PSYB57H3 Chapter Notes - Chapter 9: Vocal Folds, Speech Perception, Speech Segmentation
Chapter 9: Language
The Organization of Language
•Language involves a special type of translation, translating ideas into sounds and sounds into ideas
and this occurs because language relies on consistent and well-defined patterns
•Language structure: sentence—> phrase—> word—> morpheme—> phoneme
•Highest level being ideas intended by the speaker or the ideas derived by the listener from the input
•These ideas are typically expressed in sentences—coherent sequences of words that express the
intended meaning of a speaker
•Sentences, in turn, are composed of phrases (e.g., The umpires, talked to the players, which are
composed of words
•Words are composed of morphemes, the smallest unit of language units carrying meaning (The,
umpire, s, talk, ed, to, the, play, er, s). Some morphemes are “free morphemes”, such as “umpire” or
“talk” can stand alone and typically refer to particular objects or ideas or actions. Other morphemes get
“bound” onto free morphemes to add information (e.g., ed and s)
•When morphemes are pronounced, the units of sounds are called phonemes, the smallest unit of
sound that can serve to distinguish words in language (correspond to how actual sounds are produced,
regardless of how these sounds are expressed in ordinary writing)
Phonology
The Production of Speech
•Air flows quietly out of the lungs, through the larynx and up through the nose and mouth
•Noise is produced by altering or interrupting airflow, allowing humans to produce different sounds
•For example, within the larynx are two vocal folds (vocal chords) that can be rapidly opened and closed,
producing a buzzing sort of vibration known as voicing
•Sounds are also produced by narrowing the air passageway within the mouth itself, such as placing the
tongue’s tip near the roof of the mouth just behind the teeth to produce the [s] sound
•We can distinguish sounds according to how the airflow is restricted; this is called the manner of
production
•We can also distinguish between sounds that are voiced (produced with the vocal folds vibrating, such
as [v], [z], and [n]) and sounds that are unvoiced, e.g., [f], [s], [t], [k]
•Sounds can also be categorized according to where the airflow is restricted, called the place of
articulation
•Lips are closed for “bilabials”, top teeth close to bottom lip for “labiodentals”, and tongue behind
upper teeth for “alveolars”
•Phonemes are created by simple combinations of the features above, and the English language has
about 40 different phonemes
The Complexity of Speech Perception
•Speech is fast, normal speech consisting of about 180 words per minute—about 15 phonemes per
second—but people can follow speech as fast as 250 words per minute
•Amplitudes, in the form of air-pressure changes reach the ear. There are no markers to indicate where
one phoneme ends and the next begins, even in moment-by-moment sound amplitude figures.
Likewise, there are no gaps to indicate the boundaries between successive syllables or successive
words. Therefore, prior to phoneme identification, you need to “slice” this stream into the appropriate
segments—a step called speech segmentation (see Figure 9.3 & page 328-329)
•Most of use think that there are pauses between words that mark word boundaries, however this is an
illusion. This is evident in segmentation ambiguity, when we “hear” (slice) the pauses in the wrong
places and segment the speech stream in a way the speaker didn’t intend (“The sky is falling” for “this
guy is falling”), or when we listen to foreign speech where we cannot “supply” the word boundaries, and
therefore hear a continuous, uninterrupted flow of sound. This is why foreign speech sounds so fast
•Coarticulation: in producing speech, you don’t utter one phoneme at a time. Instead, the phonemes
“overlap”, so while you producing [s] for “soup”, your mouth is getting ready for the vowel
•This overlap allows faster and more fluent speech
•The [s] produced before one vowel is different for the [s] produced for another vowel, and so we
cannot point to a specific acoustical pattern and say “this is the pattern of an [s] sound”; the
acoustical pattern is different in different contexts. Speech perception has to “read past” these
context differences in order to identify the phonemes produced
Aids to Speech Perception
•How do you manage to perceive speech accurately and easily?
•Part of the reason is that the speech you encounter day-by-day is limited in its range; we know tens of
thousands of words, but most are rarely used
•In fact, it is estimated that the 50 most commonly used words in English make up more than half of
the words you actually hear
•Also, you don’t rely only on the stimuli you receive, but supplement this input with other knowledge
•One proposal is that when you hear the first phoneme, you activate all the words in your vocabulary
that begin with that phoneme, narrowing the activation down when you hear the next, and so on
•In other cases, speech perception is guided by knowledge that relies on the context in which a word
appears
•Phoneme restoration effect: participants supplying missing sounds (e.g., legi*lature) on their own
without realizing the the [s] was not there; this occurs more when the stimuli appears in context that
guides them rather than in isolation
Categorical Perception
•Speech perception also benefits from categorical perception: you’re much better at hearing the
differences between categories of sounds than you are at hearing the variations within a category of
sounds
•This means that you’re very sensitive to differences between, say, [d] and [t] sounds, but insensitive
to differences within these categories, like one [p] sound from a somewhat different [p] sound. This
allows you to hear the differences that matter without being distracted by insignificant variations
•In a study, the first stimulus is a [ba] sounds, and the upcoming stimuli are gradually changed to
sound slight more like [pa] until a clear [pa] sound is reached. We would expect that participants
would become less and less likely to identify the sound as [ba] every time; however, even though
gradual changes are made, participants “hear” an abrupt shift, recognizing half as [ba] and half as
[pa]—they seem indifferent to differences within each category, hearing either [ba] or [pa]
•Chinchillas also show the pattern of greater sensitivity to between-category differences than within-
category differences
Morphemes and Words
•Speakers know the word’s sound—the sequence of phonemes that make up the word. The speaker
also generally knows the word’s orthography—the sequence of letters that spell the word. Third, the
speaker knows how to use the word within various phrases, governed by rules of syntax. Finally, the
speaker needs to know the meaning of a word (semantic representation for the word)
Word Meaning
•What a word or phrase refers to is called the word’s (or phrase’s) referent. One may propose that if you
know the referent of, say, “bird”, then you know what the word “bird” means
•Still, there are key differences between a word’s referent and its meaning. Some phrases have no
referent because they refer to something that doesn’t exist (“unicorn”), also sometimes the referent is
temporary or a matter of coincidence (“president of United States”)
•For these reasons, word meanings must involve more than reference
•For present purposes, we can say that a large part of “knowing a word” is knowing the relevant concept
Building New Words
•The size of someone’s vocabulary is quite fluid, because new words are created all the time (e.g., a
new style of music or clothing demanding a corresponding new vocabulary)