User Manual
User Manual
User Manual
We thank Sarra Ghazel for preparing the first version of this manual.
Table of Contents
1 Introduction.............................................................................................................................................1
2 Participants..............................................................................................................................................1
3 Data Used in the Database......................................................................................................................1
4 Articulation measures..............................................................................................................................1
4.1 GFTA................................................................................................................................................1
4.1.1 Scoring the GFTA (Percentile, Standard, And Raw Scores) ....................................................2
4.2 EAT..................................................................................................................................................2
5 Descriptive variables in the database......................................................................................................2
5.1 Descriptive Variables for the Children ............................................................................................2
5.1.1 ID..............................................................................................................................................2
5.1.2 Age............................................................................................................................................2
5.1.3 Gender.......................................................................................................................................2
5.1.4 Longitudinal..............................................................................................................................2
5.2 Descriptive Variables of Phonemes .................................................................................................2
5.2.1 Phoneme/cluster .......................................................................................................................2
5.2.2 Token.........................................................................................................................................3
5.2.3 Phoneme position......................................................................................................................3
5.2.4 Phonetic characteristics of phonemes and clusters...................................................................3
5.2.5 Order of Presentation................................................................................................................4
5.2.6 Sound files.................................................................................................................................5
5.3 Descriptive Variables of Articulation Quality..................................................................................5
5.3.1 Accuracy ...................................................................................................................................5
5.3.2 Change in accuracy...................................................................................................................5
5.3.3 Type Of Mis-articulation...........................................................................................................5
Appendix 1: Phonemes And Their Characteristics....................................................................................7
Language Development Database Manual 1 S. Ghazel and M. Sénéchal
1 Introduction
This database presents samples of accurate and inaccurate articulations of a variety of the English
phonemes as pronounced by English speaking children between the ages of 3.5 and 5 years. The
purpose of this database is to provide a research tool of accurate and inaccurate articulations of
particular phonemes in different word positions for researchers interested in early speech and language
skills. The database could also serve as a training tool for specialists and speech-language pathologists.
The database will also provide access to the articulation of the same words and phonemes at two times,
one year apart, for each of the 37 children who participated in the study in both years. This will allow
researchers and specialists to detect differences and progress in articulation over time.
2 Participants
110 children participated and specific permission for inclusion in the database has been obtained for 72
of these children (28 girls, and 44 boys). 37 of these children were follow up participants (i.e.
participated both in 2002 as well as in 2003). They were assessed twice, once in 2002 and again 2003,
while the rest of them (35) participated only in 2003. Year 1 children ranged from 42 to 55 months old.
While year 2 children ranged from 48 to 65 months old. Excluded from this sample were children
receiving speech therapy or for whom parents reported hearing difficulties.
3 Data Used in the Database
Data used in the database was collected for a study carried out in the summer of 2002 and 2003
examining early speech and language abilities. Experiments took place in the Ottawa-Carleton region,
where children were tested twice, with a one-year interval separating the two testing sessions. Children
were tested individually in their homes or in day care centres. Articulation measures administered
included the Sounds in Words Subtest of the Goldman Fristoe Test of Articulation (GFTA) and an
Experimental Articulation Task (EAT). The former test assessed children’s articulation of the consonant
sounds of the English language in initial, medial and final word positions and consonant clusters (in
initial word position only). The latter task assessed children’s articulation of the target phoneme /r/ and
three distractor phonemes, /b/, /p/ and /l/. In both tasks, responses were elicited by showing children
pictures or objects and asking them to name them. Their responses for both tasks were recorded on
minidisks and then evaluated and scored by a speech pathologist. These recordings were the source of
data for the sound files used in this database. Through hyperlinks, the database will provide access to
the audio recordings of the pronunciations of 62 words for each child (a total of 6624 words).
4 Articulation measures
4.1GFTA
The Sounds-In-Words Subtest of the GFTA was used to assess articulation quality. This test assessed
children’s articulation of all of the consonant sounds in the English language in the initial, medial, and
final word positions, as well as consonant clusters. Children were shown pictures and asked to name
them. A total of 39 phonemes and clusters were included in the GFTA test and 53 words were elicited
from each child (exp: wagon, ball, yellow…).
A maximum of 3 words were used to test for each particular phoneme, depending on whether the
phoneme exists in all three possible word positions (word initial, word medial, word final position).
