topics: |  cognitive | language-acquisition-infant | developmental | phonology

An absolute classic!

Excerpts

Infants are not ready for speech. [in general, motor development does not mature rapidly in humans]. also, relative structure of the oral cavity is significantly different - resembles primates in that the curve of the oropharyngeal canal is gradual (and not 90 deg like in adults).

"The larynx comes up like a periscope and engages the nasal passage, forcing the infant to breathe through the nose and making it anatomically possible to drink and breathe at the same time." [Pinker 94].

Tongue is a stubby mass filling up the mouth w only limited movement possible. [Kent/Murray 82] Also infant doesn't control breathing yet. 15-16

At three months, the soft palate comes lower - child can now close off the nose. tongue is lengthened and its musculature is more developed, and opening of pharynx permits it to move from front to back. Child gains control of respiratory cycle. by five months, control of phonation has been acquired [Koopmans, van Beinum, and van Der Stelt 1979].

articulatory control - the whole machine - tongue, lips, pharynx, larynx - takes longer. 2nd half of 1st year. But full adult capability comes only at 5-6 years. This is why french children keep saying obelisk instead of obelix. 17

Listening: Sound preference experiments

Einar Siqueland and Clement de Lucia 1969 --> use the only behaviour mastered
by the newborn - enthusiasitc sucking.

    --> High-amplitude sucking (HAS), usually nonnutritive.

A rubber nipple is instrumented, and frequency of sucking activity is used
to produce sounds.  If familiarized to the sound, sucking falls off.  If a
novel sound is introduced, sucking increases. p.19

[Eimas, Siqueland etal 1971] : found 4-mo babies are sensitive to /ba/ vs
/pa/ - boundary is close to that used by adults to distinguish these
phonemes.  Sucking increased for a change, but not for a differnt type of p,
although voicing lag between /p/ and /p'/ was of the same order as /p/ and
/b/. - p.21

 

Since then, dozens of expt - w babies as young as 3/4 days
old - show that infants can distinguish almost all phonetic contrasts found
across all natural languages - see [Jusczyk 1985).  [most adults can't] p.20

prosody: infants only a few days old prefer to listen to the voice of their
mother in contrast to another mother talking to her baby.  However, the
intonation must be natural - isn't sensitive to the voice played backward.
Thus, sensitivity is not to static elements of sounds.  [Mehler Bertoncini
etal 1978].

Also, they are sensitive to their mother's language.  Babies 4 days are
presnted sequences of French speech and Russian speech, by same bilingual
woman.  They show stronger sucking when French follows Russian, and not vice
versa.  The preference is maintained when the phonetic info is removed
leaving only intonations.  p.22 [language originates in song.]

Effects of Prenatal exposure

Physicisans long dismissed as maternal imagination the observations of
mothers who felt the fetus react to sharp noises and jump at the sound of a
loud telephone ring.  It is now known that the child's auditory syustem is
functional from the 25th week of gestation and its level of hearing
approaches that of an adult by the 35th week.

What sounds does it hear?  Hydrophone recordings in a pregnant mother's
uterus show that the mother's internal sounds (respiratory, heartbeat and
gastrointestinal) are typically lower frequencies, and the mother's voice is
heard almost as loudly as ex utero.  Other sounds can also be detected,
prosody is particularly well preserved. [Querleu, Renard Versyp 1981] Even ~
30% of phonemes can be recognized from the recordings by adults. 22-23

For in-utero testing, can monitor heartbeat.  Lower heartbeat --> increased
awareness.  Speaker positioned 20cm from mother's abdomen, repetitive sound till fetus
habituates to it.  Initial arousal at the beginning causes cardiac
deceleration, but this finally disappears and resumes normal rhythm.  If,
after habituation, the sound is changed,  a new round of deceleration is
observed  [Lecaunet et al 1987]. 24

[Lecaunet and Granier-Deferre 1993] : fetuses between 36 and 40 weeks.
Sixteen repetition of disyllable [babi] - habituated.  then presented with
[biba] - results in decleration.  tested in a state of calm sleep. [of the
baby? or mother?]  Thus the two sequences were distinguished.

Thus, exposing fetuses to their mother's language before birth favours
perceptual adjustment to the phonetic and prosodic parameters that
characterize this lg and diff it from others.

