Errors in Speech Production
“A Tanadian from Toronto”. We are all guilty of producing such speech errors and other slips of the tongue in our day-to-day communications. Speech errors have long been a source of amusement for many, a source of frustration for some, and more recently a source of serious study in the field of psychology. Although these errors are good for a laugh now and then, they prove to be of much greater value to the field of linguistics. Speech errors are providing linguists with insight into the mechanisms behind speech production. There are limitations to how much is available for study; the process of speech production is largely inaccessible for observation. However, by analyzing errors individually and in the context of their surroundings, we may better learn the underlying mechanisms that occur to produce our speech, and investigate the reality of speech production units in word formation . This chapter will introduce the different types of commonly documented speech errors, the rules that govern error-generation, and how these errors provide insight into some of the proposed speech production models. The nature of speech errors in this chapter will be based on speakers who have no pre-existing speech delays or disorders. These are errors made by speakers whose language and speech production systems are thought to be fully intact. First, let us start by introducing the smaller errors and then work our way up through a hierarchy based on size of units subject to the error.
An Overview of Speech Errors
Early estimates suggested upwards of 10 000 different speech errors are committed in the English language . These errors have become the source of investigation and experimentation in search of explanation of the basic processes that conduct speech production; from the basic stages of planning to the finished motor plan that produces audible speech  . A preliminary finding from error observations is that errors occur mainly within the same level of speech production rather than between levels of production. For example, this means in the occurrence of units being exchanged that one phoneme will change with another phoneme, but will not change with a syllable as it is a speech unit existing on a separate level of production . First, let us familiarize ourselves with nine commonly documented speech errors:
Addition: adding a unit
"optimal number→ moptimal number"
Anticipation: a later speech unit takes place of an earlier one
"reading list→leading list"
Blends: two speech units are combined
Deletion: a unit is deleted
Exchange: two units swap positions
thunder and lightning→lunder and thightning
Misdeviation: a wrong unit is attached to a word
Intervening node→intervenient node
Perseveration: speech unit is activated too late
Rule of thumb→ rule of rum
Shift: affix changes location
she decides to hit it→she decide to hits it
Substitution: unit is changed into a different unit
Set the table→set the chair
All different types of speech units are victims of speech error. Above we see changes in features, syllables, morphemes, affixes, words, and syntax. It must be acknowledged that classification of errors is no easy task as some errors are co-occur between different units of speech. Classification can at times be ambiguous. For example, an error such as "hit the spot" (as opposed to the intended word pot) could be considered a phoneme addition, or also a word substitution.
Errors made at the level of the phoneme, whether it be substitution, addition, deletion, or any others for that matter, are by far the most common speech errors . An error at this level can occur within a word but more frequently will occur between separate words. The majority of these phonemic errors are anticipations, in which a substitution occurs of a sound that is supposed to occur later in the sentence. In this case, the speaker produces the target phoneme earlier than intended and it interferes with the intended original phoneme; the interfering segment follows the error.
a) also share→ alsho share
b) sea shanty→ she shanty
Second in frequency to phonemic anticipation errors are perseverance errors (the interfering segment precedes the error), which are as follows:
a) walk the beach→walk the beak
b) Sally gave the boy→Sally gave the goy
The very nature of these errors, and the fact that they occur indicate that speech is well planned before it is articulated. As words get confused, like we saw above, we could speculate that all words of a sentence exist as part of a single representation in production and are therefore susceptible to being mixed at that stage in planning. Of course this is intuitive as a sentence could not be created if words were held as separate representations; at some point down then line the words must be integrated and related to create and complete the sentence. Dell et al., noted a difference between perseverations and anticipations depending on the context of the sentence. If one is speaking a novel sentence, they are more prone to perseverations, where as anticipations are more common amongst practiced and recited phrases.