For example, the words ball, rabbit, and bathtub were used to test for the articulation of the phoneme
/b/ when in initial, medial, and final position respectively. The phoneme /h/, however, was tested
through the word house representing the word initial position only, since this phoneme rarely exists
otherwise in English. All clusters were tested in word initial positions only. Some words were used
Language Development Database Manual 2 S. Ghazel and M. Sénéchal
more than once to test for different phonemes. For example, the word ball was used twice. Once to test
for the phoneme /b/ in initial position, and another time to test for the phoneme /l/ in final position.
4.1.1 Scoring the GFTA (Percentile, Standard, And Raw Scores)
The GFTA articulations were given raw, standard, and percentile scores. The raw score represents the
total number of errors each child makes (this falls in the range between 0 to 77). This number (77)
represents the total possible number of errors a child could make knowing that the sum of all phonemes
in all possible word positions (initial, medial, and final) adds up to 77. This score is then converted to a
standard score following the scale used in Goldman and Fristoe (2000). The latter provided age and
gender dependent tables where one could convert a raw score to a standard and/or a percentile score.
For example for any 2-year old female child who has a total number of 48 errors (raw score), the
standard score is 86 and the percentile is 26. For a male child in the same age and with the same raw
score, however, the standard score is 93 and the percentile is 42. For a slightly older female child (3
years old) with the same raw score (48), the standard score is 74 and the percentile is 14).
4.2EAT
In the experimental articulation task, children were administered 16 testing items. Two of the nine
testing items began with the phoneme /b/ and two with /p/, six began with /r/, while the rest (six) began
with the phoneme /l/.
Articulation of each target phoneme in word-initial position was tested by having children label
familiar objects (for example, articulation of the phoneme /r/ was tested for by having children label
things like rake, ring, and red). In both the GFTA and the EAT, children’s articulations were recorded
using a microphone and a minidisk recorder so that scoring of the task could later be verified.
5 Descriptive variables in the database
5.1Descriptive Variables for the Children
5.1.1 ID
Each child was given a fake ID number to protect the real identity. ID numbers in the 100s represent
follow up children, and those in the 500s represent the non-follow up children (who only participated in
year 2).
5.1.2 Age
For follow up participants age is described in months for year 1 (2002) and year 2 (2003). For non-
follow up participants, age was described in month for year 2 only (2003). This will allow researchers
and specialists to compare differences and progress in articulation across the different ages.
5.1.3 Gender
Described in the values male or female.
5.1.4 Longitudinal
Described with the values yes for those who participated in both year 1 and year 2 of the study and no
for those who participated in year 2 only.
5.2Descriptive Variables of Phonemes
5.2.1 Phoneme/cluster
Phonemes used in this database included all of the consonant sounds in the English language as well as
consonant clusters in the initial, medial, and final word positions. Appendix 1 includes all tested
Language Development Database Manual 3 S. Ghazel and M. Sénéchal
phonemes and clusters, their phonetic characteristics, the words that represent them, and their position
within each word). Below we describe each variable in the database and the values that that these
variables can take.
5.2.2 Token
For each phoneme or cluster, the corresponding word that was used to test for its articulation was
illustrated.
5.2.3 Phoneme position
Each phoneme was tested for all possible positions in the English language (initial, medial, final). The
phoneme /b/, for example, was tested in initial position through the word ball, in medial position
through the word rabbit, and in final position through the word bathtub. Some phonemes do not occur
in English in some positions. The phoneme /y/ as in yellow, for example, only occurs in initial position.
5.2.4 Phonetic characteristics of phonemes and clusters
Phonetic characteristics of each phoneme in the database were described according to 3 dimensions: its
voicing quality, its place of articulation, and its manner of articulation (O’Grady and Archibald, 2004).
One value is presented for each phoneme according to each dimension. Two values, on the other hand,
are presented for each cluster to describe the characteristics of each phoneme in the cluster (e.g.,
bilabial + alveolar, or stop + glide…).
For the GFTA data, and on the voicing quality dimension, 15 of the phonemes can be classified as
voiced, while 14 are voiceless. On the place of articulation dimension 4 of these same phonemes (that
can be either voiced or voiceless) are also bilabial, 7 alveolar, 2 labiodental, 3 velar, 1 glottal, 5 palatal,
1 labial velar, and 2 dental. On the manner of articulation dimension, 8 of these phonemes can also be
classified as stops, 8 fricatives, 2 glides, 2 liquids, 1 nasal, and 2 affricates.
For the EAT data, 3 phonemes were voiced while 1 was voiceless. On the place of articulation
dimension 2 of these same phonemes (that can be either voiced or voiceless) are also bilabial, 1
alveolar, 1 palatal. On the manner of articulation dimension, 2 of these phonemes can also be classified
as stops, and 2 as liquids.
