Speech production is the process by which thoughts are translated into speech. This includes the selection of words, the organization of relevant grammatical forms, and then the articulation of the resulting sounds by the motor system using the vocal apparatus. Speech production can be spontaneous such as when a person creates the words of a conversation, reactive such as when they name a picture or read aloud a written word, or imitative, such as in speech repetition. Speech production is not the same as language production since language can also be produced manually by signs.
In ordinary fluent conversation, people pronounce roughly four syllables, ten or twelve phonemes and two to three words out of their vocabulary (that can contain 10 to 100 thousand words) each second. Errors in speech production are relatively rare occurring at a rate of about once in every 900 words in spontaneous speech. Words that are commonly spoken or learned early in life or easily imagined are quicker to say than ones that are rarely said, learnt later in life, or are abstract.
Normally speech is created with pulmonary pressure provided by the lungs that generates sound by phonation through the glottis in the larynx that then is modified by the vocal tract into different vowels and consonants. However, speech production can occur without the use of the lungs and glottis in alaryngeal speech by using the upper parts of the vocal tract. An example of such alaryngeal speech is Donald Duck talk. also published as
The vocal production of speech may be associated with the production of hand gestures that act to enhance the comprehensibility of what is being said.
The development of speech production throughout an individual's life starts from an infant's first babble and is transformed into fully developed speech by the age of five.Harley, T.A. (2011), Psycholinguistics. (Volume 1). SAGE Publications. The first stage of speech does not occur until around age one (holophrastic phase). Between the ages of one and a half and two and a half, the infant can produce short sentences (telegraphic phase). After two and a half years, the infant develops systems of lemmas used in speech production. Around four or five, the child's lemmas are largely increased; this enhances the child's production of correct speech and they can now produce speech like an adult. An adult develops speech in four stages: activation of lexical concepts, select lemmas needed, morphologically and phonologically encode speech, and the word is phonetically encoded.
Three stages The production of spoken language involves three major levels of processing: conceptualization, formulation, and articulation.
The first is the process of conceptualization or conceptual preparation, in which the intention to create speech links a desired concept to the particular spoken words to be expressed. Here the preverbal intended messages are formulated that specify the concepts to be expressed.Levelt, W. (1999). "The neurocognition of language", p.87 -117. Oxford Press
The second stage is formulation in which the linguistic form required for the expression of the desired message is created. Formulation includes grammatical encoding, morpho-phonological encoding, and phonetic encoding. Grammatical encoding is the process of selecting the appropriate syntactic word or lemma. The selected lemma then activates the appropriate syntactic frame for the conceptualized message. Morpho-phonological encoding is the process of breaking words down into syllables to be produced in overt speech. Syllabification is dependent on the preceding and proceeding words, for instance: I-com-pre-hend vs. I-com-pre-hen-dit. The final part of the formulation stage is phonetic encoding. This involves the activation of articulatory gestures dependent on the syllables selected in the morpho-phonological process, creating an articulatory score as the utterance is pieced together and the order of movements of the vocal apparatus is completed.
The third stage of speech production is articulation, which is the execution of the articulatory score by the lungs, glottis, larynx, tongue, lips, jaw and other parts of the vocal apparatus resulting in speech.
Neuroscience
The motor control for speech production in right-handed people depends mostly upon areas in the left cerebral hemisphere. These areas include the bilateral supplementary motor area, the left posterior inferior frontal gyrus, the left insula, the left primary motor cortex and temporal cortex. There are also subcortical areas involved, such as the basal ganglia and cerebellum. The cerebellum aids the sequencing of speech syllables into fast, smooth and rhythmically organized words and longer utterances.
Disorders Speech production can be affected by several disorders:
Aphasia Anomic aphasia Apraxia of speech Aprosodia Auditory processing disorder Cluttering
Developmental verbal dyspraxia Dysprosody Infantile speech Lisp Malapropism Mispronunciation
Speech disorder Speech error Speech sound disorder Spoonerism Stuttering
History of speech production research
Until the late 1960s, research on speech was focused on comprehension. As researchers collected greater volumes of speech error data, they began to investigate the psychological processes responsible for the production of speech sounds and to contemplate possible processes for fluent speech. Findings from speech error research were soon incorporated into speech production models. Evidence from speech error data supports the following conclusions about speech production.