Another possible phonemic error is the exchange of two segments, where the order of sound segments gets changed. Exchange errors have been interpreted as the possible combination of an anticipation and a perseverance .
a) feed the dog→ deed the fog
b) left hemisphere→ heft lemisphere
These phonological errors always involve the exchange of like units; a vowel exchanges with a vowel and a consonant with another consonant. Never is there an exchange between a vowel and a consonant. This is known as the consonant-vowel category effect .
Beyond vowels and consonants
All of the above examples involved the anticipation, perservation, or exchange of single segments. Errors consisted of small segments such as a vowel or a consonant. These individuals segments can further be combined. As individual segments, two consonants can be transposed. By addition of a consonant to a word, a cluster can be produced as opposed to an intended single segment, as follows:
Fish grotto→ Frish gotto
This is similar in all respects to the previously shown single segmented errors, the only difference now being that the affected segment has become a consonant cluster. A cluster however is not a single unit in speech production, but consists of a sequence of separable segments.
Although our focus on speech errors has thus far been on small-segment phonemic errors, this does not mean that errors amongst phonemes are the only source of speech error. Larger than phonemes are syllables that are also units of speech performance and susceptible to error. Nooteboom (1969) was the first to suggest that syllables could be a unit of measure in speech programming. He found that speech errors generally occur within seven syllables distance between the origin and target. This corresponds and fits with our understanding of a short-term memory span that allows us to comfortably remember seven consecutive items . Anything beyond this magic number of seven becomes challenging. Nooteboom supported the notion that segmental slips yield to a structural law of syllable placement. If we have two words, each with an equal amount of syllables, the corresponding syllables will be the ones to exchange in the event of an error. The first syllable of the origin word will replace the first syllable of the target word. Likewise, the final syllable of the origin word will exchange with the final syllable of the target word.
Moran and Fader→ Morer and Fadan
In further support of syllables being a unit of articulation, syllabic errors also occur as blends, substitutions, deletions, and additions.
Tremendously→tremenly (deletion of syllable)
Shout+yell= shell (blending of syllables)
As we continue up our hierarchy of speech units, we now see that units of meaning are susceptible to speech errors. Such errors tend to happen subsequent to the syntactic planning of the sentence . Even units as large as an entire word can be subject to an error such as exchange.
Bowl of soup→soup of bowl
Plant the seeds→plan the seats
Substitutions and exchanges of whole words occur but do so with like-constituents. A noun will take place of a noun, and the same goes for an adjective or verb. When there is a change in word placement but no change in morphemes, the error is said to consist of inflectional morphemes. However, when the root of the words remains and there is an error due to a morpheme addition or substitution, the error is known as a derivational morpheme error.
Bed time→time bed (inflectional)
Easily enough→easy enoughly (derivational)
Such derivational speech errors show that semantic intentions are intact, however, the choice of semantic features has been incorrect. Substitutions can also occur where the substituted word is structurally similar but semantically different from the intended word.
He was very productive→ he was very productful
Documentation of errors involving word affixes provides us with insight as to how words are stored and later produced in speech. An error such as the one above leads one to believe that the word 'productive' may be stored in the mental lexicon as two separate constituents. It is possible that the correct version is stored as product + ive, which is suggestive of rules for word formation. From such errors we may infer that there exists separate vocabularies for stems and affixes. The improper pairing of an affix (product+ful) then leads to a word that is impermissible by the rules of our language. This evidence supports the hypothesis that affixes are a source of speech error and that they may exist as a separate component of one’s lexicon.
In articulation of a sentence, there is a segment of primary stress at which one syllable will be stressed more than the others. Regardless of whether it is a vowel, a whole syllable, part of a syllable, or even a whole word being involved in substitution, the pattern of stress within the sentence does not change. Take for example the following error:
How great things were→ how things were great
In articulation of a sentence, there is a segment of primary stress at which one syllable will be stressed more than the others. Boomer and Laver suggest that despite an error of word exchange, the position of the primary stress in the sentence remains the same (in this case on the second word of the sentence).
