The Organization of Language
10
Language
The Organization of Language Language use involves a special type of translation. I might, for example, want to tell you about a happy event in my life, and so I need to translate my ideas about the event into sounds that I can utter. You, in turn, detect those sounds and need to convert them into some sort of comprehension. How does this translation-from ideas to sounds, and then back to ideas-take place?
The answer lies in the fact that language relies on well- defined patterns-patterns in how individual words are used, patterns in how words are put together into phrases. I follow those patterns when I express my ideas, and the same patterns guide you in figuring out what I just said. In essence, then, we’re both using the same “codebook” with the result that (most of the time) you can understand my messages, and I yours.
But where does this “codebook” come from? And what’s in the codebook? More concretely, what are the patterns of English (or whatever language you speak) that-apparently-we all know and use? As a first step toward tackling these issues, let’s note that language has a well-defined structure, as depicted in Figure 10.1. At the highest level of the structure (not shown in the figure) are the ideas intended by the speaker, or the ideas that the listener derives from the input. These ideas are typically expressed in sentences-coherent sequences of words that express the speaker’s intended meaning. Sentences, in turn, are composed of phrases, which are composed of words. Words are composed of morphemes, the smallest language units that carry meaning. Some morphemes, like “umpire” or “talk,” are units that can stand alone, and they usually refer to particular objects, ideas, or actions. Other morphemes get “bound” onto these “free” morphemes and add information crucial for interpretation. Examples of bound morphemes in Figure 10.1 are the past-tense morpheme “ed” and the plural morpheme “s.” Then, finally, in spoken language, morphemes are conveyed by sounds called phonemes, defined as the smallest unit of sound that serves to distinguish words in a language.
Language is also organized in another way: Within each of these levels, people can combine and recombine the units to produce novel utterances-assembling phonemes into brand-new morphemes or assembling words into brand-new phrases. Crucially, though, not all combinations are possible-so that a new breakfast cereal, for example, might be called “Klof but would probably seem strange to English speakers if it were called “Ngof.” Likewise, someone might utter the novel sentence “I admired the lurking octopi” but almost certainly wouldn’t say, “Octopi admired the I lurking” What lies behind these points? Why are some sequences acceptable-even if strange-while others seem awkward or even unacceptable? The answers to these questions are crucial for any understanding of what language is. Phonology
Let’s use the hierarchy in Figure 10.1 as a way to organize our examination of language. We’ll start at the bottom of the hierarchy-with the sounds of speech.
The Production of Speech
In ordinary breathing, air flows quietly Figure 10.2). There will usually be some sort of sound, though, if this airflow is interrupted or out of the lungs and up through the nose and mouth (see altered, and this fact is crucial for vocal communication.
For example, within the larynx there are two flaps of muscular tissue called the “vocal folds.” (These structures are also called the “vocal cords,” although they’re not cords at all.) These folds can be rapidly opened and closed, producing a buzzing sort of vibration known as this vibration by putting your palm on your throat while you produce a [z] sound. You’ll feel no vibration, though, if you hiss like a snake, producing a sustained [s] sound. Try it! The [z] sound is voicing. You can feel voiced; the [s] is not.
You can also produce sound by narrowing the air passageway within the mouth itself. For example, hiss like a snake again and pay attention to your tongue’s position. To produce this sound, you placed your tongue’s tip near the roof of your mouth, just behind your teeth:; the [s] sound is the sound of the air rushing through the narrow gap you created.
If the gap is somewhere a different sound results. For example, to produce the [sh] sound (as in “shoot” or “shine”), the tongue is positioned so that it creates a narrow gap a bit farther back in the mouth; air rushing through this gap causes the desired sound. Alternatively, the narrow gap can be more toward the front. Pronounce an [f] sound; in this case, the sound is produced by air rushing between your bottom lip and your top teeth. These various aspects of speech production provide a basis for categorizing speech sounds. We can distinguish sounds, first, according to how the airflow is restricted; this is referred to as manner of production. Thus, air is allowed to move through the nose for some speech sounds but not others. Similarly, for some speech sounds, the flow of air is fully stopped for a moment (e.g., [p], [b], and [t). For other sounds, the air passage is restricted, but air continues to flow (e.g., [f], [z], and [r]).
