A normative study of the acquisition of consonant sounds in Portuguese

MISSING IMAGE

Material Information

Title:
A normative study of the acquisition of consonant sounds in Portuguese
Physical Description:
Book
Creator:
Santini, Celia R. Salviano, 1960-
Publication Date:

Record Information

Rights Management:
All applicable rights reserved by the source institution and holding location.
Resource Identifier:
aleph - 22035603
oclc - 34461591
System ID:
AA00020417:00001

Full Text










A NORMATIVE STUDY OF THE ACQUISITION
OF CONSONANT SOUNDS IN PORTUGUESE













BY

CELIA R. SALVIANO SANTINI


A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL
OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT
OF THE REQUIREMENTS FOR THE DEGREE OF
DOCTOR OF PHILOSOPHY

UNIVERSITY OF FLORIDA

1995


UNIVERSITY OF FLORIDA LIBRARIES






























Copyright 1995

by

Celia R. Salviano Santini













ACKNOWLEDGEMENTS

First and above all, I would like to thank my loving God who provided for

all of the following persons who made this dissertation possible. Without His help

and guidance, this work would not have been possible.

The people who contributed to the completion of this work are many. This

has been an intense experience and the involvement of so many family and

friends was necessary to keep me going. There is no way that I can express my

feelings of gratitude to everyone, but some credit, at least, should be given.

First, I would like to express my deepest and most sincere gratitude to Dr.

Alice Tanner Dyson, my advisor, not only for her invaluable assistance during this

study, but for her time and support throughout my doctoral program. Dr. Dyson's

patient guidance through all the stages of this study helped me learn that

everything is possible if we take one problem at a time. Her constant uplifting

humor, her trust and confidence, made me feel special and capable. Indeed, the

knowledge I obtained from Dr. Dyson about research as well as computer use is

beyond measure. We can say now that we did beat the machine.

I also would like to express my warmest thanks to the members of my

committee, Dr. Linda Lombardino, Dr. Charles Perrone, Dr. Howard Rothman,

and Dr. Christine Sapienza. Their support and input were invaluable.








I owe special thanks to my husband for his love and support that kept me

going; without his help and encouragement none of this would have been possible.

I would also like to thank my son Lucas for making me so happy, every day. I

thank my mom and my dad, for the constant love and support, and especially my

mom who took care of my baby while I finished writing this paper. I would like to

thank my entire family-my brothers, my sisters-in-law, and my in-laws-for their

encouraging phone calls, their cheering, their prayers, and their understanding of

my absence.

I appreciate the financial support provided by my country through CNPq

(Conselho Nacional de Desenvolvimento Cientifico e Tecnologico-Brazil). Their

support during those last four years was sharp, valuable, and friendly.

The Department of Communication Processes and Disorders has offered

me friendship, support, and encouragement at every step of my studies. I offer

special thanks to all of these terrific people. I would like to thank my dear friend

Jennifer Dutka for her enormous help teaching me how to be a student, a

researcher, and a mother. I would also like to thank my friend Mousa Al

Amayreh for showing me the way and inspiring this study. My thanks go to Jizela,

who helped me enter the data, and to Monica who was a great tutor. The

Graduate School editorial staff must also be thanked for their sharp eyes and

valuable assistance in editing.








I greatly appreciate the support and advice of Drs. Yavas and Lamprecht.

My calls and questions to them were always handled with consideration and

information.

Last, but not least, my warmest thanks and appreciation go to the 192

children who participated in the study and especially to Anilu and Tereza, the two

speech therapist who spent hours with me helping to test the children.













TABLE OF CONTENTS


ACKNOWLEDGMENTS ........................

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

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

KEY TO SYMBOLS ...........................

ABSTRACT ..................................


CHAPTERS

1

2


INTRODUCTION ..........................

REVIEW OF THE LITERATURE ..............

Methodology of Studies of Sound Acquisition .......
Normative Research ....................
Purposes of Normative Studies .............
Subject Selection and Sample Size ..........
Word Selection and Elicitation Methods ......
Methods of Analysis and Presentation of Norms
Phonological Characteristics of Portuguese .........
V ow els ..............................
Consonants ..........................
Syllables Type and Composition ............
Studies of Acquisition of Portuguese .............
Statement of the Problem .....................
The Need for This Research ..............
Purpose of Study .......................


Siii

x


xiv

Sxv









3 METHODOLOGY ................................


Subjects ...................
The Assessment Instrument .....
Stim uli ...............
Procedure ..................
Duration of the Exam ....
Test Environment .......
Examiners ............
Tape Recording Equipment
Transcription ..........
Data Entry and Tabulation
Validity of the Test ......
Presentation of the Data ..


o..........


...,.......
...........


and Procedure
...........
...........
...... .....
...........


4 R ESU LTS ......................................

Question 1. What percentage of each consonant was produced
correctly by children at each age level? .............
SIWI, Syllable-Initial, Word-Initial ................
SIWW, Syllable-Initial, Within-Word ..............
Comparison of Percentages of Correct Production by
Position ..............................
C lusters ...................................
Comparison of Percentages of Correct Production of
Boys and Girls .........................
Question 2. What is the age of "customary production" for each


sound? .......
Question 3. What is the
Question 4. What is the
/b/-/p/ .......
/d/-It/ ........
Ig/-/k/ ........
fm/-In/-I /a ....
/v/-/f/ ........
/z/-/s/ ........

/tf//d-13/ ......
/tt$I-IdsI!......

L/I7-//-/r-R/ . .
Reliability .........


.......................
age of "acquisition" for each sound?
age of "mastery" for each sound? ..
. . . . . . . . . . . . . .
.............o..........
.,.................o......
.........o...............oo
...........................
.........................
.........................

.....o......................
.,......................


57
. 59
. 59

. 78
105

107


. 107
S107
. 107
. 109
. 109
111
111
. 111
.. 111
. 111
* 111
.. 112
. 112








5 DISCUSSION ...................

Comparison with Other Studies .......
Percentages of Correct Production
Customary and Mastery Ages ....
Strengths, Weaknesses, and Limitations .
Suggestions for Further Research ......
Conclusions .....................


APPENDICES

A STIMULUS PICTURES . .

B TRANSCRIPTION SHEET

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

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


113

113
113
116
116
120
120


122

127

132

136













LIST OF TABLES


page


Table


2-1 Summary of purposes of major studies reviewed ............ 6

2-2 Articulatory classification of Porguguese vowels ........... 23

2-3 Examples of Portuguese diphthongs and triphthongs ........ 25

2-4 Nasals vowels in Portuguese ......................... 26

2-5 Syllabic types of Portuguese .......................... 33

2-6 Distribution of Portuguese consonants .................. 34

2-7 Examples of Portuguese clusters ...................... 35

2-8 Consonant sequences in Portuguese .................... 37

3-1 Distribution of subjects' age and gender in each group ...... 46

3-2 List of words targeted by each picture .................. 48

3-3 Distribution of consonants by position .................. 49

4-1 Percentage of each consonant produced correctly in the SIWI
position in each group .............................. 60

4-2 Percentage of each consonant produced correctly in the
SIWW position in each group ........................ 69

4-3 Distribution of consonants by position .................. 79















LIST OF FIGURES


page


2-1 Places and manners of articulation of Portuguese consonants

...................................4-1 Overall percentage of consonants produced correctly by each
4-1 Overall percentage of consonants produced correctly by each


group .


4-2 Percentages of accuracy
position for Group 2:0.

4-3 Percentages of accuracy
position for Group 2:6 .

4-4 Percentages of accuracy
position for Group 3:0.

4-5 Percentages of accuracy
position for Group 3:6 .

4-6 Percentages of accuracy
position for Group 4:0.

4-7 Percentages of accuracy
position for Group 4:6 .

4-8 Percentages of accuracy
position for Group 5:0 .

4-9 Percentages of accuracy
position for Group 6:0 .

4-10 Percentages of accuracy
position for Group 2:0.

4-11 Percentages of accuracy
position for Group 2:6 .


of each consonant in the SIWI

of each consonant in the SIWI. .


of each consonant in the SIWI

of each consonant in the SIWI... .....


of each consonant in the SIWI

of each consonant in the SIWI........


of each consonant in the SIWI

of each consonant in the SIWI..


of each consonant in the SIWW

of each consonant in the SIWW...
of each consonant in the SIWI
. . . . . . . .. .. . . . . . ..

of each consonant in the SIWI


of each consonant in the SIWI
. . . . . . . . . . . . . ..

of each consonant in the SIWW


of each consonant in the SIWW
. . . . . . . . . . . ... o .. .


Figure


. . . . . . . . . . . . . . . . . . 58








4-12 Percentages of accuracy of each consonant in the SIWW
position for Group 3:0 .............................. 72

4-13 Percentages of accuracy of each consonant in the SIWW
position for Group 3:6 .............................. 73

4-14 Percentages of accuracy of each consonant in the SIWW
position for Group 4:0 .............................. 74

4-15 Percentages of accuracy of each consonant in the SIWW
position for Group 4:6 .............................. 75

4-16 Percentages of accuracy of each consonant in the SIWW
position for Group 5:0 .............................. 76

4-17 Percentages of accuracy of each consonant in the SIWW
position for Group 6:0 .............................. 77

4-18 Comparison of accuracy of stops in SIWI and SIWW posi-
tions for Group 2:0 ................................ 81

4-19 Comparison of accuracy of nasals and liquids in SIWI and
SIWW positions for Group 2:0 ........................ 82

4-20 Comparison of accuracy of fricatives and affricates in SIWI
and SIWW positions for Group 2:0 .................... 83

4-21 Comparison of accuracy of stops in SIWI and SIWW posi-
tions for Group 2:6 ................................ 84

4-22 Comparison of accuracy of nasals and liquids in SIWI and
SIWW positions for Group 2:6 ........................ 85

4-23 Comparison of accuracy of fricatives and affricates in SIWI
and SIWW positions for Group 2:6 .................... 86

4-24 Comparison of accuracy of stops in SIWI and SIWW posi-
tions for Group 3:0 ................................ 87

4-25 Comparison of accuracy of nasals and liquids in SIWI and
SIWW positions for Group 3:0 ........................ 88

4-26 Comparison of accuracy of fricatives and affricates in SIWI
and SIWW positions for Group 3:0 .................... 89








4-27 Comparison of accuracy of stops in SIWI and SIWW posi-
tions for Group 3:6 ................................ 90

4-28 Comparison of accuracy of nasals and liquids in SIWI and
SIWW positions for Group 3:6 ........................ 91

4-29 Comparison of accuracy of fricatives and affricates in SIWI
and SIWW positions for Group 3:6 .................... 92

4-30 Comparison of accuracy of stops in SIWI and SIWW posi-
tions for Group 4:0 ................................ 93

4-31 Comparison of accuracy of nasals and liquids in SIWI and
SIWW positions for Group 4:0 ........................ 94

4-32 Comparison of accuracy of fricatives and affricates in SIWI
and SIWW positions for Group 4:0 .................... 95

4-33 Comparison of accuracy of stops in SIWI and SIWW posi-
tions for Group 4:6 ................................ 96

4-34 Comparison of accuracy of nasals and liquids in SIWI and
SIWW positions for Group 4:6 ........................ 97

4-35 Comparison of accuracy of fricatives and affricates in SIWI
and SIWW positions for Group 4:6 .................... 98

4-36 Comparison of accuracy of stops in SIWI and SIWW posi-
tions for Group 5:0 ................................ 99

4-37 Comparison of accuracy of nasals and liquids in SIWI and
SIWW positions for Group 5:0 ....................... 100

4-38 Comparison of accuracy of fricatives and affricates in SIWI
and SIWW positions for Group 5:0 ................... 101

4-39 Comparison of accuracy of stops in SIWI and SIWW posi-
tions for Group 6:0 ............................... 102

4-40 Comparison of accuracy of nasals and liquids in SIWI and
SIWW positions for Group 6:0 ....................... 103

4-41 Comparison of accuracy of fricatives and affricates in SIWI
and SIWW positions for Group 6:0 ................... 104








4-42 Percentages of correct production of two types of clusters
and the /nd/ sequence by each group................... 106

4-43 Comparison of accuracy of consonants produced by boys and
girls in each group ............................... 108

4-44 Customary, mastery, and acquisition ages of each consonant 110

5-1 Comparison of customary and mastery ages of Portuguese,
Arabic, and English stops and nasals .................. 117

5-2 Comparison of customary and mastery ages of Portuguese,
Arabic, and English fricatives, affricates, and nasals ....... 118













KEY TO SYMBOLS

Throughout this paper, several symbols have been substituted for Interna-

tional Phonetic Alphabet (IPA) symbols unavailable either in the text or figure

fonts.