Freud focused on the common errors we make in our day-to-day processes and made these errors a central point of his studies. Verbal errors (or more commonly: slips of the tongue) have since been titled Freudian slips. These are errors in speech (or memory and physical action) that are said to occur due to the interference of an unconscious wish, need, or thought. For example, a man calling his spouse by the name of his previous partner.
At first glance, Freudian slips seem like a gold mine for speech error research, however, they pose some difficulty in regards to research with the model of speech production. With our current linguistic tools, we are unable to tap into the unconscious processes of language production . With no access; we do not know the intentions that lie behind these errors, unlike the other speech errors we previously examined. Therefore, we cannot make any inference about these errors. To use these slips would require a vast knowledge of the inner-self of the speaker, something that is currently largely inaccessible. Until we develop such methods to do so, this resource of unconscious errors will remain largely untapped.
Speech errors in support of Language Production Models
A general consensus exists amongst linguistic theorists that words and sentences exist as a combination of structure and content. A complete sentence requires words to create meaning, and that the syntax and relation amongst words be permissible within the language. Meaning and syntax reflect content and structure, respectively. The content of words contains a series of phonological features and the structure entails the combination of these features into larger units of speech organization. Modern psychological theories stress the importance of separation between structure and content. Chomsky (1957) stated that creating a sentence requires different levels of representation. Agreeing with later theorists, it is suggested that a semantic representation is created first. Succeeding this representation are two linguistic representations, one with syntactic information and the other with phonological information. These representations are what eventually direct motor activity in the production of speech . The following evidence from speech provide support for this theory of language production:
Phonotactic Regularity Effect
Errors at the level of the phoneme habitually end up being sound sequences that are possible within the rules of that given spoken language. This effect was established to be the “first law” of speech errors by Wells (1951) and has been a central focus of speech error research since. There are,however, violations to this first law. There are 37 found examples of violations (Stemberger 1983), but this amounts to less than one percent of all noted phonological speech errors . This proves the phonotactic regularity effect to be significant as its prevalence is seen in such a vast majority (99%) of errors.
a) A reading list→ a leading list (no violation of English rule)
b) dam→ dlam (a rare case of violation)
In consideration of speech production theories with the phonotactic regularity affect, frames are created with only letter combinations that are permissible within the language. Impossible sound sequences are prohibited in word construction. It is assumed that there is no available frame for an illegal sequence such as 'dlam' to be created . This is understood as concrete evidence that phonological rules are actively considered in the process of speech production .
Consonant-Vowel Category Effect
As mentioned earlier, like-units are prone to exchange, but not differing units. A noun slips with another noun, and a verb with another verb. In a similar fashion, vowels and consonants (basic phonological units) only slip with their similar partner; a vowel for a vowel, and a consonant for a consonant. The rate at which this occurs in speech errors is even greater than the phonotactic regularity effect, meaning there are even fewer exceptions to the consonant-vowel category effect (<1%) . Instances of cross-category errors are extraordinarily rare, in fact, some say they do not occur at all . The consonant-vowel category effect is considered to be evidence of labeled slots in the frame for production. Labels indicate whether a vowel or a consonant will be accepted into a particular slot, and will only accept the segments (consonant or vowel) that correspond with the correct category. This is a preventative measure that disallows the event of a cross-category error.
Initial consonants (onset consonants) are consonants that begin a syllable or word. There is a far greater inclination for word-initial consonants to slip than those from other regions of the word, with 80% of consonant slips coming from word-initial consonants . In explanation of this effect, MacKay (1972) and Shattuck-Hufnagel (1987) hypothesize that initial consonants of syllables and words have a distinct representation in the phonological frame. This idea suggests that initial consonants in syllables and words are more detachable from the remainder of the word and would therefore be more susceptible to error and exchange. For example, in the word 'fog', the f-sound is more easily extracted and isolated than the sound of the g. The ease with which the initial consonant f can detach is thought to be in correspondence with a principal division in the structure of the word frame. Segments following the division prove to be less accessible and more buried in the word.
