WHEN ARE SPEECH SOUNDS LEARNED?

Eric K. Sander
Cleveland State University, Cleveland, Ohio

The age at which specific consonant articulations emerge differs greatly among
children. Widely quoted age tables represent upper age limits rather than average
performance, and stress-sound mastery in all word positions as opposed to customary
usage. A new graphic way of summarizing information about age in relation to
consonant development is proposed.

When do children acquire the various speech sounds of our language? Con-
sidering the wealth of schedules in such areas as motor behavior, physical
~ o w t h , and cognitive skill, the question might be assumed to be an easy one
for the student of child behavior to answer. Surprisingly, however, a concise
reply to the question is difficult to provide, for reasons that have only partly
to do with the available research knowledge.

AVERAGE VS N O R M A L AGE LIMITS

One area of confusion arises from the fact that most normative accounts of
articulatory development stress the developmental ages at which a child's
failure to produce a given speech sound ought to concern a speech clinician or
a parent; they are not portraits of average performance.
T h e first large sample investigation of the articulatory status of U.S. children
at various preschool ages, by Wellman, Case, Mengert, and Bradbury (1931),
established norms reflecting not median performance but rather the ages at
which 75% or more of the children produced a sound correctly. Poole's widely
quoted 1 articles (1934, 1998) carried the Wellman logic one step further by
listing the ages at which correct articulation of specified speech sounds was
achieved by all (100%) of the children tested.
T h e pioneer reports by Wellman et al. and by Poole were updated when
T e m p l i n (1957) published additional articulatory test findings from 480
children. Before assigning an earliest age level to a sound, T e m p l i n insisted

1Widely quoted because Poole was the first to attach a single age level to an entire sound
category. Wellman et al. were quite content merely to list the ages at which a required per-
centage of children produced a sound correctly in a specified word context.

55

Downloaded From: http://jshd.pubs.asha.org/ by a Universite Laval User on 05/09/2016

Terms of Use: http://pubs.asha.org/ss/rights_and_permissions.aspx
 56 JOURNAL OF SPEECH AND HEARING DISORDERS -- XXXVII, 1

that three-fourths of the children of a particular age produce that sound cor-
rectly in all positions of a word (initial, medial, and final).
These widely publicized group standards reflecting upper age limits rather
than average performance are a recurring source of misunderstanding in dis-
cussions of articulatory development. A second cause of confusion is that most
authors who wish to speak at less than monographic length about articulatory
development fail to distinguish different indices of speech-sound achievement.

MASTERY VS C U S T O M A R Y PRODUCTION

A child is usually said to have mastered a consonant when he produces the

sound correctly at three different positions (for example, the /p/ in pie, open,
and soap) in words in which a vowel follows and/or precedes the consonant.
Yet this arbitrary definition is only one of several possible ways of measuring
progress in sound mastery. At the upper end of the continuum one might
rigorously assess sound mastery in terms of consonant blend, as well as non-
blend, contexts. Correct production of the ]s/ in a word such as scratch, for
example, usually involves a skill more advanced than that needed for sun.
However, even a child who articulated ]s] correctly in every observable con-
text might still conceivably have climbed short of the highest rung of the
mastery ladder. If, for instance, such a hypothetical child were also heard
substituting see for she, thereby confusing one sound with the other, would one
not be compelled to state that productively, at least, he had not yet adequately
differentiated or delimited the [s/ as a phoneme category?
At a lower end of the continuum, long before a child reaches a final stage of
sound mastery, it is possible to distinguish at least three earlier developmental
landmarks: (1) age of first appearance of the sound form, (2) age of earliest
correct articulation in words, and (3) age of customary production.
T h e sound form may occur first in the child's early babbling, or in his in-
correct utterances of words. Irwin (1957), for example, reports that infants
produce babbled approximations of many English consonants months before
uttering their first words. And a child who later says, "Thit," but means "sit"
has produced a th sound but in an incorrect context.
T h e earliest age at which a sound is correctly articulated in a word would
seem an important learning milestone. Practically, however, gathering de-
pendable data even for a single child on the first correct appearance of some
two dozen different consonants would be a formidable task. More importantly,
the mere isolated appearance of a sound in a correct context is not evidence
of substantial learning. Given patient listening, even persistent misarticulators
may be overheard uttering their "error sound" correctly in one context or
another.
When one says that a child has acquired a sound, the definition most easily
conveyed is that the child is producing the sound correctly most of the time.
Customary production refers to that point when a child is producing a sound
correctly more often than he is misarticulating or omitting it. Ideally, this