L! = /XA I, alveo-palatal, lateral liquid, voiced consonant

/f/ = /p/, alveo-palatal, nasal, voiced consonant

/R/ = /x/, lingua-velar, lateral liquid, voiced consonant













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

A NORMATIVE STUDY OF THE ACQUISITION
OF CONSONANT SOUNDS IN PORTUGUESE

By

Celia R. Salviano Santini

December, 1995

Chair: Alice T. Dyson, Ph.D.
Major Department: Communication Processes and Disorders

Phonological acquisition is a topic that has fascinated researchers trying to

solve the puzzle of how children learn to talk. A review of the literature provides

a large body of information about the acquisition of English phonology. The study

of Portuguese phonology is a rapidly growing discipline with several studies dating

from 1992 to the present. Most of these studies have addressed the need to build

a more theoretical basis to support clinical work with individuals with disorders.

The purpose of this study was to collect normative data on the acquisition

of consonant sounds in Portuguese as spoken in Brazil. The data were used to

answer four questions: (1) What percentage of children at each age level pro-

duced each consonant correctly? (2) What is the age of "customary production"

for each sound? (3) What is the age of "acquisition" for each sound? (4) What is

the age of "mastery" for each sound?








An existing articulation test of Portuguese was used to collect samples

from 192 normally developing children between the ages of 2:0 and 6:11. The

children represented eight age groups with 12 boys and 12 girls in each group.

The children's responses were tape recorded and transcribed by two listeners for

later analysis.

The consonants were considered separately in all positions in which they

occurred. The percentages of accuracy of each consonant sound were analyzed

and indicated a rapid development between 2:0 and 2:6 with continued but slower

development in older groups. No significant differences were found between the

performances of girls and boys. The ages of acquisition were compared to English

and Arabic. Customary production tended to come later in Portuguese but

mastery occurred earlier than in the other two languages. The voiceless conso-

nants were produced more accurately than their voiced cognates. In the younger

groups the word initial consonants were less accurate than the same consonants

within words. Unlike the findings of other languages, the liquids were learned

before the fricatives. Other patterns of acquisition found were similar to those

reported in other languages.














CHAPTER 1
INTRODUCTION


The acquisition of the phonology of languages has been studied for several

purposes. Much of this research has been undertaken to add to the theoretical

point of view of the researcher. However, such studies can also have practical

purposes. Precise information on the ages of acquisition of sounds and sound

sequences is needed for formulating articulation test instruments, for making

diagnostic decisions about the status of the speech of individual children, and for

planning remediation when problems are found. Since 1960, assessment batteries

for articulation have included sound inventories-articulation tests-usually

elicited by picture naming. Although such tests are useful to the speech-language

pathologist, they have been widely criticized in the literature (Ingram, 1989; Irwin

& Wong, 1983; Olmnsted, 1971; Stoel-Gammon & Dunn, 1985).

Frequent criticisms have focused on the lack of validity of testing speech by

naming pictures. Problems with picture naming include the choice of words, the

type of words, and the testing of only one sound in the word. The selection,

screening, and numbers of subjects tested have also been criticized by those who

would like to generalize data to other groups. Based on some of these criticisms,

recent researchers have made attempts to remedy the problems by conducting







2

further normative studies and by developing newer forms of the tests. Some of the

changes introduced by these researchers were minor, such as including more

items, using vocabulary more appropriate to the subjects tested, controlling word

structure and familiarity, targeting more than one sound per word, and using

more realistic pictures (Al Amayreh, 1994; Ingram, Christensen, Veach, &

Webster, 1980; Smit, Hand, Freilinger, Bernthal, & Bird, 1990; Yavas, 1988). In

addition, major changes have been made. These included a shift toward more

phonemic detail in scoring instead of correct/incorrect, a shift from sound by

sound analysis to pattern analysis, the inclusion of normative data that allow

clinicians to compare a child's performance at a particular age to norms from the

instrument (Khan & Lewis, 1986; Lamprecht, 1993; Prather, Hedrick, & Kern,

1975; Preisser, Hodson, & Paden, 1988; Sander, 1972; Smit et al., (1990).

Most of the commonly reported studies of phonological acquisition have

been conducted in English. However, it is clear that the phonologies of languages

different from English may develop in different ways and at different rates

(Wertzner, 1992). The language of interest in this study was Portuguese.

Portuguese is a Romance language spoken in Portugal, Brazil, Mozambique,

Angola, and parts of Africa and Asia. In Brazil alone, approximately 160 million

people speak this language. The practice of speech-language pathology is well

developed in Brazil, but the number of normative studies of the sound system are

few. Clearly, the development of Portuguese phonology is of interest to a great

many people.








3

In the following chapter some of the most important methodological issues

in the collection and analysis of phonological acquisition data will be discussed,

especially as these have been addressed by researchers in the United States. An

overview of the phonological system of Portuguese will be presented and

compared briefly to English. Finally, the few available reports on phonological

acquisition of Portuguese as spoken in Brazil will be presented.

The purpose of this study was to collect further normative data on the

acquisition of Portuguese. In this study an attempt was made to remedy some of

the problems found by researchers in other normative studies.














CHAPTER 2
REVIEW OF THE LITERATURE

Methodology of Studies of Sound Acquisition

The empirical problems in normative research are many. Some of those

problems have been addressed by researchers in terms of how they apply to the

study of sound acquisition (Ingram, Christensen, Veach, & Webster, 1980; Irwin

& Wong, 1983; Morrison & Shriberg, 1992; Smit, 1986). In this review the focus

will be on subject selection and sample size, word selection and elicitation

methods, and methods of analysis and presentation of norms. Those aspects will

be discussed in the context of several studies of English and the limited studies

that have been reported on Portuguese.

Normative Research

Normative research was defined by Hegde (1987) as the type of research in

which the distribution of selected dependent variables across age groups is

observed and recorded. A study of the acquisition of sounds could report, for

example, the number of phonemes correctly produced (dependent variable) by 2-,

3-, and 4-year-old children from a particular population. According to Hegde, the

major purpose of normative research is to arrive at norms that are the averaged

performance levels of presumably typical reference groups.









For clinical disciplines, such as speech-language pathology, developmental

norms are extremely important because they tell us how children's speech

behaviors change as they grow older. This information helps in making clinical

judgments as to whether a given child's behavior is within the expected age range

or not. However, the literature review shows that there has been a variety of

purposes for conducting studies of phonological acquisition, and the studies have

differed somewhat to accommodate their purposes. Because of such differences in

methodology, they are not always comparable to each other and are not of equal

value for the speech-language pathologist.

Purposes of Normative Studies

A summary of some purposes of selected studies can be found on Table 2-

1. It should be noted that this table includes studies of both English, summarized

by Al Amayreh (1994), and of Portuguese. In this paper, emphasis will be given to

studies of the type included under Purposes 1, 2, 4, 6. Smit (1986) reviewed seven

of the major normative studies of English (Arlt & Goodman, 1976; Irwin &

Wong, 1983; Olmsted, 1971; Poole, 1934; Prather, Hedrick, & Kern, 1975;

Templin, 1957; Wellmnan, Case, Mengert, & Bradbury, 1931), looking especially at

their methodologies as these might affect the ages of acquisition reported. She

concluded that "the major differences among the elicited speech investigations

occur in the area of subject selection, method of obtaining the speech sample, and

analysis procedures" (p. 177). Further discussion regarding the differences in

methodology between the studies will be addressed in sections below.










Table 2-1. Major purpose of the studies reviewed.


Major Purpose of Study Conducted by

1. To provide normative data on speech sound acquisition for a Al Amayreh, 1994; Smit et al., 1990; Preisser et al., 1988;
specific group of children (e.g., particular state or region); those Stoel-Gammon, 1985, 1987; Yavas, 1988; Wellman et al.,
in a particular age group; or a particular language. 1931; Poole, 1934.; Silverio et al., 1993; Hernandorena,
_______________________ 1993; Wertzner, 1992

2. To provide normative data on speech sound acquisition as a part Templin, 1957
of a larger study of other aspects of language development.

3. To examine a particular aspect of phonology at different times. Lamprecht, 1993; Vihman & Greenlee, 1987; Dyson, 1988

4. To replicate previous normative studies to take into account Arlt & Goodban, 1976; Prather, et al., 1975
changes in age of acquisition due to current rearing practices,
technology, and educational procedures.

5. To study the effect of methodological factors on age of Ingram et al., 1980
acquisition

6. To provide normative data on speech sound and distinctive Irwin & Wong, 1983; Prather et al., 1975; Olmsted, 1971
feature acquisition using a different elicitation procedure than
that typically used

7. To examine errors made by children learning their language Templin, 1957

8. To identify the phonological processes that account for various Khan & Lewis, 1986; Preisser, et al., 1988; Mota, 1993
types of errors in a group of normal children

9. To develop normative data for a test, either the development of Khan & Lewis, 1986; Templin, 1957
a test or follow-up of an existing one









Subject Selection and Sample Size

Normative research has typically used the stimulus and response sampling

procedure to establish the statistically averaged response patterns across age

groups. Theoretically, norms are established on randomly selected subjects who

are representative of the population. Several researchers have emphasized that

certain variables must be controlled in order for the results of normative studies

to have value (Ingram et al., 1980; Smit et al., 1990). Specifically, the sample must

be representative if the conclusions are to be extended to all children in the

population.

Subject characteristics

In their selection of subjects, most researchers make an effort to ensure

that the sample reflects the socioeconomic distribution of the population as a

whole. Also, audiometric screening and parental report are typically used to

exclude children with hearing losses or delayed language development. However,

different criteria for subject selection have been used depending upon each

researcher's belief about factors important to his or her purpose. For example,

Smit et al. (1990) were only concerned with studying children from a particular

region of the United States-Kansas and Nebraska; Stoel-Gammon (1985) was

looking at a particular age group, children under the age of 2.

Khan and Lewis (1986) pointed out the need for subjects in different age

groups to have similar backgrounds (gender, ethnic, and geographical). Their

subjects were also screened to exclude physical abnormalities and to guarantee









normal function of the speech mechanism. They did not express a particular

concern about screening or controlling the socioeconomic status. Templin (1957),

however, was much more concerned with socioeconomic level. She weighted her

sample heavily on the lower end of the socioeconomic status scale (70% lower,

30% upper, based on father's occupation) and included only urban children, all

monolingual. Although all of her subjects were from Iowa, she stated that she

tried to represent the general population of the United States in her subject

group in the distribution of socioeconomic status. Templin (1957) did not screen

hearing or language development even though there are indications that even a

mild problem in either area might affect the age of sound learning (Ingram,

1989).

Prather et al. (1975) also attempted to mirror the socioeconomic makeup

of the general population of the United States in their subject sample. They

selected one-third of each age group from three different classes based on

education and occupation. Only Caucasian monolingual children were included

and those with hearing loss or language disorders were excluded. Arlt and

Goodban (1976), on the other hand, claimed that their subjects represented an

average socioeconomic population but did not describe how this was determined.

The majority of their subjects were white with some other races in each subgroup.

They included only monolingual subjects and screened for normal I.Q., emotional

problems, and/or hearing problems.










Both of the two early studies, Wellman et al. (1931) and Poole (1934)

observed children from university laboratory schools, making their populations

very selective. All subjects were from upper socioeconomic status with fathers who

were graduate students. A very narrow sample of children such as this, all drawn

from a specific group, is probably not a representative sample from which

generalizations to larger populations could be made. No other factors were

reported to have been screened in these two studies to determine subject

exclusion.

