daily Measure of

Geography Terms' Definitions Device, was used during each

instructional session to help compare the rate of acquisition

for subjects in each condition. The third instrument, the

Teacher Survey to Evaluate Content Validity, was sent to 21

Orange County middle schools. Each school had approximately

six social studies teachers who received the survey in order

to obtain an external evaluation of content validity as

recommended by Ary, Jacobs, and Razavieh (1985). The

selected number of words used for instructional conditions

was based upon the survey results and validation of

professionals in the educational system. The fourth

instrument, the Geography Terms' Definition Device, included

all targeted terms and was administered weekly during this

investigation. Varied forms of the Geography Terms'

Definition Device were used with the words randomly ordered

on each form. These criterion measures were used to

determine subjects' performance regarding acquisition, short-

term, and long-term retention.

Ninety words were used during the screening (see

Appendix A), and a percentage of these words was randomly

selected for use in this study dependent upon the results of

the teacher survey and the screening device. The words that

were not selected for the study were used as example words

for the first day of instruction during each of the four

instructional strategy training weeks (see Appendix D).

Written Geography Term Definition Screening Device

The Written Geography Term Definition Screening Device

(see Appendix B) required the subjects to write the

definitions of as many of 90 seventh grade geography terms as

they could. Students had as much time as they needed to

complete this screening device. Students were only given

credit for complete definitions. Partial credit was not

awarded. Subjects with scores of less than 40% were eligible

for participation in this investigation.

Teacher Survey to Evaluate Content Validity

The Teacher Survey to Evaluate Content Validity (see

Appendix C) was used to obtain an external evaluation of

content validity. The survey was sent to all social studies

teachers in each Orange County middle school. Each assistant

principal of instruction or principal was asked to distribute

the survey to the social studies teachers in their school.

Approximately 120 surveys were sent out to be given to

University of Florida Home Page
© 2004 - 2010 University of Florida George A. Smathers Libraries.
All rights reserved.

Acceptable Use, Copyright, and Disclaimer Statement
Last updated October 10, 2010 - Version 2.9.7 - mvs