Downloaded From: http://jshd.pubs.asha.org/ by a Universite Laval User on 05/09/2016

Terms of Use: http://pubs.asha.org/ss/rights_and_permissions.aspx
 SANDER: WHEN ARE SPEECH SOUNDS LEARNED? 57

ought to be assessed by listening to a sample of the child's spontaneous spoken

language, not a predetermined word list.
Although this concept of customary production as opposed to mastery pro-
vides a~ basis for assigning newly learned sounds to age profiles, the
often-quoted summations of sound development remain, instead, traditionally
stringent and easily misinterpreted. Examine a classic schedule of sound de-
velopment (in this instance, Templin's revision of Poole) arrived at by match-
ing consonant sounds with the earliest ages at which they are correctly produced
by 75% or more of the children in each of three different word positions:

Age Consonant Sound

3.0 m, n, ng (as in going), p, f, h, w
3.5 y (as in yellow)
4.0 b, d, k, g, r
4.5 s, sh (as in shirt), ch (as in chew)
6.0 t, 1, v, th (as in think)
7.0 th (as in this), z, zh (as in pleasure), j (as in job)

According to this, t h e / t / is not learned until age six, t h e / s / is mastered at

age 4.5, and so on. Such generalizations are valid only if qualified by Templin's
demanding criterion of what it means to learn a sound. T h e ]t[ is somewhat
misleadingly placed at age six only because children younger than six failed
to articulate it clearly in the medial position of words such as skating or outing
(where t h e / t / is seldom unvoiced anyway). Not one child among Templin's
480 children failed to produce It/ correctly in the initial position of a word.
As a matter of fact, even the final / t / was said clearly by three-fourths (45 of
60) of the three-year-olds (the youngest age level tested).
As f o r / s / , it is usually classified among the later sounds to be acquired by a
child, perhaps because Poole (1934) did not assign its final mastery until age
7.5. Using Templin's 75% criterion, / s / is placed at 4.5 years. Yet well over
half of Templin's three-year-olds produced the sound correctly at each of three
word positions. Among children a half-year older (3.5 years), 51 of 60 articu-
lated the initial /s/, 49 of 60 the medial /s/. Yet age placement of /s/ is de-
layed until the 4.5 year level, where 46 of 60 children (75% or more) at last
produce the sound correctly in a final position. Templin's findings for ]t/ and
/ s / are presented in Table 1.
T h e point of all this is merely to emphasize by way of example the extremely
arbitrary nature of the age norms that have been set up for consonant develop-
ment. Change the definition of mastery to one of customary production (which
might be defined in this instance as correct articulation in two of the three
word positions) and placement of / t / descends from age six to age three.
Change the group percentage criterion for the age placement of a sound and
/ s / may be assigned anywhere along the total age range (3-8) studied by
Templin. T h e average age at w h i c h / s / is acquired (50% correct performance
at each of three positions) turns out to be a surprisingly early age three; but
insist that 90% (in this case 54 of 60) of the children articulate it consistently

Downloaded From: http://jshd.pubs.asha.org/ by a Universite Laval User on 05/09/2016

Terms of Use: http://pubs.asha.org/ss/rights_and_permissions.aspx
 58 JOURNAL OF SPEECH A N D HEARING DISORDERS -- XXXVII, 1

TABLE I. Number of children (out of 60) at various age levels correctly articulating / t /
a n d / s / in three different word positions (from Templin, 1957).

Sound Chronological Age (in Years)

3.0 3.5 4.0 4.5 5.0 6.0 7.0 8.0

Initial/t/
(toes, table) 60 60 60 60 60 60 60 60
Medial/t/
(skating, outing) 16 16 20 17 19 53 50 50
Final/t/
(bat, skate) 45 41 45 45 46 51 56 57
Initial/s/
(seat, silk) 42 51 46 48 47 46 55 58
Medial/s/
(myself, glasses) 39 49 46 45 47 43 53 57
Final/s/
(mouse, grass) 35 41 44 46 46 47 53 56

(not an unreasonable limit of normal performance) and the age placement

soars to eight years.