Olmsted (1971) was the only researcher who specifically included some

children with parents who spoke English as a second language or non-American

dialects. However, he observed primarily children of professors and graduate

students. No control for hearing status or language development was reported.

A brief review of subject selection procedures in some other studies of

English found that Irwin and Wong (1983) selected subjects with middle social

economic status, all Caucasian. They screened for hearing, previous language

therapy, and dentofacial abnormalities. Stoel-Gammon (1985, 1987) selected

subjects who were monolingual and screened for hearing, cognitive development,

and motor development. Preisser, Hodson, and Paden (1988) included subjects

from middle and lower-middle socioeconomic levels, all monolingual. They

screened for hearing, "general function," language development, and voice. In the

most recent study, Smit et al. (1990) stated the population density of their

subjects' residences and the parental education of their subjects, all of whom were










monolingual and spoke a "standard Midwestern dialect." The subjects were

screened for hearing, motor speech problems, and oral-facial abnormalities.

A study conducted by Silverio, Parlato, Mourao, Altmann, and Chiari

(1994) looked at the occurrence of Portuguese phonemes produced by preschool

children of public versus private schools. The researchers controlled the subject

selection to ensure the same number of males and females in each group. A

screening was conducted to control for normal language development and normal

speech mechanism. It should be noted that in Brazil the majority of children who

enroll in public schools are from the lower social economic layer of the

population. Other studies looking at the phonological acquisition of Portuguese

(Hemrnandorena, 1993; Lamprecht, 1993; Yavas, Hernandorena, & Lamprecht,

1991) controlled the subject selection in terms of gender, age, and normal

development. One study of Portuguese phonology (Mota, 1993) looked at a

specific group. She evaluated the validity of a phonological therapy model

(Hodson & Paden, 1986); therefore, only children with developmental

phonological disorders were selected.

Age range

Normative research is typically concerned with the distribution of behaviors

across age groups. The age groups are usually selected according to the purposes

of the study (e.g., to include very young children to extend existing age norms).

The age ranges in the studies reviewed have generally fallen between 2 and 8

years. Also, studies have varied in the earliest age tested and in the intervals








11

between and within age groups. Some studies started at a very young age, from 12

to 18 months or even younger (e.g., Ingram et al., 1980; Irwin & Wong, 1983;

Preisser et al., 1988; Stoel-Gammon, 1985, 1987; Vihman & Greenlee, 1987;

Olmsted, 1971). Another group of studies started at 2 years or at 2:6 (Khan and

Lewis, 1986; Poole, 1934; Prather et al., 1975; Wellman et al., 1931). Arlt and

Goodban (1976), Smit et al. (1990), and Templin (1957) started at 3 years of age.

The age range of subjects is important when considering the usefulness of

the data. For example, inclusion of young children allows more opportunities to

see if sounds were acquired earlier than had been previously reported.

Unfortunately, in studies of very young children, it was often reported that a large

number of young children did not respond to all stimuli. In addition, studies that

included only very young children usually reported a smaller sample size and were

typically undertaken for special purposes. For example, Preisser et al. (1988)

studied a particular age group to identify phonological processes, and Stoel-

Gammon (1985, 1987) studied a particular age group to extend previous norms

into younger age groups. ArIt and Goodban (1976), who started at age 2:6, and

Poole (1934), who started at age 3, did not report on the completeness of their

data (Smit, 1986). The frequently used norms collected by Templin (1957) started

at age 3. All sounds that had already been acquired by her subjects before that

age were simply reported as being acquired at 3 years. Such a presentation of

data can be misleading for clinical diagnosis.










In cross-sectional research such as most of the phonological acquisition

studies, a decision must be made about the age interval within groups. Most of

the studies reviewed used 6-month or 1-year intervals, often with smaller intervals

for younger children, assuming that more rapid changes would occur in these

groups. For example, Templin (1957) allowed 6-month groups up to age 5 and 1-

year groups between 6 and 8 years. Narrower age groups were used by Prather et

al. (1975), four age groups per year (e.g., 24, 28, 32, 36 months) with each subject

falling within one month before or after the mid-point (e.g, 23-25, 31-33, etc.)

The studies of Portuguese phonology reviewed presented age ranges

similar to the ones used by studies of English. Hernandorena (1993) studied

children as young as 2 and up to 4:3. The children were divided into 14 groups

with a 2-month age range in each group. Yavas et al. (1991) studied children from

2:4 to 4:4 years of age. Silverio et al. (1994) studied preschool children from 2:6

to 5:6 years of age divided into three groups with a window of one year each.

Lamprecht (1993) studied the phonological acquisition of Portuguese speaking

children between age 2:9 to 5:5.

Sample size

The sample size is a very important consideration in any research study. In

normative research, decisions about the number of subjects to study represent a

real challenge. Hegde (1987) reminds us that any population is heterogeneous.

When a large number of persons are studied, there is much variability. Therefore,

a good sample must be heterogeneous, but the more heterogeneous the sample,










the more variable the performance. Unfortunately, the more variable the

performance, the less meaningful are averaged data (the norms). "In other words,

even when an investigator achieves a representative random sample, the resulting

norms will be highly variable, a contradiction of terms" (Hedge, 1994, p. 85).

The great variability in sample sizes included in the normative studies

reviewed can, in most cases, be attributed to differences in the purposes of the

studies. Templin (1957), Khan and Lewis (1986), and Smit et al. (1990) used very

large samples (480, 852, and 997, respectively) to provide normative data on

speech sound acquisition. These samples have been divided almost evenly

between boys and girls.

Two studies that were intended primarily to replicate previous normative

studies used smaller samples for this purpose; Arlt and Goodman (1976) included

240 subjects, and Prather et al. (1975) included 147 subjects. Both Olmsted (1971)

and Irwin and Wong (1983) included 100 subjects in their studies, which

attempted to show the use of different elicitation procedures to provide data on

speech sound and distinctive feature acquisition. Al Amayreh (1994) included 180

subjects, starting with children as young as 2:4 in his normative study of Arabic.

Preisser et al. (1988), Stoel-Gammon (1985, 1987), and Vihman and Greenlee

(1987), observed much smaller samples of children to study development in very

young children with more detail than is possible in the large sample studies.

Mota (1993) studied a small sample of three children because she was

looking at the effects of a detailed phonological process therapy. Lamprecht








14

(1993) also reported a fairly small sample of 12 children in her longitudinal study

of acquisition of Portuguese phonology. Wertzner (1992) looked at 56 children

and Yavas (1988) included 72 children in their studies of the phonological

development of Portuguese speakers. Hernandorena (1993) studied a larger

sample of 134 Portuguese speaking children when looking at the stages of

phonological acquisition.

Word Selection and Elicitation Methods

Elicitation methods

The method used for eliciting the speech sample from the subjects in

normative research has varied somewhat and is usually determined by the purpose

of the study. Elicitation methods can be classified under two major categories:

single words and conversational samples. Single words can be collected either in

spontaneously evoked samples (picture naming, reading, and sentence

completion) or in imitative samples (direct imitation or with some intervening

time delay or words between the model and the imitation).

To obtain single-word samples, children typically are asked to name a

picture representing a target word. If the child does not produce the word

spontaneously, Bernthal and Bankson (1993) have suggested that a response

should be elicited by giving some prompts first without giving a model. If the

child still does not produce the word spontaneously, a delayed imitation can be

used whereby the examiner names the picture and then says something to

diminish the influence of the spoken cue; the last recourse is to use direct










imitation. It is important that words elicited by different methods should be

analyzed separately if possible instead of combining all responses as if they were

elicited in the same way.

The majority of studies have used single words as a method for collecting

data using pictures or objects. However, different procedures have sometimes

been used when no spontaneous response was obtained. A brief review of some

frequently reported studies indicated that Wellman et al. (1931) used spontaneous

picture naming; Poole (1934), Preisser et al. (1988), Khan and Lewis (1986), and

Smit et al. (1990) used spontaneous naming of pictures, objects, actions, or

questions with delayed imitation, or imitation, if needed; Templin (1957) used

either spontaneous or imitated picture naming; Arlt and Goodban (1976) used

imitation throughout. Prather et al. (1975) followed a sequence of spontaneous,

cues, forced choice questions, and then imitation, or no response. Ingram et al.

(1980) used sentence completion, sentence recall, and imitation.

Conversational samples have also been used to collect data about

articulation and phonological acquisition. Such samples have the advantage of

using spontaneous connected speech, and the words are used in meaningful and

real contexts. There are also some disadvantages because this method requires

more time to obtain and analyze the sample, some sounds may not be represented

in the sample, and the results cannot be compared to most of the norms available.

Of the studies reviewed in this paper, Olmsted (1971), Irwin and Wong

(1983), Stoel-Gammon (1985, 1987), and Vihman and Greenlee (1987) used









conversational samples. The method of data collection varied. Olmsted (1971)

used play sessions with the children; Irwin and Wong (1983) used non-structured

conversation exchanges and tape recorded the sessions; Stoel-Gammon (1985,

1987) tape recorded interactions with caretakers and children using the same set

of toys for all subjects; and Vihman and Greenlee (1987) tape recorded and

videotaped play sessions with mother and family peers and studied the production

of fricatives and clusters with pictures.

Spontaneously evoked samples of single words and imitated samples have

been criticized for several reasons. Critics point out that production of single

words may differ from the production of the same words by the same children in

connected speech-a problem of validity (Morrison & Shriberg, 1992); each sound

is targeted only once in each position and the consistency of production cannot be

assessed-a problem of reliability; and children's performance may vary based on

the familiarity and/or the structure of the word-a problem of both reliability and

validity (Bernthal & Bankson, 1984; Ingram, 1989). Some attempts have been

made to evaluate these possible differences in results due to the method of

elicitation (e.g., Ingram et al., 1980; Ingram, 1989; Kenney & Prather, 1984; Stoel-

Gammon & Dunn, 1985; Templin, 1947), and each study contributed to important

methodological considerations to be taken into account. However, the conclusions

of most of these studies indicated that only very small differences will appear in

the final results and that these differences should not rule out the use of a single-

word elicitation method if it appears to be the most useful and usable for a









particular study (Morrison & Shriberg, 1992). The purpose of a study should

determine the elicitation method used. To emphasize this point, Ingram (1989)

noted that regardless of the differences in the elicitation and analysis procedures,

Templin (1957), who used single words, and Olmsted (1971), who used only

conversation, obtained almost the same results.

Word selection

The selection of words used in normative studies has been dependent

mainly on the inclusion of target sounds needed in the corpus to satisfy the

purpose of the studies. Most normative studies have been based on samples

elicited from one single word for each sound in three positions-initial, medial,

and final. However, Prather et al. (1975) and Smit et al. (1990) tested only two

positions, initial and final. Arlt and Goodban (1976), Khan and Lewis (1986), and

Preisser et al. (1988) tested more than one sound per word in an attempt to keep

the number of words small. Ingram et al. (1980) and Smit et al. (1990) tested

each sound of interest in each position of interest more than once.

Sounds tested

Most normative studies reviewed included the majority of consonant

sounds of the language. Consonant clusters were included to some extent in most

studies, but vowels and diphthongs often were not tested. Wellman et al. (1931)

and Poole (1934) excluded infrequently occurring sounds. Ingram et al. (1980)

included only a selected group of sounds, because the study had a specific

purpose in examining these sounds.










Methods of Analysis and Presentation of Norms

Criterion for acquisition

In studies of sound acquisition, a criterion is set to determine "age of

acquisition" for the group as a whole. The criterion for acquisition is set by the

level of correctness of the sound specified by the researcher and varies

considerably from study to study. Precise definition of acceptable responses is

essential and can affect the results. Ingram criticized Templin (1957) for not

considering normal variants of phonemes as acceptable in scoring. For example,

an acceptable variant such as the flapped I/ was considered to be incorrect. The

issue of acceptable responses is also very important in languages that have various

dialects and for normative studies conducted nationwide.