[deCasper and Fifer 1986]: have mother read a prose passage in a loud voice
  several times during last 6 weeks of pregnancy.  Then within 12 hours of
  birth, they conduct an expt using a finer HAS test: sucking fast - one
  reaction, sucking slow, another.  babies can then choose the stimuli
  through sucking at a particular rate.  They show a clear preference for the
  passage read by the mother before birth, even if it was read by another
  woman during the test.  The fetus therefore seemed responsive to the
  general acoustic properties of the speech and not just the voice /
  intonations of the mother.

[deCasper and Spence 1986]: in utero expt, 33d wk: asked mother to read a
poem in a loud voice every day for four weeks.  At end, (37th wk), fetus
listened to this poem, in alteration w another novel poem, both recorded on a
speaker by a diff voice.  speakers positioned near the head of the fetus.
Heart bets systematically decreased only in response to the poem read by
mother earlier. 25-6

Phoneme and syllable recognition

By 5 months, infants can neglect effects of speaker (bass/high/gender),
accent, diff contexts - can recognize /a/ sound.  [Kuhl 1983].

Infants recog syllable (anyhow, syllables are the units of perception, see
[massaro-1996] in [dijkstra-smedt-1996]).

2-months: form inventory of familiarized syllables - bu is diff from
bo,ba,be, but also from du.  Thus they did not habituate to the /b/-initial
sounds, but to the syll as a whole.  [Juscyzk and Derrah 87]

sentence sensitivity: 2mos : detect change of phonemes more effectively when
they occur in short sentences [rat/cat chases the white mouse) rather than in
a list of words. [rat/cat king idle stop] -->  effect of prosody.

Preference to own name

Mandel Jusczyk Pisoni 1995: 4.5 mos - prefer sound of own name compared to
those of their friends.
  [spkrs to left and right, with lights at top.  when baby gazes towards a
  light, it makes a sound.  thus she can indicate preference through gazing. 28-9

Brain structure

sound structures are specialized in the left hemisphere whereas the right is
more relevant to longer term sounds - lesions there affect prosody and music.
In the left, the articulatory system is controlled by the Broca's area in the
sylvan sulcus, and the Wernicke's area in the temporal lobe has a role in
comprehension.  Without it, one produces sentences, but these are
incomprehensible.  Without Broca's one can hardly produce sounds.

The RH matures faster than the Left Hemisphere; this maturational discrepancy
leads to differing skill sets.

In 65% the LH area including Wernicke's is larger.

But is lateralization genetically determined?  Perhaps not.  infant brains are
plastic.  If even LH-lobotomy
needs to be performed, if it is before age 1, then there is complete
recuperation including language ability.

However, in normal brains, the LH is specialized to handle langauge, and only
profound changes will result in alternate mechanisms emerging.

Which hemisphere is more active?  Can investigate using dichotic listening -
diff auditory inputs in diff ears.  Although somewhat controversial, [Entus
1977] expted with (2-3mos): music in one ear and sound in the other.  After
habituation, change of speech sound in the right ear (and music on the left
ear) had a greater effect.  These results have been reproduced more often than
not - and appear to indicate that the LH does a better job of discriminating
between speech sounds and the RH on music.

[Dehaene-Lambertz and Stanislas Dehaene 1994]: EEG potentials: 3mos infants
can very rapidly (within 400ms after onset) detect changes in the first
consonant of a syllable.  The ERP responses prefer the LH slightly, but
considerable indiv variations.

2 Emergence of Speech

The newborn comes into the world crying.  Prodn of sounds is constant in
humans, from the first cry to the last breath.

First two months: reactional sounds signalling well-being or discomfort.  But
infants are extraordinarily attentive to speech [observns by Antoine Gregoire
1937] - imitate mouth shapes - listen attentively

2-5 months: sounds while lying down.  arrheu and ageu - from larynx and
soft palate.  by 4th-5th month begin varying intonation.  Then vocalizations
become progressively more voluntary. 38

By 6th month they can interrupt their vocalization, control phonatory and
supraglottal adjustments.  Also tunes the pitch of their vocalizations; their
voice is pitched higher when with mother than w father.  39

between 4 and 7 months - articulatory games - like practising scales -
quasi-consonants [aw:a], [abwa], [am:am] as well as isolated vowelsthat are
modulated.