VARIABILITY IN EMERGENCE OF ARTICULATION

This brings us to a third reason it is often difficult to assign specific ages to

specific sounds: while it is possible to predict rather precisely an acquisition
age for a variety of behaviors during the first few years of a child's life, the
v a r i a b i l i t y a m o n g c h i l d r e n in the ages at w h i c h they successfully p r o d u c e
specific sounds is so great as to discourage p i n p o i n t statistics.
T a b l e 2 charts as a f u n c t i o n of age (in m o n t h s ) the percentages of c h i l d r e n
e x h i b i t i n g three different types of vocal behavior: l a u g h i n g aloud, s h o u t i n g
for a t t e n t i o n , a n d saying one w o r d clearly. O n the basis of these findings,
collected by Griffiths (1954), the average (median) child first laughs a l o u d
a r o u n d the f o u r t h m o n t h , first shouts for a t t e n t i o n a r o u n d eight m o n t h s , a n d

TABLE 2. Percentage of children at various ages who have laughed aloud, shouted for atten-
tion, and said one clear word (from Griffiths, 1954).

Vocal Chronological Age (in Months)

Behavior 1-2 3-4 5-6 7-8 9-10 11-12 13-14 15-16 17-18

Laughs
Aloud 0 35 98 100 100 100 100 100 100
Shouts for
Attention 0 0 4 37 84 94 98 100 100
One Clear
Word 0 0 6 28 57 90 94 98 100

Downloaded From: http://jshd.pubs.asha.org/ by a Universite Laval User on 05/09/2016

Terms of Use: http://pubs.asha.org/ss/rights_and_permissions.aspx
 SANDER: WHEN ARE SPEECH SOUNDS LEARNED? 59

says his first clear words at nine or ten months. Not much danger of misinter-
pretation here.
Now compare these findings with what occurs when one attempts to chart
the specifics of articulatory skill. Templin, unfortunately, studied no children
below age three, by which time they had already averaged 60% correct produc-
tion of the consonant elements. A few sounds, however, including the voiced
and voiceless th and the / v / , 2 were still said incorrectly by the majority of
three-year-olds. Correct production of these sounds (initial position) as a
function of age (in years) is charted in Table 3.

TABLE3. Percentage of children at various ages correctly articulating initial voiced th, initial
/v/, and initial voiceless th (from Templin, 1957).

Chronological Age (in Years)

Sound 3.0 3.5 4.0 4.5 5.0 6.0 7.0 8.0

Initial/~5/
(those, there) 23 32 57 60 62 83 97 98
Initial/v/(valentine,
vacuum cleaner) 12 30 40 47 55 80 88 98
Initial/o/
(thinner, thumb) 27 28 48 60 67 85 90 97

Viewed against the uniformity in time of appearance of laughter, shouting,

and talking, children display an amazing variability in emergence of articulatory
skill. For example, it takes not four months but four years to move from an age
level (three) where 23% of the children are making the initial voiced th to a
point (age seven) where 97% of them are doing so. In contrast (Table 2), no
children evidence laughter at two months, yet nearly all of them do so at five
months. T h e observed percentage jumps from zero at 1-2 months to 98 at 5-6
months.
Similar comparisons can be made for the other behaviors in Tables 2 and 3.
Within a span of four months (5.5 to 9.5) the percentage of children shouting
for attention climbs from four to 84; in contrast, it takes four years (from
3.0 to 7.0) to move from 12% to 88% correct production of the initial / v / .
Whereas only 28% of children have said their first words at 7.5 months, 90%
have done so at 11.5 months. Again by way of comparison, one must wait four
years (from 3.0 to 7.0) for a similar rise (from 27% to 90%) in the correct
production of initial voiceless th.