Poole (1934) set the highest criterion for acquisition, 100% correct

production in all positions. Wellman et al. (1931) and Templin used correct

production by 75% of the children in each of three word positions. Ingram et al.

(1980) criticized the 75% criterion as too strict and suggested 70%. They required

correct production of 70% of the child's four attempts to produce a sound.

Prather et al. used correct production in two positions by 75% of the children,

and the scores were averaged over the two positions. Smit et al. (1990) used

correct production by 90% of the children in each age group in all positions,

defending this high criterion as appropriate for state guidelines where their data

were collected.







19

An important aspect in the discussion of defining a level of acquisition was

brought up by Sander (1972). He argued that the typically used 75% criterion

represented the "upper age limits rather than average performance" (p. 56). He

suggested that different indices of speech sound achievement should be

distinguished, and normative data should report at least two measurements:

customary age and age of mastery. Mastery is generally preceded by at least three

stages: (1) the first appearance of the sound, (2) the earliest correct articulation

of the sound in words, (3) customary production (51% correct production in two

positions), and finally (4) mastery (90% correct production in three positions).

Arlt and Goodban (1976), Ingram et al. (1980), Prather et al. (1975), and

Smit et al. (1990) compared their results with those found in earlier studies.

Agreements between the findings in the various studies were striking when the

differences in methodology mentioned above are considered. However, the more

recent studies have tended to report earlier ages of acquisition (e.g., Arlt &

Goodban, 1976; Prather et al., 1975; Smit et al., 1990).

Analysis procedures

The normative studies reviewed have reported different types of scoring,

different methods for reporting the data, different statistical analyses employed,

and different procedures for handling error production. The simplest type of

scoring in a test of sound production is to use "correct" and "incorrect" or 1 and 0.

Templin (1957) gave one point for each correct sound out of a total possible of

176. She then looked at correct percentages for each consonant, for each position,









and for each type of consonant by manner. Smit et al. (1990) refined this

procedure by scoring three types of responses: acceptable, not acceptable but with

marginal differences, and incorrect.

Another commonly used scoring procedure (Prather et al., 1975) is a four-

way system based on categories or degrees of correctness, such as Correct,

Omission, Substitution, and Distortion (e.g., numerical values 0, 1, 2, 3). In other

studies (Ingram et al., 1980; Preisser et al., 1988; Stoel-Gammon, 1985, 1987;

Yavas et el., 1991), an IPA (International Phonetic Alphabet) symbol was

assigned to the substituted or distorted sound. This gives more detailed

information but suits small groups better than large groups. Such detailed

information cannot be used in most statistical procedures, and the data are

difficult to report in tables.

Reporting the data

A variety of methods for reporting data can be found in normative

research on sound acquisition. Studies have often included the use of tables of

norms (e.g., Khan & Lewis, 1986; Silverio et al., 1994; Templin, 1957), figures or

graphs representing change over time (e.g., Prather et al., 1975; Smit et al. 1990),

and percentages, percentiles, or standard scores (e.g., Khan & Lewis, 1986; Yavas

et al., 1991). The data are usually reported separately by age group and by sex

(e.g., Poole, 1934; Prather et al., 1975; Silverio et al., 1994; Yavas, 1988) and by

the position of the sound in the word (e.g., Lamprecht, 1993; Templin, 1957;

Wellman et al., 1931). Some studies also included data by socioeconomic status or









geographical area (e.g., Khan & Lewis, 1986; Silverio et al., 1994; Smit et al.,

1990), by order of difficulty of sound classes (e.g., Silverio et al., 1994; Templin,

1957; Wellman et al., 1931; Yavas et al., 1991), and by phonemic environment,

cluster, or singleton (e.g., Ingram et al., 1980; Smit et al., 1990; Templin, 1957).

A different method of reporting was used by Dyson (1988), Stoel-Gammon

(1985, 1987), and Yavas et al. (1991). The phonetic inventories used by the

children were examined entirely, and the overall percentage of consonants

produced correctly was reported. In addition, the syllable and word shapes that

commonly occurred were reported.

In a number of studies, error productions have been either ignored or

simply counted. More recent research has emphasized the description of these

errors by using narrow phonetic transcription. Errors have been described in

terms of phonological processes, distinctive features, or exact substitutions and

distortions used by the children (e.g., Irwin & Wong, 1983; Khan & Lewis, 1986;

Mota, 1993; Prather et al., 1975; Preisser et al., 1988; Smit, 1993; Smit et al.,

1990; Vihman & Greenlee, 1987; Yavas et al., 1991).

Due to their descriptive nature, most normative studies have not made

extensive use of statistical procedures. If comparison between and among groups,

such as differences between genders or socioeconomic status, were needed for the

study, researchers have typically used a t-test or ANOVA to determine the

significance of differences. Sometimes percentile ranks and standard scores have

been computed as well.










Phonological Characteristics of Portuguese

This section of the paper presents an analysis of the main characteristics of

Portuguese sounds based on their phonetic features and the way in which those

sounds operate as a system. When the phonemes are combined in meaningful

sequences, they interact with one another in systematic ways that can be explained

by phonological rules. In this paper all references to Portuguese refer to the

variants of the language spoken in Brazil unless otherwise mentioned. The

inventory of phonemes used in Portuguese will be described.

The phonemic inventory of Portuguese is typically described as consisting

of 19 consonants and 7 vowels. Following is a brief account of these sounds

extracted from Mascherpe (1970).

Vowels

The vowels are classified by using the height and position of the tongue,

from highest to lowest and from front to back, in conjunction with the shape of

the lips, from spread or less rounded to very rounded. The vowels used in

Portuguese can be seen in Table 2-2.

Portuguese vowels are all syllabicic]; that is, they can be the nucleus of a

syllable. This feature distinguishes them not only from consonants, which are all [-

syllabic], but also from the glides /w/ and /j/, which appear in diphthongs and

triphthongs. Vowels can appear either in stressed or unstressed positions. In

Portuguese, as in other Romance languages, some stress opposition are












Table 2-2. Articulatory Classification of Portuguese Vowels.


Unrounded


Front


Rounded


Back


High


Higher mid

Lower mid


/t/


Low


Examples of Portuguese Vowels


Initial
patolpatu/
pelo/pelu/
pelelpeli/
mico/miku/
bolo/bolu/
m6to/m tu/
luva/luva/


Medial
batata/batata/
cabelo/kabelu/
boneca/bunEka/
camisa/kamiza/
tijolo/ti3olu/
mar6la/mar la/
coluna/coluna/


Final
vacalvaka/
*fome/fome/
cafg/kafE/
bule/buli/
*tenho/tej ff o/
mocot6/mocot /l
tatu/tatu/


* The vowel /e/ is frequently pronounced as /i/ in final position and the vowel /o/
is frequently pronounced as /u/in final position. Only if the /e/ and the /o/ are
followed by an /r/ or /s/, the original characteristic sound will be preserved (e.g.,
morrer, comer, calor, mrs, p6s; /moRer/, /kumer/, /kalor/, /mes/, /pos/).










neutralized under weak stress, and the vowel inventory consequently is reduced.

Examples of Portuguese vowels in common words are presented in Table 2-2.

Diphthongs and triphthongs

Besides the single vowel nuclei described above, Portuguese has a rich

variety of complex syllable nuclei, formed by a vowel phone accompanied by one

or two glides. Phonetically, a glide is produced by a movement of the tongue

toward or away from the area of articulation of one of the high vowels /i/ or /u/.

Such a complex nucleus is a rising diphthong (Table 2-3) if the glide is followed

by a vowel, as in quatro /'kwatru/, or a falling diphthong if the glide follows the

vowel as is pai /paj/. If two glides are involved, the syllable nucleus is a triphthong

as in the last syllable of Paraguai /para gwaj/.

Nasalization

Three types of vowel nasalization occur in Portuguese (Table 2-4). First,

there are nasalized vowels following nasal consonants in the same syllable, in

words like doma /d6ma/ or boina / b6jna/. Second, there are nasalized vowels

also in syllable-final position but adjacent to a nonnasal consonant in the

following syllable. In such cases, standard orthography represents the nasalized

nucleus as a vowel followed by a nasal consonant campo k x pu/, canto /k tu/.

The third and final case is that of nasalized syllable nuclei in word-final position.

These are pronounced in a manner similar to the second case; that is, as a

sequence of a simple or complex nucleus followed by an underlying /n/, as in the










Table 2-3. Examples of Portuguese diphthongs and triphthongs.


Rising Diphthongs


ui /wi/
ue /we/
ue /we/
ua /wa/
uo /w3/
uo /wo/


sagui
dueto
cueca
suave
quota
qiiociente


iu /ju/
ie /je/
ie /je/
ia /ja/
io /jo/
io /jo/


Falling Diphthongs.


iu /iw/
ei /ej/
eu /ew/
ei /Ei/
eu /ew/
ai /aj/
au /aw/


faliu
sei
seu
r6is
veu
pai
mau


ul /uw/
ui /uj/
oi /oj/
ou /ow/
oi /:j/
ou /Ow/


Triphthongs


uai /waj/ quais iai /jaj/ fiais
uei /wej/ suei iei /jej/ guiei
uou /wow/ su6u iou /jow/ guiou
uiu /wiw/ ruiu


mi6do
piedade
viela
quiabo
idi6ta
pi6lho


pulga
Rui
foi
sou
d6i
sol











Table 2-4. Nasal vowels in Portuguese.



Syllabic Nucleus Word-final Position


i fino /finu/ fim /fi/
e pena /'p na/ tern /t j/
a lama /'lxma/ 1 /IA/
o t6ma / t6ma/ som /s6/
u fundo /'ffindu/ um /h/
ai amAina /a mAjna/ mre /m 7j/
ei reino / R jnu/ sem /snj/
oi ac6ima /a k6jma/ p6e /po T/
ui arruina /a Ru T na/ ruimn /RuT/









words fim and ruim. These two words are pronounced as /fin/ and /Ru n/,

respectively, although orthographically they appear to end with a vowel followed

by /m/.

Consonants

The Portuguese consonant system has been variously described as including

from 16 to 21 distinctive units. This variation is due to competing interpretations

of the phonological status of /1/, /n/,/r/ (the "double" or "strong" r in words like

roupa, carro, and honra), and the glides /j/ and /w/. In this study the figure of 19

consonant phonemes has been adopted, with /1/, /n/, and /r/ considered as

independent phonological units, and the glides /j/ and /w/ as positional variants of

the high vowels /i/ and /u/.

The description of the consonants takes into account two broad

parameters: manner and place (or area) of articulation. Each contrast in place

and manner corresponds to a feature common to a group of consonants.

Manner of articulation

Continuant vs. occlusive. Continuant sounds, such /s/, /f/, /R/, are

articulated without interruption of the flow of air. If the articulators form an

obstacle to air flow, the consonant is an occlusive, such as /p/, /d/, /k/.

Nasal vs. oral. Nasal consonants are articulated with the velum lowered, so

that the air enters the nasal cavity through the nasopharynx. If the velum is

raised, the sound is oral.









Sonorant. This feature refers to the possibility of spontaneous voicing

taking place during the articulation of a sound, and it is shared by vowels, glides,

and certain consonants such as /1/ and /r/.

Lateral. In the articulation of lateral sounds, the air flows around the sides

of the obstruction in the oral cavity as in /I/ and /L/.

Vibrant. This feature refers to sounds produced by vibration of the tongue.

In Portuguese it applies to only two sounds, the flapped /r/ of cara / kara/ and the

"double r" of carro /'kaRu/. In one of its predominant phonetic manifestations,

the /R/ is produced as an alveolar trill /R/, but variations are common.

Voicing. Depending on whether or not the vocal folds vibrate, sounds are

classified as [+voiced] or [-voiced]. Those in which the vocal folds vibrate are

[+voiced], e.g., /b/, /d/, /v/. Examples of [-voiced] consonants include /p/, /t/, and

/f/.

Place of articulation

As can be seen in Figure 2-1, six places of articulation are relevant in

Portuguese. Each of these is described briefly below.