Syllabic Constituent Effect
The syllabic constituent effect occurs when a neighboring vowel and consonant are exchanged as a VC or CV unit with another similar pair. The sequence of vowel-consonant proves to be more susceptible to error than the sequence of consonant-vowel . Nootboome (1969) noted that of 24 collected phonological errors, 19 involved a VC sequence and only four involved a CV sequence. This effect provides support for a phonological frame that has structure within its syllables. A typical sequence of consonants and vowels follows the CVC pattern, where the first consonant is the onset consonant of the syllable, and the succeeding vowel and consonant combine to form a single unit; the rhyme constituent of the syllable. The fact that VC slips are far more common than CV slips provides further evidence that phonological structure plays a role in the production of the speech frame .
Taken into consideration separately, each of these effects unveil the functioning of different phonological rules and structures that may be at work in language production. Evidence from the corpus of naturally occurring speech errors and the underlying effects within speech errors supports multiple levels of linguistic analyses in the process of speech production. This growing body of evidence from naturally occurring speech errors suggests that speech production begins begins with a semantic and structural plan. Following this foundation is the progression to accessing the proper words, and finally the application of proper phonological information. With a greater understanding of speech errors comes an understanding of the speech production process. This will ultimately lead to an increased understanding or our means of communication; language.
- Fromkin, V.A. (1973). Speech errors as linguistic evidence. The Hague: Mouton
- Meringer, R., & Mayer, K. (1895). Misspeaking and Misreading: A psycholinguistics study. Stuttgart, Germany: Goschense Verlagsbuchhanlung.
- Lashley, K.S. (1951). The problem of serial order in behavior. In L.A. Jeffress, Cerebral mechanisms in behavior (pp. 112-136). New York: Wiley.
- Dell, G.S., Reed, K.D., Adams, D.R., & Meyer, A. (2000). Speech errors, phonotactic constraints, and implicit learning: A study of the role of experience in language production. Journal of Experimental Psychology: Learning, Memory, and Cognition, 26, 1355-1367.
- Dell, G.s., Burger L.K., & Svec, W.r. (1997). Language production in serial order: A functional analysis and a model. Psychological Review, 104, 123-147.
- Jay, T. (2003). The Psychology of Language. Upper Saddle River, New Jersey: Pearson Education
- Nooteboom, S.G. (1969). The tongue slips into patterns. Leyden studies in linguistics and phonetics. The Hague: Mouton
- MacKay, D.G. (1978). Derivational rules and the internal lexicon. Journal of Verbal Learning and Verbal Behavior, 17, 61-71.
- Chomsky, N. (1857). Syntactic structures. The Hague, Netherlands: Mouton.
- Dell, G.S., Juliano, C., and Govindjee, A. (1993). Structure and content in language production: A theory of frame constraints in phonological speech errors. Cognitive Science, 17. 149-195.
Chomsky, N. (1957). Syntactic structures. The Hague, Netherlands: Mouton.
Dell, G.S., Juliano, C., and Govindjee, A. (1993). Structure and content in language production: A theory of frame constraints in phonological speech errors. Cognitive Science, 17. 149-195.
Dell, G.s., Burger L.K., & Svec, W.r. (1997). Language production in serial order: A functional analysis and a model. Psychological Review, 104, 123-147.
Dell, G.S., Reed, K.D., Adams, D.R., & Meyer, A. (2000). Speech errors, phonotactic constraints, and implicit learning: A study of the role of experience in language production. Journal of Experimental Psychology: Learning, Memory, and Cognition, 26, 1355-1367.
Fromkin, V.A. (1973). Speech errors as linguistic evidence. The Hague: Mouton
Jay, T. (2003). The Psychology of Language. Upper Saddle River, New Jersey: Pearson Education
Lashley, K.S. (1951). The problem of serial order in behavior. In L.A. Jeffress, Cerebral mechanisms in behavior (pp. 112-136). New York: Wiley.
MacKay, D.G. (1978). Derivational rules and the internal lexicon. Journal of Verbal Learning and Verbal Behavior, 17, 61-71.