Voicing and voicelessness
Phonemes can be voiced or voiceless. They are voiced when the vocal folds are brought together in a
way that is not tightly closed and when the air passing through them causes them to vibrate (/b/, /d/, /g/,
/l/, /m/, /n/, /η/, /r/, /v/, /w/, /z/, /dʒ/, and /δ/). A sound is said to be voiceless when it is produced with
the vocal folds being pulled apart in a way that air passes directly through the glottis (/f/, /k/, /p/, /s/, /t/,
/∫/, / t∫/, and /θ/).
Place of articulation
1. Bilabial
Sounds that are produced with closure or near-closure of the lips are said to be labial. Sounds that
involve both lips are said to be bilabial (/p/, /b/, /m/).
2. Alveolar
Sounds produced with the tongue being brought to the alveolar ridge behind the upper front teeth (/s/,
/z/, /t/, /d/, /l/, /n/)
3. Labiodental
Sounds involving the lower lip and upper teeth (/f/, /v/).
Language Development Database Manual 4 S. Ghazel and M. Sénéchal
4. Velar
Sounds made with the tongue touching the soft area towards the near roof of the mouth (the velum)
(/k/, and /g/).
5. Glottal
Sounds produced using the vocal folds as primary articulators (/h/).
6. Palatal
Sounds produce with the tongue being brought to the highest roof of the mouth called “palate” (/y/, /r/,
/∫/, / t∫/, and /dʒ/).
7. Labial-velar
Sounds produced with the tongue raised near the velum and the lips rounded at the same time (/w/).
8. Dental
Sounds produced with the tongue placed against or near the teeth (/δ/, and /θ/).
Manner of articulation
1. Stops
Sounds made with a complete closure in the oral cavity or the glottis (/b/, /d/, /g/, /p/, /t/, /k/, /m/, and
/n/).
2. Fricative
Sounds produced with a continuous airflow through the mouth accompanied by a continuous audible
noise because the air used in their production passes through a very narrow opening either at the glottis
or in the vocal tract (/f/, /h/, /s/, /v/, / ∫/, / z/, / δ/, and / θ/).
3. Glide
Very rapidly articulated non-syllabic segments made with the tongue raised and pulled back near the
velum and with the lips protruding or rounded (/w/, and /y/).
4. Liquid
Sounds articulated when air escapes through the mouth along the lowered sides of the tongue, with the
tongue raised to the dental or alveolar laterals (/l/, and / r/).
5. Nasal
Sounds produced through the nose with the mouth closed (/m/, /n/, and /η/).
6. Affricate
Composite sounds consisting of a stop and a fricative articulated at the same point (as `ch' in `chair' and
`j' in `joy', / t∫/, and / dʒ/).
5.2.5 Order of Presentation
Order of presentation describes (in numerical values) when during the test each phoneme was
presented. Since the GFTA and the EAT tests were conducted separately from each other, each group of
phonemes within the one test had its own order of presentation.
Language Development Database Manual 5 S. Ghazel and M. Sénéchal
Stopping
Occurs when fricative consonants as (/f/ /v/ /s/ /z/, /sh/, /zh/, /th/ or /h/), or affricates as (/ch/ or /j/) are
replaced by stop consonants as (/p/ /b/ /t/ /d/ /k/ or /g/) (e.g. /f/ in funny is pronounced as /p/, and /j/ in
jump is pronounced as/d/).
Gliding
Occurs when liquids as /l/ and /r/ are replaced by /w/ or /y/ (e.g. real is pronounced as weal, and leg is
pronounced as yeg).
Assimilation
Occurs when the pronunciation of the whole word is influenced by the presence of a particular sound in
the word (e.g. "Cupboard" is pronounced as pubbed
dog is pronounced as gog).
Syllable reduplication
Occurs when a syllable is doubled often because of the difficulty in coordinating two distinct
articulations in the same word (e.g. the word water is pronounced wawa, or bottle pronounced baba).
Syllable reduction
Occurs when a child omits a weak syllable (unstressed syllable) such as the second syllables in the
words telephone and tidying (telephone is pronounced as teffone, and tidying is pronounced as tying).
Language Development Database Manual 7 S. Ghazel and M. Sénéchal
References
Bowen, C. (1998). Developmental phonological disorders. A practical guide for families and
teachers. Melbourne: ACER Press.
Creaghead, N., A., Newman, P., W., & Secord, W., A. (1989). Assessment and remediation of
articulatory phonological disorders. Columbus, OH.: Merrill Publishing Company Press.
Goldman, R., & Fristoe, M. (2000). Goldman Fristoe 2 test of articulation. Circle Pines, MN.:
American Guidance Service, Inc.