Bilabial. In this manner of articulation the articulators are the lips, which

act together to block the airflow, as in /b/, /p/, and /m/.

Labiodental. The upper front teeth form an obstacle to airflow as they

contact the lower lip, as in /v/ and /f/.

Apicodental. The apex of the tongue touches the inner face of the front

teeth, as in /t/ and /d/.














Portuguese Bi-labial Labio- Dental/ Alveo- Palatal Velar
Consonants Dental Alveolar Palatal
Stop pi b t d k g
(Plosive) "

Nasal m n i

Fricative f v s z f (x)


Affricate tj d3


Lateral 1 L* (R)

Vibrants r R


* /L/ has been used in figures and text to represent the sound/X/, because its symbol is
not available in this font.


Places and manners of articulation of Portuguese consonants.


Figure 2-1.










Laminoalveolar. The blade of the tongue is placed against the alveolar

ridge, as in /!/, /n/, /s/, I/z/.

Palatal. The pre-dorsum of the tongue touches the hard palate, as in /L/.

Dorsovelar. The dorsum of the tongue touches the velum, blocking the

airflow, as in /k/ and /g/.

Phonetic realization of consonants

Like the vowels, the consonant phonemes are affected, in varying degrees,

by the phonetic environments in which they occur. The following is a list of some

important variants of their phonetic realizations.

1. The /d/ and /t/ are palatalized before /i/ as in dia /d3ja/, and tia /tfja/. In

such cases, palatization may be described as resulting from a change in the

value of the feature [anterior].

2. The /s/ becomes voiced in syllable-final position before a voiced consonant

or vowel, as in as duas /az dwas/, as armas /a zarmas/, and it remains

voiceless if followed by a voiceless consonant or pause, as in este /'estfi/,

duas /'dwas/. The voicing rule applies regularly in syllable-final position,

generating a voiced fricative I/z/, as in desde /'dezd3i/, as well as in word

final position, if the next word begins with a voiced consonant: os dois

/uz dojs/.

3. The /r/ is a voiced apicoalveolar flap, as in caro I karu/. In final position, it

is maintained only in educated speech and slow, deliberate styles; otherwise

it is greatly reduced, devoiced, or dropped altogether, as in falar, /fa lar/,








31

/fa'la/, /fa la/. If final /r/ is followed by a word beginning with a vowel, the

/r/, if pronounced, links with that vowel, forming a new syllable-falar

alguma coisa /fa laraw gumakojza/. Some speakers use either /r/ or /x/ in

final position: amor /a mor/, /a mox/. In some rural speech, /r/ has a

palatal retroflex realization very similar to that of English postvocalic

rt---carpa / karpa/.

5. The /R/, which corresponds to an alveolar trill, has been used above to

represent the so-called "double r." In fact, this phoneme has several

different phonetic renderings, namely:

a. /r/ voiced apicoalvelar trill,

/r/ voiceless apicoalveolar trill,

b. /61 voiced alveopalatal fricative, slightly retroflexed,

/)/l voiceless alveopalatal fricative, slightly retroflexed,

c. /x/ voiceless velar fricative,

d. /R/ voiceless uvular trill, and

e. /h/ voiceless glottal fricative.

All of these variants are regionally distributed allophones of the same phoneme,

/R/. However, there is a noticeable amount of overlapping, so they are found in

free variation in the same dialect or even in the speech of the same individual

(Azevedo, 1981). The distinctive contrast between /r/ and /R/ is maintained only in

intervocalic position, as in caro vs. carro /'karu/ vs. / kaRu/. Elsewhere, /R/ occurs









in word initial position or after a consonant (Roberto, honra, Israel, guelra;

/Ro berto/, /' Ra/, /is Raew/, /gew Ra/).

Syllable Type and Composition

The simplest syllable type in Portuguese contains only one vowel (e.g., hd

/'a/). If the nucleus is a diphthong, one of the vowels must be both [+high] and [-

stressed] and articulated as a phonetic glide (e.g., oi /oj/). In Table 2-5 the syllable

nuclei are represented by V and different consonants by C.

In Portuguese, not all consonants occur in every position. The restrictions

of occurrence are best understood if single consonants and clusters are considered

separately. Table 2-6 illustrates the positions within words in which consonant

singletons can occur in Portuguese.

The number of possible consonant sequences in Portuguese is limited, and

the ones that hold the most interest are those that form cohesive clusters within

the same syllable. In clusters of the general type CC, the consonants are

necessarily different from each other; there are no sequences of identical

consonants in Portuguese. The first consonant must be a nonnasal stop (/p/, /b/,

/t/, /d/, /k/, or /g/) or a labiodental fricative (/f/ or /v/), and the other can be only

the anterior lateral /l or the flap /r/.

There are some co-occurrence restrictions for clusters (Table 2-7). First,

there are no /dl/ clusters, except marginally in foreign lexical items and












Table 2-5. Syllabic types of Portuguese.



V hi

VC is

CV da

CVC das

CCV pri

CCVC cruz

CCVCC trans











Table 2-6. Distribution of Portuguese Consonants.


Syllable Word
Word Initial Medial Final Final


pato
bata
toca
data
cata
gato
mato
nata
nhonh6
faca
vaca
saga
zaga
chaga
jto
lado
lhama


/p/
/b/

/t/
/d/

/k/
/g/
/m/
/n/
/n/
/f/
/v/
/s/I
/%/
/$/
//





/R/
If-


IIJ


1W


rato


capa
cabo
mato
cada
laca
toga
canma
cana
minha
alfa
alva
ago
vezo
acho
haja
mala
r6elho
caro
carro


canto





deste
desde




alto


farto


si, sAio





mas





sal


ser











Table 2-7. Examples of Portuguese clusters.


Initial Medial


p1 planet replica
pr prazo compra
bl bloco rublo
br brinco abraqo
tl tlintar atlas
tr ... atris
dl ... adleriano
dr dragao adro
kl clarin eclesifistico
kr crime recreio
gl glosa sigla
gr groselha regra
fl flauta rifle
fr frances africano
vl Vladimir ...
vr ... lavra








36

derivatives, such as Adler and adlenriano. Second, in initial position, /vr/ and /tl/ are

very rare. Third, the cluster /vl/ appears initially only in the foreign name Vladimir

and medially in borrowed words, such as the trademark Revlon.

Other non-cluster consonant sequences are not frequent in Portuguese

(Table 2-8). The only frequent phonological consonant sequences in Portuguese

are /ns/ and /nd/. Here, however, after nasalization of the preceding vowel, the

nasal consonant /n/is usually deleted, barely leaving a phonetic trace, as in the

plurals sons, bons and the progressive form of verbs such as nadando.

Studies of Acquisition of Portuguese

Very little empirical research about the acquisition or disorders of

phonology among Portuguese speaking children in Brazil has been reported. The

need for such research and for the publication of methodologies for the practicing

speech-language pathologist was recognized by Yavas, Hernandorena, and

Lamprecht in their 1991 book, Avaliagdo Fonol6gica da Crianqa Reeducaqdo e

Terapia (Phonological evaluation of children). Santini and Dyson (1995) also

stressed the importance of such research in their article, "Discussdo sobre

Metodologia de Pesquisa Relacionada a Area de Aquisigdo Fonol6gica" (Discussion

of methodological issues related to phonological acquisition). Yavas et al. (1991)

prepared their manual to provide a method for phonological analysis as a clinical

tool. They emphasized the relationship between theoretical phonological

principles and practical therapy issues. Two important contributions of this

manual are the presentation of an instrument for obtaining a sample of linguistic












Consonant sequences in Portuguese.


Initial(very rare)


Medial


ptialina
pneu
psicologia
bdelio










tmese


/pt/
/pn/
/ps/
/bd/
/bg/
/bz/
/bv/
/bm/
/bn/
/bs/
/tm/
/dg/
/dv/
/dm/
/dn/
/ds/
Id3/
/kt/
/kn/
/ks/
/kz/
/gf/
/gn/
/ft/


rapto
dispneia
decepcAo
abdicar
Abgail
abjeto
6bvio
abmigragAo
abnegaqao
absorver
atmosfera
Edgar
adventista
admitir
abdominal
adstringente
adjunto
ectoplasma
tecnico
ficao
eczema
Agfa
agn6stico
aft6se


Djalma
ctenideo
cnemio


czar


gn6mo
ftaleina


Table 2-8.









data (consonants) and the presentation of different theoretical and

methodological bases for analyzing the data collected.

The clinical instrument described by Yavas et al. (1991) was first reported

by Yavas (1988) in a pilot study of phonological acquisition in 72 Portuguese-

speaking children between the ages of 2:4 and 4:4. This instrument consisted of

five theme pictures used to elicit 120 words by spontaneous naming. The words

elicited and the procedures for their elicitation and analysis were revised and

presented more fully by Yavas et al. (1991).

In 1993, Lamprecht presented a study on the phonological acquisition of

Portuguese in children aged 2:9 to 5:5. She stated that the majority of current

studies of phonological development-normal or disordered-are based on the

Natural Phonology Theory (Stampe, 1969, 1973). Lamprecht reported that in the

last 10 years, 9 out of the 14 most important publications have been based on

Stampe's model. Two others analyzed distinctive features, two were founded in

generative phonology, and one on auto-segmental phonology. Because of this

emphasis on natural phonology and because its application to Portuguese was not

clear, Lamprecht (1993) conducted a longitudinal study describing the

phonological acquisition of Portuguese based on natural phonology theory. She

emphasized that such information is relevant, not only to enhance our knowledge

about the normal parameters of Portuguese phonology for clinical speech therapy,

but also as a data resource for educators in general.









Lamprecht included 12 children aged 2:9 to 5:5 in her study. The speech

samples were collected using the five theme pictures proposed by Yavas et al.

(1991) and described above. The data were analyzed using contrastive analysis

and phonological processes analysis. Her findings indicated most phonological

processes occurred less than 25% of the time by 4:2. The last processes to be

suppressed were cluster reduction, fronting, devoicing, and deletion of liquids in

the syllable final position within words.

Wertzner (1992) pointed out the lack of descriptions of the articulatory

acquisition of Brazilian children. Her study of 56 children (equally divided

between girls and boys) was intended to verify the use of phonological contrasts

and phonological processes in children between 3:0 to 7:0, seen at the Health

Center "Serviqo de Pediatria do Centro de Saiide Escola Prof. Samuel B. Pessoa."

All subjects were from the same "district" (Bairro do ButantA), and all were from

a low socioeconomic status. Her specific research objectives were: (1) to describe

the order of phoneme acquisition, the occurrence of phonological processes in the

children's speech, and differences in the occurrence of phonological processes in

the speech of the children and their mothers, (2) to verify the viability of using

the Khan-Lewis (1986) analysis model to describe the use of phonological

processes in Portuguese, and (3) to compare the first two objectives in terms of

the imitation and naming situation. Only the first objective is of primary interest

in the proposed study. Her results indicated some individual differences but a

general agreement with previous literature. Plosives and nasals were acquired










first, followed by fricatives, and then liquids. In terms of place of articulation,

labials preceded dentals/alveolars, which preceded palatals and velars. The lateral

liquids were acquired before the non-lateral liquids. She hypothesized that the

individual differences were due to different "strategies" used by the children. All

phonological processes, with the exception of consonant sequence reduction, were

suppressed by about 4:1 or 4:2. Consonant sequence reduction persisted until

about 5:2.

Mota (1993) looked at the practical application of phonological theory with

Portuguese-speaking children. She conducted a study to evaluate the efficacy of a

therapy model based on phonological processes in the treatment of children with

developmental phonological disorders. Three subjects were selected from the

children who went to one university speech and hearing clinic in Brazil with the

complaint of "speech problems." All three subjects were male, ages 6:2, 5:8, and

5:0. The subjects had been diagnosed as presenting developmental phonological

disorders before being treated with the phonological remediation model selected

(Hodson & Paden, 1986).