S U M M A R I Z I N G AGE I N F O R M A T I O N

Clearly, dangers of misinterpretation greet any attempt to convey in a single

2Templin did report correct production o f / v / in the medial position (as in driving) by
36 of 60 three-year-olds. For two-year-olds, Wellman et al. observed no correct productions of
initial/v/, 7% correct articulation of the initial voiced th, and 20% correct articulation of
initial voiceless th.

Downloaded From: http://jshd.pubs.asha.org/ by a Universite Laval User on 05/09/2016

Terms of Use: http://pubs.asha.org/ss/rights_and_permissions.aspx
 60 JOURNAL OF SPEECH A N D HEARING DISORDERS -- XXXVII, 1

table the essential information about articulatory development. Given the ex-
pected variability among children, tables with average ages of customary
consonant production will send some needlessly anxiqus mothers scurrying to a
speech clinician. To insist that three-fourths or 100% of all children articulate
a sound in all word positions before assigning it to an age level tends to
promote equally erroneous interpretations of normal sound acquisition. This
dilemma may be easily resolved, however, by including in the age summaries
of consonant development both an average age of customary production and
a measure reflecting the traditional upper age limits for sound acquisition.
Since the most significant progress in articulatory development takes place
between the ages of one and three, age norms ought to arrange for a stop or two
spaced along this busy route. Bullowa, Jones, and Duckert (1964), for example,
provide a fascinating narrative of one child's progress (from 11 to 22 months)
in articulating the word shoe. At 13-14 months, report Winitz and Irwin
(1958), only 16% of the words in an average child's limited repertoire are
free of phonetic error. T h e percentage of phonetically accurate words rises
to 24 at 15-16 months and to 38 at 17-18 months.
By age two, the average child has sufficient vocabulary and intelligibility to
allow specific testing of a variety of consonants. Published data on the articula-
tion of two-year-olds comes mainly from Wellman et al. (1931), who, re-
grettably, tested only 15 (probably brighter-than-average) children, not all of
them completely. (Future investigations at this age level, using more children,
might alter these pioneer findings somewhat, although probably not greatly.)
By including data from the Wellman et al. two-year-olds in addition to that
collected by T e m p l i n at later ages (three through eight) a tentative table of
average age estimates for customary consonant production may be constructed.
Customary production of a sound might reasonably be assumed if a majority
of children articulate it clearly in at least two of three word positions. One
method, then, of translating previous research findings into a table of average
ages of customary consonant articulation would be to assign each sound to the
earliest age at which it is correctly produced in two positions by 5I% or more
of the children. Although these calculations are in most cases easily accom-
plished, many sounds are testable in only two different word positions. T o
standardize matters, each sound has therefore been placed at an age level
where the combined test average at the various word positions exceeds 50%
correct production. 8 Sounds assessed at age two (the earliest age level tested)
and showing a combined average exceeding 70% correct production have been
somewhat arbitrarily placed in a "before age two" category.
For convenience, and to avoid a misleading sense of precision, fractional age
norms have been disregarded. As added information, each sound is followed

3For example, although 53% of Templin's three-year-olds correctly produced /" or /d3/ in
the initial position, the comparable percentages in the medial and final positions are 30% and
22%, respectively, resulting in a combined average of 38%. The minimum combined average
for age placement is 51%. The sound is therefore placed at age four where the combined
average is 69%.

Downloaded From: http://jshd.pubs.asha.org/ by a Universite Laval User on 05/09/2016

Terms of Use: http://pubs.asha.org/ss/rights_and_permissions.aspx
 SANDER: WHEN ARE SPEECH SOUNDS LEARNED? 61

in parentheses by the percentage of children of that age correctly articulating

it in the initial (for ng o r / r j / , the final; for zh o r / 5 / , the medial) word posi-
tion. (The "before age two" percentages are, of course, for two-year-olds.)