Some of these ideas include:
Speech is planned in advance. The lexicon is organized both semantically and phonologically. That is by meaning, and by the sound of the words. Morphologically complex words are assembled. Words that we produce that contain morphemes are put together during the speech production process. Morphemes are the smallest units of language that contain meaning. For example, -ed on a past tense word. Affixes and functors behave differently from context words in slips of the tongue. This means the rules about the ways in which a word can be used are likely stored with them, which means generally when speech errors are made, the mistake words maintain their functions and make grammatical sense. Speech errors reflect rule knowledge. Even in our mistakes, speech is not nonsensical. The words and sentences that are produced in speech errors are typically grammatical, and do not violate the rules of the language being spoken.
Aspects of speech production models Models of speech production must contain specific elements to be viable. These include the elements from which speech is composed, listed below. The accepted models of speech production discussed in more detail below all incorporate these stages either explicitly or implicitly, and the ones that are now outdated or disputed have been criticized for overlooking one or more of the following stages.
The attributes of accepted speech models are:
a) a conceptual stage where the speaker abstractly identifies what they wish to express.
b) a syntactic stage where a frame is chosen that words will be placed into; this frame is usually sentence structure.
c) a lexical stage where a search for a word occurs based on meaning. Once the word is selected and retrieved, information about it becomes available to the speaker involving phonology and morphology.
d) a phonological stage where the abstract information is converted into a speech-like form.
e) a phonetic stage where instructions are prepared to be sent to the muscles of articulation.
Also, models must allow for forward planning mechanisms, a buffer, and a monitoring mechanism.
Following are a few of the influential models of speech production that account for or incorporate the previously mentioned stages and include information discovered as a result of speech error studies and other disfluency data, such as tip-of-the-tongue research.
Model
The Utterance Generator Model (1971)
The Utterance Generator Model was proposed by Fromkin (1971). It is composed of six stages and was an attempt to account for the previous findings of speech error research. The stages of the Utterance Generator Model were based on possible changes in representations of a particular utterance. The first stage is where a person generates the meaning they wish to convey. The second stage involves the message being translated onto a syntactic structure. Here, the message is given an outline. The third stage proposed by Fromkin is where/when the message gains different stresses and intonations based on the meaning. The fourth stage Fromkin suggested is concerned with the selection of words from the lexicon. After the words have been selected in stage four, the message undergoes phonological specification. The fifth stage applies rules of pronunciation and produces syllables that are to be outputted. The sixth and final stage of Fromkin's Utterance Generator Model is the coordination of the motor commands necessary for speech. Here, phonetic features of the message are sent to the relevant muscles of the vocal tract so that the intended message can be produced. Despite the ingenuity of Fromkin's model, researchers have criticized this interpretation of speech production. Although The Utterance Generator Model accounts for many nuances and data found by speech error studies, researchers decided it still had room to be improved.
The Garrett model (1975)
A more recent (than Fromkin's) attempt to explain speech production was published by Garrett in 1975. Garrett also created this model by compiling speech error data. There are many overlaps between this model and the Fromkin model from which it was based, but he added a few things to the Fromkin model that filled some of the gaps being pointed out by other researchers. The Garrett Fromkin models both distinguish between three levels—a conceptual level, and sentence level, and a motor level. These three levels are common to contemporary understanding of Speech Production.
Dell's model (1994)
In 1994, Dell proposed a model of the lexical network that became fundamental in the understanding of the way speech is produced. This model of the lexical network attempts to symbolically represent the lexicon, and in turn, explain how people choose the words they wish to produce, and how those words are to be organized into speech. Dell's model was composed of three stages, semantics, words, and phonemes. The words in the highest stage of the model represent the semantic category. (In the image, the words representing semantic category are winter, footwear, feet, and snow represent the semantic categories of boot and skate.) The second level represents the words that refer to the semantic category (In the image, boot and skate). And, the third level represents the phonemes (syllabic information including onset, vowels, and codas).
Levelt model (1999)