The subject selection was based on a speech-language and hearing

evaluation that included: an interview with parents for a case history, a

psychomotor evaluation, an orofacial exam, a language evaluation, an auditory

discrimination test, and complementary exams as needed (e.g., ENT, hearing

evaluation, neurological evaluation). The results from all exams except the

phonological evaluation were within normal limits. The phonetic inventories of all









three subjects were almost totally complete before treatment. However, their

contrastive phone systems showed the absence of important contrasts, which made

their speech unintelligible at times.

The phonological treatment plan consisted of two cycles with re-evaluation

after each cycle. Mota (1993) suggested that the results showed some

improvement between the initial evaluation and the second re-evaluation.

Although no information was given regarding length of treatment, the author said

that:

The analysis of the results obtained with this study led me to a conclusion
that supports the validity of the method used. The value of the model
used could be seen in terms of the short time needed to obtain significant
improvement in the phonological system of the subjects and, therefore, in
their pronunciation, and also in the facilitation of the appearance of carry-
over from the small amount of sounds trained. (Mota, 1993, p.49,
translated by this author)

In 1994, Silvirio et al. conducted a study to describe the occurrence of

Portuguese phonemes in preschool children and to compare the phonological

performance of children in private schools as opposed to public schools. The

variables considered in their subject selection were: gender, age, and type of

school attended. The study included 235 children, 117 from private schools and

118 from public schools. The subjects were almost equally distributed between

males and females. Their ages ranged from 2:6 to 5:6 divided into three groups:

2:6 to 3:6, 3:6 to 4:6, 4:6 to 5:6. The testing procedure consisted of three parts:

(1) spontaneous speech (with a set of questions), (2) picture naming (25 two-

syllable words), and (3) word repetition (22 two-syllable words). All sounds were








42

tested in the initial position except for /L/, /r/, /n/, which were tested in the medial

position. The /s/ and /r/ were tested in the word final position. The criteria for

correct production (acquisition) by each child was one out of the three possible

elicitations of a sound. Distorted productions were considered to be incorrect.

According to the authors such a criterion shows that the child was capable of

correct emission of the sound, but one cannot evaluate the systematic use of the

sound.

According to Silverio et al. (1994) the analysis of the data collected showed

that: (1) the phonemes were earlier then usually reported in the literature; (2)

there is a difference between the two groups in terms of ages when phonemes

occur, although the hierarchy of phonemes is preserved; and (3) most of the time,

the phonemes were used by the children without distortions or substitutions from

the start. The authors of the study suggested that the superior performance of the

subjects from private schools was probably due to higher expectations for children

among those in higher social economic (and cultural) levels.

Statement of the Problem

The Need for this Research

Phonological acquisition can be seen as an interesting topic that has

fascinated researchers trying to solve the puzzle of how children learn to talk. The

review of the literature presented above provides some examples of a large body

of literature devoted to child phonology showing several different attempts to

improve aspects of the investigation. It is safe to say that speech is an extremely








43

complex behavior that encompasses many variables, making the investigator's task

an enormous one.

Although several different theoretical frameworks have been applied to

data from children's speech in English, not much has been done in terms of

evaluating the validity of those results when applied to other languages. In terms

of Portuguese it can be seen that child phonology is an expanding area with

several studies dating from 1992 to the present. In addition, most studies

addressing the acquisition of Portuguese phonology have emphasized the need for

the collection of more data in the area (Lamprecht, 1993; Silverio et al., 1994;

Wertzner, 1992), and especially in the area of phonological disorders.

According to the last census, there are nearly 160 million people living in

Brazil. Portuguese is the only language spoken in that country and no major

dialects are found, except for small accent variations that do not affect meaning.

This language was chosen for several different reasons. First, Portuguese operates

differently than English; therefore, its study may provide cross-linguistic

information on how the phonological system develops. Second, in the few studies

found addressing the phonological acquisition of Portuguese (Hernandorena,

1993; Lamprecht, 1993; Silverio et al., 1994; Wertzner, 1992; Yavas, 1988) the

need for further studies that would provide large scale normative data was

pointed out. Finally, the area of phonological disorders, more specifically, the

field of speech-language pathology has been trying to build a more theoretical

basis to support and legitimize clinical findings. Therefore, the present study










proposed a detailed control of many of the methodological issues raised by

previous researchers in the attempt to provide a collection of data about

acquisition of phonology by Portuguese speaking children.

Purpose of Study

The general purpose of the study was to collect normative data on the

acquisition of consonantal sounds of Portuguese. Samples were collected from

normally developing children in Brazil. The data were be used to answer the

following questions:

1. What percentage of each consonant was produced correctly by children at

each age level?

2. What is the age of "customary production" for each sound?

3. What is the age of "acquisition" for each sound?

4. What is the age of "mastery" for each sound?













CHAPTER 3
METHODOLOGY


The purpose of this study was to collect normative data on the acquisition

of the consonant sounds of Portuguese. These data were used to answer a number

of questions about the ages and order of acquisition.

Subjects

The subjects included 192 monolingual, Portuguese-speaking children.

Their ages ranged from 2:0 to 6:10 (years:months) in the following intervals:

2:0 to 2:4,

2:6 to 2:10,

3:0 to 3:4,

3:6 to 3:10,

4:0 to 4:4,

4:6 to 4:10,

5:0 to 5:10, and

6:0 to 6:10.

Twenty-four children were included in each of the eight groups. The ages and

composition of each group can be seen in Table 3-1. Each age group was

intended to include an equal number of girls and boys, although a few minor

exceptions to this rule had to be made among the youngest groups.










46








Table 3-1. Distribution of subjects' ages (years:months) and gender (M=--male,
F=female) in each group.


Groups


2:0 2:6 3:0 3:6 4:0 4:6 5:0 6:0
Age Gender Age Gender Age Gender Age Gender Age Gender Age Gender Age Gender Age Gender


2:0 M 2:6 M 3:0 M 3:6 F 4:0 F 4:6 M 5:0 F 6:0 M
2:0 M 2:6 F 3:0 F 3:6 M 4:0 F 4:6 F 5:0 F 6:0 F
2:0 M 2:6 F 3:0 F 3:6 F 4:0 F 4:6 F 5:0 F 6:0 M
2:0 F 2:6 M 3:0 F 3:8 F 4:0 F 4:6 M 5:1 M 6:0 F
2:0 F 2:6 F 3:0 F 3:8 F 4:1 F 4:7 M 5:1 M 6:0 F
2:0 M 2:6 F 3:2 F 3:8 M 4:1 M 4:7 F 5:3 F 6:0 M
2:2 F 2:6 F 3:2 M 3:8 M 4:2 M 4:8 M 5:3 F 6:2 M
2:2 M 2:6 M 3:2 M 3:9 F 4:2 F 4:8 F 5:4 F 6:3 M
2:2 F 2:6 F 3:2 F 3:9 M 4:2 F 4:8 M 5:4 M 6:3 F
2:2 M 2:7 M 3:2 M 3:9 F 4:2 M 4:8 F 5:4 F 6:4 F
2:2 F 2:8 M 3:2 F 3:10 M 4:2 M 4:8 F 5:4 F 6:5 M
2:2 M 2:8 F 3:2 M 3:10 M 4:2 F 4:8 F 5:4 M 6:6 M
2:2 F 2:8 F 3:3 F 3:10 F 4:2 M 4:8 M 5:5 M 6:6 F
2:2 M 2:8 F 3:3 F 3:10 F 4:2 F 4:8 M 5:5 F 6:6 F
2:2 F 2:8 F 3:4 M 3:10 M 4:2 M 4:8 M 5:5 M 6:7 F
2:3 F 2:9 M 3:4 M 3:10 F 4:2 M 4:8 F 5:6 M 6:7 M
2:3 F 2:10 M 3:4 M 3:10 M 4:3 F 4:9 F 5:7 F 6:7 F
2:4 F 2:10 F 3:4 F 3:10 M 4:3 M 4:9 M 5:7 M 6:7 F
2:4 M 2:10 M 3:4 M 3:10 F 4:3 F 4:9 F 5:9 M 6:8 M
2:4 F 2:10 F 3:4 F 3:10 M 4:3 F 4:9 M 5:10 M 6:8 F
2:4 F 2:10 F 3:4 M 3:10 M 4:4 M 4:10 M 5:10 M 6:8 M
2:4 M 2:10 M 3:4 F 3:10 M 4:4 M 4:10 F 5:10 F 6:9 M
2:4 M 2:10 F 3:4 M 3:10 F 4:4 M 4:10 M 5:10 M 6:9 F
2:4 F 2:10 M 3:4 F 3:10 F 4:4 M 4:10 F 5:10 F 6:9 M








47

An oro-facial screening examination was conducted to exclude subjects who

showed any abnormalities in the speech mechanism that could interfere with

speech production and/or language development. No child considered as a subject

exhibited such physical abnormality.

The selection of subjects was intended to represent the socioeconomic

status of the population of the State of Sdo Paulo, Brazil. An attempt to mirror

the entire population was made by testing children from both one public and two

private schools.



The Assessment Instrument

Stimuli

A picture-naming test developed by Yavas et al. (1991) was used to collect

the data. The test consists of five theme pictures (Vehicles, Living Room, Bath

Room, Kitchen, and Zoo) used to elicit spontaneous single word responses. The

pictures can be seen in Appendix A, and the word lists in Table 3-2. The five

theme pictures encouraged the elicitation of 125 words, 97 basic words plus 28

optional words shown in italics. The words sampled are commonly present in

children above 3 years of age, test the target sounds necessary, and are easily

represented through drawings (Yavas et al., 1991). The numbers of each sound

tested in each position are presented in Table 3-3. According to Yavas et al., the

words chosen had the following phonological criteria: (1) a balanced













Table 3-2. List of words targeted by each picture.


Folha de GravaAo
Nome: Data:
Idade:

Zoo Kitchen Living Bath Vehicles


Room


Room


borboleta
cachorro
cobra
comer
dois
dragao
fl6r
floresta
grama
grande
latir
olhar
passarinho
pedra
peixe
pular
rabo
sol
tigre
verde
zebra
zool6gico
orelha
voar


abacaxi
aqficar
cafe
estrela
feijio
fogao
frio
fruta
garrafa
geladeira
janela
prato
soprar
vela
vidro
banana
bolo
fogo
OVO
tamovopa
tampa


brinquedo
cruz
dinheiro
disco
globo
guard
chuva
igreja
journal
lipis
livro
martelo
mesa
palhaqo
plant
prego
quadro
radio
tapete
televised
tesoura
antena
botdo
estate
franja
poltrona
telhado


banquinho
blusa
bolso
bravo
calqa
camisa
chave
chinelo
dedo
dente
escovar
nariz
pescogo
rel6gio
sabonete
toalha
esperar
armdrio
azulejos
cabelo
cano
espelho
menino
perna
porta
saia
sapato
tomrneira


andar
bicicleta
brincar
carro
criancas
dizer
dirigir
estrada
frente
fumaqa
microfone
nadar
nuvem
placa
tia/tio
tocar
trator
trem
dncora
chaming
navio
roda
trilho
sino















Table 3-3. Distribution of consonants by position. SIWI=syllable-initial,
word-initial; SIWW=syllable-initial, within-word; SFWW=
syllable-final, within-word; SFWF=syllable-final, word-final.

Total
SIWI SIWW SFWW SFWF per sound


I1

23


3


3

3

3


. . . .


x
Total per
position


5

3
5 ": :", :- .

5

12

9 6

3


151 26 8


8

6

5

15

18

7








50

representation of the target phonological system of Portuguese (that is, the adults'

phonological system); (2) more than one occurrence of each possible different

target; and 3) opportunities for the sounds to occur in different positions within

the words and in words with differing syllable structure and number of syllables.

Four syllable and word structures were considered: SIWI (syllable-initial, word-

initial), SIWW (syllable-initial, within word), SFWW (syllable-final, within word),

and SFWF (syllable-final, word-final).

The five theme pictures allowed the elicitation of 125 words along with the

opportunity to elicit more data in the form of narratives and descriptions.

According to Yavas et al., the set of pictures allows the elicitation of a

representative sample because it is not limited to one representative of each

target sound. Contrary to most articulation tests, this test provides at least three

possible occurrences of each consonant sound of Portuguese in all positions. The

only instances where the principle of three occurrences in all positions is not met

are described below.