Age Consonant Sound and Percentage

Before 2 h (87), m (87), n (100), w (79), b (93), p (87)
2 t (80), k (66), g (57), ng (60), d (87)
3 f (88), y (70), s (70), r (58), 1(67)
4 ch (72), sh (75), j (85), z (62), v (40)4
5 voiceless th (67), voiced th (62)
6 zh (72)

These are assumed average ages of customary consonant production. Most

of these sounds are first spoken correctly in m a n y words at earlier average ages.
Also, at years later than those indicated, the average child continues to mis-
articulate some of these sounds in certain contexts.
A table of average age estimates alone is hardly less subject to misunder-
standing than one listing upper age limits. Figure 1 summarizes in a form most
appropriate for parents and professionals both the average ages and the upper
(90%) age limits of customary production for each of 24 consonants. T h e solid
bar corresponding to each sound starts at an age level where the combined test
average at the various word positions exceeds 50% correct production. ~ T h e
bar stops when the combined test average reaches 90% correct production. 6
T h e length of the bar thus indicates in rough fashion the extent of variability
among children in their ages of acquisition of specific consonant articulations:
the longer the bar, the greater the variability. As a group, children tend to vary
less in acquisition age for those sounds that are mastered earliest. Greatest
age variability occurs on the / s / consonant.
Use of Figure 1 avoids the pitfalls of attaching a single acquisition age to
each consonant. Unlike previous summaries, it depicts at its starting points
median or average performance; however, it also preserves in graphic form the
complementary view that normality in articulatory development encompasses
an impressively broad age range.

ACKNOWLEDGMENT
Requests for reprints should be addressed to Eric K. Sander, 915 Nelaview Road, Cleveland,
Ohio 44112.

4At age four,/v/received a combined average of 51% despite the low initial position per-
centage. Three-fourths of Templin's children of that age produced it correctIy in the medial
position.
5In constructing this figure, use was made of the available fractional age norms. (Mainly for
convenience of memory, fractional age norms were disregarded in the previous table of average
ages.) In addition, / f / and / y / were interpolated at 2.5 years, since they showed combined
age-three averages exceeding 70% correct production. Consonants whose averages exceeded
70% at age two were extrapolated downward to 1.5 years.
6The zh o r / 3 / sound received a combined age eight average of 86%; its upper limit has
been extrapolated to 8.5 years.

Downloaded From: http://jshd.pubs.asha.org/ by a Universite Laval User on 05/09/2016

Terms of Use: http://pubs.asha.org/ss/rights_and_permissions.aspx
 62 JOURNAL OF SPEECH AND HEARING DISORDERS -- XXXVII, 1

AGE LEVEL
3 + ~ G ? 8

[ Em

Figure 1. Average age estimates and upper age limits of customary consonant produc-
tion. The solid bar corresponding to each sound starts at the median age of customary,
articulation; it stops at an age level at which 90% of all children are customarily pro-
ducing the sound. (From Templin, 1957; Wellman et ah, 1931.)

Downloaded From: http://jshd.pubs.asha.org/ by a Universite Laval User on 05/09/2016

Terms of Use: http://pubs.asha.org/ss/rights_and_permissions.aspx
 SANDER: WHEN ARE SPEECH SOUNDS LEARNED? 63

REFERENCES
BULLOWA, M., JoNrs, L., and DUCKERT, A., The acquisition of a word. Lang. Speech, 7, 107-111
(1964).
GRIFFITHS, R., The .4bilities of Babies. New York: McGraw-HiU (1954).
IRWIN, O. C., Phonetical description of speech development in childhood. In L. Kaiser (Ed.),
Manual of Phonetics. Amsterdam: North Holland (1957).
POOLE, I., Genetic development of articulation of consonant sounds in speech. Elem. Eng.
Rev., 11, 159-161 (1934).
POOLE-DAVIS, I., The speech aspects of reading readiness. Yearb. Dept. elem. schl. Princ., natl.
educ..4ss., 17, 282-289 (1938).
TEMPLIN, M., Certain Language Skills in Children. Minneapolis: Univ. Minnesota (1957).
WF.LLMAN, B., CASE, I., MVNGERT, I., and BRADBURY, D., Speech sounds of young children.
Univ. Iowa Stud. Child Well., 5, 1-82 (1931).
WINITZ, H., and IRWIN, O. C., Syllabic and phonetic structure of infants' early words. J. Speech
Hearing Res., 1, 250-256 (1958).

Downloaded From: http://jshd.pubs.asha.org/ by a Universite Laval User on 05/09/2016

Terms of Use: http://pubs.asha.org/ss/rights_and_permissions.aspx