    As Shunzya Suzuki wisely observed in connection with problem solving,
    there are many possibilities in the mind of the beginner but only a few in
    the mind of the expert.
To become an expert in the native lg, the child has to select the right
movements to make and select the right sounds to listen to. 40

[Shunryu Suzuki]: Sōtō Zen priest. San Francisco Zen center [w]:
In the beginner's mind there are many possibilities, but in the expert's
there are few.

4: Discovering the meaning of words

Christophe, Dupoux, Bertoncini Mehler 1995
studied how newborns responded to the same two syllables (e.g. "mati"), one
appearing as a disyllable in a single word (eg mathematicien), and the
other an intraword containing
a word boundary (eg schema tigre).   Newborns 3 days world, under the
non-nutritive sucking paradigm, were seen to suck harder for a novel stimulus
(intraword) after habituation (disyllable).

At 5 months, children show preference for stories with pauses at clause
boundaries than within the phrases:

A. Cinderella lived in a great big house // but it was sort of dark //
   because she had this mean mean stepmother.

B. Cinderella lived in a great big house but it was // sort of dark
   because she had // this mean mean stepmother.

A is preferred to B.

[Paradigm: The child looks at the speaker for a light to come on and the
sound to go on.  If he looks away for 2 seconds, light goes off and sound
stops.  So he learns to look if he wants it to continue. ]

English prosody: stress is mostly on content words, and within the word on
the root syllable.  Babies 3 mos have no preference, but by 6 months they are
more interested in sounds with a STRONG-WEAK stress sequence.  On the other
hand, Hebrew is the opposite, with a iambic stress, and there by 6 mos babies
are more interested in a WEAK-STRONG stress pattern.
[Jusczyk Cutler Redanz 1993]

By 9 months, children are sensitive to the phonotactic structure of their
language.  e.g. Dutch words like zwetsen and vlatke would not be possible in
Engl where [zv] and [vl] are not licensed.
Infants of 6 months did not discriminate between Engl and Dutch, but Engl
babies of 9mo listened more (8 sec) to the Engl than the Dutch (5 sec)

Combining Phonotactic with prosodic organization, mother tongue preference is
evident by 6 mos.

[Myers etal 1996]
Pauses inserted in stories, either at word boundary or intra-word.  By 11
months, children show a clear pref for boundary to intra-word.

[Morris Lewis 1936]
psycholinguist, kept diary
8-9 mo: son waved when someone said goodbye.
At 7 mo, baby studied by Boysson-Bardies clapped his hands when one said
bravo to him, and refrained from touching an object when one whispered
"hush".
Henri plays the "no" game - he doesn't quite touch the forbidden plant,
waiting for the no that isn't long to come...111-2

He would have responded the same if yes was uttered in the same tone.  So
separating the phonetic component from the prosody remains a challenge for
studies of word recognition.

Between 11-13 mos, memory of sounds - names assoc with unknown objects -
arises.
Sharon Oviatt 1980 calls it "Recognitory comprehension" 121

Augustine quote 122

-Review: Leslie Nabors Olah

		Harvard Educ http://www.hepg.org/her/abstract/134:

An incredibly detailed account of developmental phonology in the first two
years of life. De Boysson-Bardies has made her career by exploring the
earliest language acquisition — the phonology and composition of babbling by
infants seven to twelve months old. She believes that "children’s processing
of language is initially more acoustic than linguistic" (p. 129). At this
early age, a main question "has to do with the reality of innate mechanisms"
(p. 9). 

De Boysson-Bardies modifies the Chomskian notion of innate knowledge
by using Darwin’s term instinctive tendency to describe the "program of
acquisition that develops on the basis of potentials inscribed in the genetic
code of the child" (p. 7) at this very early stage in development. Once the
child can segment speech sounds, however, the author ascribes more agency to
the young language learner. This development in perceptive abilities, in de
Boysson-Bardies’s mind, lays the earliest foundation for language
development. The discrimination of vowels and consonants, for example, is
used by the ten- to twelve-month-old to register differences in word meaning,
such as the difference in English between pat and bat. Further exposure to
language-specific sounds in turn helps the child eventually become proficient
in language production.

De Boysson-Bardies’s section on prenatal linguistic development is not only a
superb review of the research, but it is also a simply fascinating read.