Iz/ SIWI. The Portuguese vocabulary has few examples of /z/ in the word

initial position. The two words included in the test (zebra and zool6gico-zebra

and zoo) can be easily produced by children as young as 3 years old.

/t f/. /d3/ SIWI and SIWW. Although these two sounds affricatess) are

normal allophones of /t/ and /d/in most variants of Portuguese, they were

included in order to provide a more complete picture of child phonology. The

pictures elicited only two words for each of the two sounds in SIWI because a










third example is expected to occur naturally when the child says tio and/or tio,

referring to the examiner, and some forms of the verb dizer, as digo or disse (to

say).

I/ SFWW and SFWF. Only two words were provided for each of the two

syllable-final possibilities. Yavas et al. suggested that, because the /1/ is produced

as a semi-vowel in final position, the two examples provided are enough for the

this non-consonantal sound.

Clusters. Most clusters were elicited in only one example, and some

clusters are not even present. This decision was made based upon the rationale of

presenting an adequate vocabulary for children. Even the limited set of word

containing clusters that are included may not be totally appropriate for young

children. For example, the words: vidro, dragao, floresta, globo, prego, igreja, cruz,

poltrona, trator, estrada, trilho, plant (glass, dragon, forest, globe, nail, church,

couch, tractor, road, rail road, plant) may not occur spontaneously in the speech of

young children. In addition, it has been observed that even older children who

produce these words, often reduce the clusters. The words fruta and brinquedo

(fruit and toy) seemed easy but are difficult to elicit because they refer to

collective nouns and not to specific names. It should also be noted that the

number of possible consonant sequences in Portuguese is limited, and clusters do

not occur often.

Some words that were not represented in the pictures needed to be elicited

by a question. These words are: dizer, dois, frente, frio, grande, latir, microfone,








52

soprar, tia/tio, verde, zool6gico (to say, two, front, cold, big to bark, microphone, to

blow, aunt, green, zoo). Some words to be elicited were directly related to a

specific picture; for example, latir (to bark) appeared in the list of words for the

zoo picture because that is where the dog is represented. Other words that are not

directly related to any one picture were distributed between the five drawings so

that no word list would carry more or less words to be elicited verbally. However,

the words that were not directly represented in the drawings could be easily

integrated in the exam as part of the conversation. These are microfone, dizer,

tia/tio, and frente (microphone, to say, colloquial name for Ms/Mr, and in front of).

Procedure

The test was administered to each child individually in a quiet room. If the

child refused to stay alone with the examiner, one adult was allowed to stay with

the child. When testing the youngest children, the class teacher or the aid usually

stayed with the child and helped the examiner. The exam was tape recorded using

a lapel microphone clipped to the child's clothing, approximately six to eight

inches from the child's mouth. A total of thirty-five 90-minute tapes were used for

the recordings.

To administer the test the examiner placed each of the five theme pictures

one at a time in front of the child. The child was instructed to look at the picture

and then to tell all that he/she could about it. The examiner followed the child's

productions using an alphabetical list of all the words to be elicited by each

picture. The examiner repeated the responses of very young children who








53

appeared difficult to understand to ensure that they would be recognized on the

tape recording. A check mark was placed on the response form beside each word

named by the child, regardless of the accuracy of the articulation. The words that

were optional appeared in italics and were not prompted if not produced

spontaneously. When a child did not produce all of the obligatory words, the

examiner used delayed imitation with some form of the question: "This is a

cabinet (target word) in the bathroom (picture name)? Can you tell me that?" In

cases where delayed imitation did not elicit the desired word, direct imitation was

used. The child was sometimes praised and/or encouraged to name the pictures

but was not reinforced for correct productions.

Duration of the Exam

The exam was performed in 20 to 30 minutes with most children. For the

younger children, a much longer session often was needed, and three subjects

were tested in two sessions to assure the validity and reliability of results.

Test Environment

The room used for the data collection was a quiet room away from street

noise or conversational background noise. The examiner and the child were

usually sitting at the table, although the younger children were often tested on the

floor.

Examiners

Three examiners, including the author of this research, collected the data.

All three examiners were speech-language pathologists with several years of








54

experience. The author trained the examiners to administer and record the exam

and provided a written protocol for them to follow to ensure a standard

procedure and to clarify possible questions. In fact, the author tested

approximately 60-70% of the children herself.

Tape Recording Equipment and Procedure

The exam was recorded using a Marantz tape recorder model CH 221 with

an external lapel microphone. The entire exam session was recorded because

words used in the greeting and the test recording were analyzed as a part of the

sample (i.e., tia, a colloquial name for an adult female; microfone, microphone;

dizer, to say). Each taped sample was checked immediately after recording to

ensure that it was audible and had a minimum of extraneous noise. None of the

samples needed to be discarded for any technical reason.

Transcription

Narrow transcription was used to transcribe the children's productions of

the target sound. Approximately 30% of the audiotape material was transcribed

by two trained and experienced transcribers, the author and supervisor, using the

consensus procedure outlined by Shriberg et al. (1984). After transcription of each

utterance, the two versions were compared. Segments on which the transcribers

agreed were accepted and entered on the data collection sheet mentioned above.

Disagreements were resolved by replaying the utterance a maximum of three

times with each transcriber attempting to hear the other's transcription as well as








55

trying to confirm their own transcription. The remaining 70% were transcribed by

the author alone with spot checks by the supervisor.

To examine the reliability of transcription, 10% of the samples, one subject

selected randomly from each group were transcribed independently by the two

listeners. A program for computing reliability using relative weights for different

types of disagreement was used. This program, which is part of the Logical

International Phonetics Program -LIPP (Oiler & Delgado, 1990) was used for the

analysis of the data and for comparison of the two transcribed versions sound by

sound. The rule set for this program was originally included in the LIPP but

several modifications were made for the sounds of Portuguese.

Data Entry and Tabulation

After transcribing each sample, the data were entered on a data collection

sheet (Appendix B) and later entered into the computer using the LIPP (Oiler &

Delgado, 1990). This program allows entry of data in IPA symbols with any

desired set of diacritic mark. The expected standard form of each target sound

was entered in the LIPP program ahead of time. For each sound tested, a set of

rules compares the child's production with the acceptable response to determine

whether or not the response can be counted as "correct."

Validity of the Test

In order to account for word elicitation, a pilot study conducted by Yavas

et al. (1991) tested 40 children between 3 and 4 years of age. The results showed

that most basic words reached at least 50% adequate elicitation. The basic words










that did not meet the 50% criteria were: zool6gico, zebra, globo, floresta, plant,

microfone, placa, fruta, cruz, claro, igreja, and dragdo (Zoo, zebra, globe, forest,

plant, microphone, sign, fruit, cross, bright, church, and dragon). Those words were

maintained on the test to provide a significant phonological sample as other

suitable words could not be found (Yavas et al., 1991). In terms of content

validity, this is implicit in this type of study when all sounds of the language in all

positions are included. To examine construct validity, the extent to which accuracy

of the sounds on the test reflects increases due to increasing age were observed as

a part of the overall purpose of the study.

Presentation of the Data

To determine the percentages of each consonant sound produced correctly

at each age level, the data were tabulated by age level and by sex, with an entry

made for each sound. The four positions were treated independently to control

for the position factor. The percentages of correct productions at each age level

in each position (e.g., SIWI) was computed separately from the percentages in

any of the other three possible positions (SIWW, SFWW, and SFWF).














CHAPTER 4
RESULTS

This study was undertaken to answer four questions. The results intended

to answer the first question will be presented first under several headings. The

remaining three questions will be combined, and the results intended to answer

these three questions will be presented together.

Question 1. What percentage of each consonant was produced

correctly by children at each age level?

The percentages of correct production at each age level for all sounds

combined can be seen in Figure 4-1. It should be noted that the youngest subjects

sampled in this research were 2 years old, which suggests that phonological

acquisition had started before the child was tested. Over 50% of the consonants

attempted were already produced correctly by the youngest group. It can be seen

that between the first group (2:0) and the second group (2:6), the most dramatic

increase in accuracy was made, from 51% to 67%. The increases between each of

the other groups ranged from 2% to 9%. By 4:0 phonological acquisition was

essentially complete. Between the ages of 4:0 and 6:0, there is a two year time

frame during which the change was very subtle. By 6:0, 97% of the consonants

were correct.




















Consonants Produced Correctly .,
9%100% 9



6 0 % - ........-....-.-.---.---................,,^ .... ...................................................
40%

8 0% - ----......------ ....................................... .................................................................................
473%

0
cc)
8 0L.....-- ^^^


4-.
40% 4 ............................--............................................... ......................................




0%-

0% I I I
2:0 2:6 3:0 3:6 4:0 4:6 5:0 6:0
Age


Figure 4-1. Overall percentage of consonants produced correctly by each group.










SIWI. Syllable-Initial. Word-Initial

The percentages of each consonant produced correctly in the SIWI

position by all groups can be seen in Table 4-1. The results for each group are

illustrated separately on Figures 4-2 through 4-9. It can be seen that the nasals

and the stops were produced correctly earlier then the fricatives and liquids, with

the exception of the two fricatives /f/ and /v/, which were acquired earlier than the

others. A tendency to produce voiceless consonants more accurately than voiced

consonants was observed across all age groups. The youngest groups showed a

marked increase in the percentage of correct production for both nasals and stops

between 2:0 and 2:6 years of age. The liquids were produced with greater than

80% accuracy at age 3:0 in SIWI position. The fricatives /s, z, f, S/ were the least

accurate sounds. The affricates, especially /tf/, were produced correctly earlier

than the fricatives, at age 3:0.

SIWW, Syllable-Initial. Within-Word

The percentages of each consonant produced correctly in the SIWW

position by all groups can be seen in Table 4-2. The results for each group are

illustrated separately on Figures 4-10 through 4-17. For syllable-initial within-word

(SIWW) position it can be seen that the nasals and stops were again produced

earlier than the fricatives and the liquids, with the exception of /f/ and /v/, which

were acquired earlier. It should be noted that three nasals were possible in this

position as opposed to SIWI, where only two occur. However, this did not change

the percentage of correct occurrence of nasals in relation with the SIWI position.














Table 4-1. Percentage of each consonant produced correctly in the SIWI position in
each age group.

Age Groups


2:0 2:6 3:0 3:6 4:0 4:6 5:0 6:0
b 46.85% 70.72% 86.88% 93.22% 94.89% 97.91% 96.17% 100.00%

p 82.88% 96.57% 96.14% 98.35% 99.19% 99.60% 99.21% 99.61%
d 42.31% 80.33% 80.68% 96.70% 96.74% 96.81% 95.79% 100.00%
t 78.22% 87.38% 96.21% 95.71% 97.20% 98.58% 99.29% 100.00%
g 25.93% 66.22% 71.96% 89.17% 94.96% 95.76% 94.07% 99.17%
k 71.92% 92.98% 92.44% 98.73% 99.15% 100.84% 100.00% 100.00%
m 75.93% 87.50% 95.71% 97.50% 98.75% 96.55% 100.00% 100.00%
n 58.00% 78.95% 90.41% 85.92% 87.84% 88.46% 96.10% 95.29%
v 31.43% 63.46% 66.67% 85.56% 92.22% 91.40% 95.79% 98.95%
f 67.86% 93.38% 94.41% 98.96% 99.48% 98.45% 100.00% 100.00%
z 29.41% 60.00% 37.50% 57.78% 70.83% 75.56% 72.92% 97.92%
s 25.93% 31.75% 40.43% 61.05% 69.00% 84.21% 93.27% 90.29%

f 13.79% 17.95% 45.83% 66.67% 87.23% 89.58% 93.75% 97.96%
3 12.96% 14.52% 25.35% 47.89% 44.44% 71.83% 83.10% 95.83%
tf 40.43% 78.18% 80.56% 89.71% 95.71% 98.55% 94.44% 100.00%
d3 14.29% 39.13% 52.46% 65.22% 82.86% 87.32% 94.20% 100.00%
1 31.25% 79.55% 86.36% 87.14% 98.59% 100.00% 98.59% 98.61%
r 15.91% 37.93% 80.82% 91.25% 96.00% 98.73% 98.77% 98.73%
All


Sounds 40.28% 61.92% 69.52% 79.29% 84.48% 87.90%


89.76% 93.28%

















SIWI-Group 2:0
100%

S) 80%
I--
o 0%- ----
0)
CO f
C.) I Ii;l.Iii
(D 20%
a.