Meringer, R., & Mayer, K. (1895). Misspeaking and Misreading: A psycholinguistics study. Stuttgart, Germany: Goschense Verlagsbuchhanlung.
Nooteboom, S.G. (1969). The tongue slips into patterns. Leyden studies in linguistics and phonetics. The Hague: Mouton
Identify the type of errors shown in the following sentences as one of the nine possible errors in speech production (ex substitution, preservation). It is possible for a sentence to contain more than just one type of error. If so, identify the multiple types of errors.
a. Walk the beach→ walk the beak
b. Hop on one foot→ Fop on one foot
c. Pat put the pot on the pable
d. On the computer→ on the commuter
e. He had an upset stummy
f. The adds up to→ that add ups to
The following story was written by Tommy, a grade two student at Atlantic Memorial Elementary. It is a recount of his trip to the store on a beautiful weekend day. He plans to submit his story as part of his year end project. Tommy has called upon you to proof read his story!
Identify the types of errors made by Tommy (i.e. perseverance, shifts, deletions, additions etc.) and at which level of production these errors occur (i.e. phonemes, syllables, morphemes). Make the appropriate corrections so he can get the A+ he so greatly desires.
Tommy woke up to a bright sun-shining say. He bate his eggs-benny and got dressed as fast as possible, with an itch to break out the front door into the sunshine. He decide to gos for a walk to Mable’s Country Store. Off he went, with his walking stick in hand, along the trail to the tore. He had the lovely company of the morning birds springing their tune to him along the way, and the river at his side. Not too far along the trail, Tommy realized he had forgot aboutten his money. Being in such a rush, he realized he forgot to deed the fog too! He would need some money to buy a treat at Mables Mountry More, so he turned around and dashed back dome! He shelled at the top of his lungs when the neighborhood hound chased him down the road, but he made it back in the nick of nime. He gathered his money and decided he would bake his ticycle this time to make up for lost time. He made his was along the trail, through the trees birch and at last to the store. For sake of his Saturday morning tradion, he purchased two scoops of strawbry ice cream. Delicious!
In the week following the reading of this chapter, closely listen for and document any speech errors made by yourself and others. Return to this chapter to analyze and relate your documented errors to the content of this chapter. Which types of errors did you find to be most common from this time? Are there any types of errors you have heard frequently in past? Deletions, anticipations, exchanges, or others? Indicate how your documented errors agree, or disagree with such theories as the phonotactic regularity effect or the consonant-vowel category effect. Make note of whether the speaker makes an effort to correct their error, or if it goes unnoticed. Can any conclusions be drawn regarding the level of speech production at which an error has occurred?
a. Walk the beach→ walk the beak (Persevaration)
b. Hop on one foot→ Fop on one foot (Anticipation)
c. Pat put the pot on the pable (persevaration)
d. On the computer→ on the commuter (deletion)
e. He had an upset stummy (blend)
f. The adds up to→ that add ups to (shift)
Sun shining say→ sun shining day (Perseverance. Level: phoneme)
Bate eggs-benny→ ate eggs benny (anticipation: Phoneme. Addition: phoneme)
Decide to gos→ decided to go (shift: affix)
Trail to the tore→ trail to the store (perseverance: phoneme)
Springing their tune→ signing their tune (Addition: phoneme. Subtitution: morpheme)
Forgot aboutten→ forgotten about (Shift: morpheme)
Deed the fog→ feed the dog (anticipation and perseverance: phoneme. Combination of anticipation and perseverance= exchange)
Mables Mountry More→ Mables Country Store (Perseverance: phoneme. Substitution: morpheme)
Shelled→ yelled (Blend: syllable)
Nick of nime→ nick of time (Perseveration: phoneme)
Bake his ticycle→ take his bicycle (exchange: phoneme. Substitution: morpheme)
Treed birch→ birch trees (exchange: morpheme)
Tradion→ tradition (Deletion: syllable and phoneme)
Strawbry→strawberry (Deletion: syllable)