Second, we can distinguish between sounds that are voiced-produced with the vocal folds vibrating-and those that are not. The sounds of [v], [z], and [n] (to name a few) are voiced; [f], [s], [t], and [k] are unvoiced. (You can confirm this by running the hand-on-throat test while producing each of these sounds.) Finally, sounds can be categorized according to where the airflow is restricted; this is referred to as place of articulation. For example, you close your lips to produce “bilabial” sounds like [p] and [b]; you place your top teeth close to your bottom lip to produce “labiodental” sounds like [f] and [v]; and you place your tongue just behind your upper teeth to produce “alveolar” sounds like [t] and [d].
This categorization scheme enables us to describe any speech sound in terms of a few simple features. For example, what are the features of a [p] sound? First, we specify the manner of production: This sound is produced with air moving through the mouth (not the nose) and witha full interruption to the flow of air. Second, voicing: The [p] sound happens to be unvoiced. Third, place of articulation: The [p] sound is bilabial. These features are all we need to identify the [p], and if any of these features changes, so does the sound’s identity. In English, these features of sound production are combined and recombined to produce 40 or so different phonemes. Other languages use as few as a dozen phonemes; still others use many more. (For example, there are 141 different phonemes in the language of Khoisan, spoken by the Bushmen of Africa; Halle, 1990.) In all cases, though, the phonemes are created by simple combinations of the features just described. The Complexity of Speech Perception This description of speech sounds invites a simple proposal about speech perception. We’ve just said that each speech sound can be defined in terms of a small number of features. Perhaps, then, all a perceiver needs to do is detect these features, and with this done, the speech sounds are identified.
It turns out, though, that speech perception is more complicated. Consider Figure 10.3, which shows the moment-by-moment sound amplitudes produced by a speaker uttering brief greeting. It’s these amplitudes, in the form of air-pressure changes, that reach the ear, and so, in an important sense, the figure shows the pattern of input with which “real” speech perception begins.
Notice that within this stream of speech there are no markers to indicate where one phoneme ends and the next begins. Likewise, there are, for the most part, no gaps to indicate the boundaries between successive syllables or successive words. Therefore, as your first step toward phoneme identification, you need to “slice” this stream into the appropriate segments-a step known as speech segmentation.
For many people, this pattern comes as a surprise. Most of us are convinced that there are brief pauses between words in the speech that we hear, and it’s these pauses, we assume, that mark the word boundaries. But this perception turns out to be an illusion, and we are “hearing” pauses that aren’t actually there. This is evident when we “hear” the pauses in the “wrong places” and segment the speech stream in a way the speaker didn’t intend (see Figure 10.4). The illusion is also revealed when we physically measure the speech stream (as we did in order to create Figure 10.3) or when we listen to speech we can’t understand-for example, speech in a circumstance, we lack the skill needed to segment the stream, so we’re unable to “supply” the word boundaries. As a consequence, we hear what is really there: a continuous, uninterrupted flow of foreign language. In the latter sound. That is why speech in a foreign language often sounds so fast.
Speech perception is further complicated by a phenomenon known as coarticulation (Liberman, 1970; also Daniloff & Hammarberg, 1973). This term refers to the fact that in producing speech, you don’t utter one phoneme at a time. Instead, the phonemes overlap, the [s] sound in “soup,” for example, your mouth is getting ready to say the vowel. While uttering the vowel, you’re already starting to move your tongue, lips, and teeth into position for producing the [P].
This overlap helps to make speech production faster and considerably overlap also has consequences for the sounds produced, ready for one upcoming vowel is actually different from the [s] you produce while getting ready for a different vowel. As a result, we can’t point to a specific acoustical pattern and say, “This is the so that while you’re producing more fluent. But the so that the [s] you produce while getting [s] sound.” Instead, the acoustical pattern is different in different contexts. Speech pattern of an perception therefore has to “read past” these context differences in order to…