Sound


Figure 4-2.


Percentages of accuracy of each consonant in the SIWI position for
Group 2:0.





















SIWI-Group 2:6

100%
-I-'

o 80%
8 0 % -- ----.. .... ... ..... ....... ..... ...... ... .

0
0 60% ......
(D
0)
"C 40% ......

D 20%
0.
0%


Sound


Figure 4-3. Percentages of accuracy of each consonant in the SIWI position for
Group 2:6.



















SIWI-Group 3:0
100%
t5
u 80% ...............
0
o 60%-
4L 0% ............. .
(D II I

0 20% I
CL


Sound


Figure 4-4.


Percentages of accuracy of each consonant in the SIWI position for
Group 3:0.
























SIWI-Group 3:6


100%


80%


60%


40%

20%


0%


Sound


Figure 4-5.


Percentages of accuracy of each consonant in the SIWI position for
Group 3:6.


C-)
0

0
0)
0)
CO
.4-1
(D
Cm
mi

2-





















SIWI-Group 4:0


Sound


Figure 4-6.


Percentages of accuracy of each consonant in the SIWI position for
Group 4:0.


100%






















SIWI-Group 4:6


120%

100%

80%

60%

40%

20%

0%


Sound


Figure 4-7. Percentages of accuracy of each consonant in the SIWI position for
Group 4:6.







67













SIWI-Group 5:0
100%-
oo%-- .-. .. --------------- .

j 80%
L..
0
0 60%
()
0)
40%

D) 20%
0-
b p d t g k m n v f z s j f x
Sound









Figure 4-8. Percentages of accuracy of each consonant in the SIWI position for
Group 5:0.
























SIWI-Group 6:0


s f 3 tf d3 1 x


J L Uq n = a a = N m m m N I


b'p'd't g'km' n'v'f'z
Sound


Figure 4-9. Percentages of accuracy of each consonant in the SIWI position for
Group 6:0.


100%-

80%

60%-

40%-

20%


0%


I














Table 4-2. Percentage of each consonant produced correctly in the SIWW position in
each age group.

Age Groups


2:0 2:6
51.09% 78.23%
94.12% 100.00%
51.28% 56.46%
94.22% 100.00%
48.15% 61.11%
75.57% 97.87%
91.43% 94.34%
87.23% 94.48%
47.71% 65.32%
82.93% 96.30%
36.25% 40.38%
30.15% 33.95%
18.33% 28.17%
16.00% 21.21%
53.57% 78.67%

17.86% 26.53%
52.71% 87.50%
35.59% 50.00%

54.68% 67.25%


3:0 3:6 4:0 4:6 5:0 6:0
77.02% 91.57% 94.01% 96.39% 94.61% 98.81%
51.09% 100.00% 100.00% 96.51% 100.00% 100.00%


69.15% 76.41% 83.95% 82.28%
99.15% 99.59% 97.57% 99.60%
67.24% 82.11% 90.91% 94.26%
94.72% 98.64% 99.03% 99.67%
96.30% 100.00% 100.00% 98.95%
98.97% 98.56% 98.08% 97.67%
65.22% 81.17% 89.10% 93.98%
96.97% 98.51% 100.00% 100.00%
42.02% 52.10% 66.12% 75.00%
47.56% 58.72% 58.58% 80.17%
54.79% 73.97% 86.11% 90.28%
28.16% 35.65% 58.47% 69.83%
79.82% 88.70% 93.50% 97.44%
47.69% 46.38% 67.61% 72.22%
85.87% 85.00% 96.83% 100.00%


83.33% 90.28%


88.62% 93.90%
98.81% 100.00%
91.20% 98.40%
98.69% 99.68%
98.95% 100.00%
99.09% 100.00%
90.91% 97.75%
98.55% 100.00%
83.19% 92.50%
85.23% 91.25%
91.67% 98.61%
76.47% 87.50%
95.80% 100.00%

81.94% 92.96%
96.15% 98.95%


92.86% 100.00% 100.00% 100.00%


71.39% 80.96% 87.37% 91.35% 92.77% 97.24%


All
Sounds





















SIWW--Group 2:0

100%
t5 -
8 0 % ......... .... ......... .............................................................

280%
I-
0


20 Q % .. ... ...... .. ... ..... .. ... .. ..... .. ...... ............. ......... ..................... ....... ...........
0 60% I_




b p d t g'km n'p v f z s f 3 t/d3L I r x
Sound


Figure 4-10. Percentages of accuracy of each consonant in the SIWW position
for Group 2:0.



















SIWW--Group 2:6

100%


0
60% ............
0)

40 ....1 1 ... ................
240% .....

~30% L + +
0% b p d t g k m n jv f z s f 3 tfd3L 1 r x
Sound








Figure 4-11. Percentages of accuracy of each consonant in the SIWW position
for Group 2:6.





















SIWW--Group 3:0


100% 1
tot
g 80% ............

6 0 % .... .. . .. ... ............. ......... .... ........ ..... - -
4_ 0 % . .. .. . ... .. .. .. .. .. .... .. .......... ... .
40%.


D20%
CL
0% I2 L

b p d t g k m n j v f z s f 3 tfd3L r x
Sound









Figure 4-12. Percentages of accuracy of each consonant in the SIWW position
for Group 3:0.






















SIWW--Group 3:6

100%

g ) 8 0 % - -- -- ---------- ---------- ------ ------ .....
4-,
2 8 0 % ................ ......................
0
S60% ..Q
0
0)1
240% .. .. .
20% -,
C..
0% 20%
0% b p d t g k mnjp v f z s 3f tfd3L I r x
Sound


Figure 4-13. Percentages of accuracy of each consonant in the SIWW position
for Group 3:6.




















SIWW--Group 4:0


100%


Sound


Figure 4-14. Percentages of accuracy of each consonant in the SIWW position
for Group 4:0.























SIWW--Group 4:6

100%
4-
2 80% .. ..
0
S60% .....
G)
0)
^- 40%

0) 20% -. ....
0%-
bp d t g k m n v f z s f 3tfd3L I r x
Sound


Figure 4-15. Percentages of accuracy of each consonant in the SIWW position
for Group 4:6.






















SIWW--Group 5:0

100%-
ID00% -*--- -- -------------------------------------------
g) 80%
0
0 60%
0)
H40%-
C

"( 20% .-
0-
0% + + d + + + + -
b dt m n v f z s f 3 tfd3L lrx
Sound


Figure 4-16. Percentages of accuracy of each consonant in the SIWW position
for Group 5:0.




















































Figure 4-17. Percentages of accuracy of each consonant in the SIWW position
for Group 6:0.










A tendency to produce voiceless sounds more accurately than voiced

sounds can be seen again for all age groups. As in the SIWI position, a growth

spurt can be seen between Group 2:0 and Group 2:6, especially on nasals and

stops. It should be noted that four liquids were possible in this position as

opposed to SIWI, where only two occur. The inclusion of these two additional

consonants, the /L/ and the /r/, in SIWW position appeared to affect the overall

performance on the liquids. The // and /R/ (indicated on figures as /x/ because of

font restrictions) showed approximately equally correct production, over 75% at

Group 3:0. However, the /L/ and /r/ lagged behind and did not reach 75% correct

until 4:0 or later. The affricates were produced correctly earlier than the

fricatives. The voiceless affricate /tf/ was again more accurate than the voiced

cognate /d3/. The fricatives were produced correctly later than the other sounds,

with the exception of /f/ and /v/.

Comparison of Percentages of Correct Production by Position

Table 4-3 (also presented as Table 3-3) illustrates that most consonants

tested are possible in only two positions, SIWI and SIWW. Portuguese has very

few consonants that are either syllable final or word final. In the SFWF position

only /s/ and /r/ occur, whereas in the SFWW position /n/, /s/, and /r/ occur.

Variants used in some geographical areas also would include final /!/. Other areas

would consider final If and /R/ instead of /s/ and /r/ as correct productions.

In the next section each group will be considered, starting with the

youngest group, 2:0. The results are illustrated on three figures for each group,















Table 4-3. Distribution of consonants by position. SIWI=syllable-initial,
word-initial; SIWW=syllable-initial, within-word; SFWW=
syllable-final, within-word; SFWF=syllable-final, word-final.


SIWI

10

11

4

6

5

10

4
A


151 26 8


4

9

2

5

2

3

3

3


32, m


x
Total per
position


SIWW

7

4

22

11

6

13

4

10

3

8

3

5

10

3

5

5

3

5

12

9

3


Total
per sound
17

15

26

17

11

23

8

26

3

12

12

7

28

5

8

8

6

5

15

18

7

277








80

the first showing stops, the second showing fricatives and affricates, and the third

showing nasals and liquids.

Group 2:0. Figures 4-18 to 4-20 show the accuracy of stops, nasals and

liquids, and fricatives and affricates, respectively, in the two positions tested. It

can be seen that children in Group 2:0 were more accurate in the SIWW position

than in SIWI for all sounds. The liquid /r/ in all positions, including the SFWF,

starts with a very low percentage of correct production by Group 2:0. The

fricative /s/ in SFWW also has a low percentage of correct production in the early

age groups.

Group 2:6. In Figures 4-21 to 4-23 it can be seen that the accuracy for

most sounds in the SIWW position is still somewhat higher than in the SIWI

position. However, a reversal can be seen for the stops /d/ and /g/, the fricative /z/,

and the affricate /d5/. All three of these voiced sounds are now produced more

accurately in the SIWI position. Group 2:6 continued to show a much higher

percentage of correct production of the fricative /s/ in SFWF than in SFWW.

Groups 3:0-6:0. In the next group, 3:0, Figures 4-24 to 4-26 show that the

accuracy of the SIWW is practically level with the SIWI position for all sounds,

except /p/. This trend towards more equal accuracy of production for the two

positions can be seen in Figures 4-27 to 4-41 to continue for all groups and all

sounds from Group 3:6 to Group 6:0.

A t-test was performed using percentages of correct production of all

sounds for each of the three youngest groups to evaluate the difference between





























Group 2:0--Stops


1 no/.--,


80%-


60%


40%


20%-


0%


p d t g

I'siw" nslvww I


Figure 4-18. Comparison of accuracy of stops in SIWI and SIWW positions for
Group 2:0.


I


I


.....................


...............................


....... ---- ...........


..........................























Group 2:0--Nasals & Liquids


100%

S80%
0
060%
(D)
40%

20%
0%.
0%


ESIWI MSIWW MSFWWOMSFWF I


Figure 4-19. Comparison of accuracy of nasals and liquids in SIWI and SIWW
positions for Group 2:0.


I



















Group 2:0--Fricatives & Affricates

100%
"5

0
68 0 % ------------------- ................ ........... ......... ........ ....... .... ............ .................. ....... ........ .............. ...... ..............
0O 60%
N 0)
C:





lsIwI__sIWW 4SFWW0SFWFi
v f z s f 3 tf d3
Sounds
*MSIWI *MSIWW ESFWWERSFWF|


Figure 4-20. Comparison of accuracy of fricatives and affricates in SIWI and
SIWW positions for Group 2:0.







84












Group 2:6--Stops

100%


8j 0 % -....-........ ......... .................... ....
(D60%
g .





I 0%
b p d t g k

,SIWI *SWW I







Figure 4-21. Comparison of accuracy of stops in SIWI and SIWW positions for
Group 2:6.




Full Text
xml version 1.0 standalone yes
Volume_Errors
Errors
PageID P65
ErrorID 4
P398
4
P401
4
P416
4
P419
4
P422
4
P425
4