Psycholinguistics Articles

Front 1

(Bock) Why do the authors characterize speech production as a “not-so-simple ability”? What is involved in the process of producing an utterance?

Back 1

Speech production is a complex. The process is described as including: thinking in a way that is able to be converted into language, retrieving the words or morphemes that can be used for the thought, putting the morphemes into the language-specific forms for the meaning of the thought, and connecting them to the sound forms.

Front 2

(Bock) What is the linguistic relativity hypothesis? What are some of the challenges it faces?

Back 2

The linguistic relativity hypothesis was developed by Whorf and Sapir. It is the position that “speakers of different languages experience their shared world in different ways, with the differences filtered through the lexical and grammatical devices of language. Its challenges: speakers of different languages differ in ways other than the differences in spoken languages; tests are difficult to design in such a way that they aren’t influenced by language effects; observed differences must be interpreted by linguistic relativity.

Front 3

(Bock) How are messages “pre-verbal”?

Back 3

Pre-verbal messages are ideas that do not yet consist of words or sentences to represent them.

Front 4

(Bock) What are lemmas?

Back 4

They are the abstract linguistic codes for words (lexical entries).

Front 5

(Bock) What are the different types of information that can be used to access lexical entries?

Back 5

Lexical entries can be accessed based on their meaning (semantics), syntactic category, or the phonological or morphological forms with which they are associated.

Front 6

(Bock) What evidence do the authors provide for the separability of semantic, grammatical, and phonological properties? How absolute is this separability?

Back 6

The tip of the tongue state provides evidence for the separability of semantic, grammatical, and phonological properties because it shows that a speaker can be aware of the meaning of a word without being able to recall the word’s complete form. Speech errors also provide evidence for the separability of these linguistic properties in that they can involve whole words or individual sounds. There are processes that manipulate words, morphemes, and phonemes. Finally, through experimental research we know that the retrieval of grammatical and semantic information precedes the retrieval of phonological information (retrieval processes are multi-step and have separation). This last temporal separate is not absolute and do not necessarily lead to discrete stages of word production.

Front 7

(Bock) How do conceptual accessibility, word-form frequency, and frequency inheritance affect lexical retrieval?

Back 7

Conceptual accessibility refers to the fact that a context with more relevant cues can facilitate matching a word. Word-form frequency refers to the fact that words that appear in one’s dialogue in higher frequency are produced faster than words that appear with low frequency. Frequency inheritance refers to the fact that the production of low frequency words is higher when homophones with unrelated meanings occur in high frequency in the lexicon; thus, reducing the importance of conceptual accessibility.

Front 8

(Bock) What must an English speaker know or determine about the structure of an utterance in order to produce correct subject-verb agreement?

Back 8

The speaker must know the subject and its number information, and the verb as well as what of the verb part carries number marking and what the number of this marking should be.

Front 9

(Bock) Explain attraction and structural priming in speech production.

Back 9

Attraction is a type of agreement error in which the verb and the noun do not agree. Verbs are more likely to err in response to nouns within their clauses than nouns in other clauses, even though they could occur within the same temporal vicinity, such as immediately preceding the verb. Structural priming is when a speaker uses structures in new utterances that were recently heard or produced.

Front 10

(Bock) What three unsolved issues/questions do the authors conclude with?

Back 10

1. How does language production related to language comprehension?
2. How do speakers put their knowledge of language to use along with their perceptual, attentional, conceptual, and motor abilities to adequately convey particular meanings at particular places and times to particular audiences?
3. Is language distinct or no different from other mental functions or skills? How general cognition can work to yield the human capacity to learn and use language? What are the nature and origins of language capacity and how does it interface with our cognition?

Front 11

(Bock)

Back 11

1. Researchers could compare groups of participants creating some component of language with participants perceiving that same component. Specifically, processes involved in coding and decoding the same language need to be studied to see if differences are observed.
2. Isolating elements could be a useful way to study the second question. Allowing certain speakers to use gestures to convey a message and prohibit others from using these gestures could lead to the identification of differences in the speakers’ abilities to convey messages by implementing motor abilities in conjunction with speech.
3. More studies of communication in other species (e.g. apes, birds) could provide more information to answer the third question. These studies should compare cognitive abilities with communication abilities. Or, studies could be completed with children born without speech and hearing, or without vision, though it would be challenging to determine which abilities are minimally essential for co

Front 12

(Samuel) Categorical perception

Back 12

When presented with stimuli that fall on a continuum, listeners perceive items as one category or another, at one end of the continuum or the other. There is an abrupt change in perception somewhere within the continuum of stimuli.

Front 13

(Samuel) Right-ear advantage

Back 13

When presented with separate stimuli to the right and left ears, listeners more often correctly report the stimuli presented to the right ear than the left ear. The right ear has better connections to the left hemisphere of the brain, which is where language is most often processed.

Front 14

(Samuel) Phoneme restoration

Back 14

Phoneme restoration is said to occur when a listener “restores” a phoneme actually replaced by a cough or white noise. In experiments, the phoneme is replaced by the other sound and listeners are asked to indicate the position of the cough or other noise. Performance is low since they automatically “restore” the missing phoneme.

Front 15

(Samuel) Statistical learning of speech

Back 15

Statistical Learning of Speech is most generally the idea that there are frequency effects and distributional patterns in speech perception. More specifically, as language is heard, if a particular sequence of speech is heard often, a chunk, or unit in this model, will develop.

Front 16

(Samuel) Perceptual learning of speech

Back 16

Perceptual Learning of Speech is the ability to modify how the units of statistical learning are used. Perceptual Learning of Speech can be divided into two types – the modification of phonemic categories and the modification of the system.

Front 17

(Samuel) Describe Motor Theory and discuss support for it and problems for it.

Back 17

Motor Theory states that there is a special processor for speech sounds; the neural mechanisms involved in processing speech sounds differ from those that process other sounds. This processor is thought to extract a speaker’s intended articulations (gestures) used to produce speech. In support of Motor Theory, there are reports of (1) trading relations among acoustic information possibly deriving from the same motor commands, (2) compensation for coarticulation, and (3) instances of duplex perception. Problems for Motor Theory include the duplex perception of nonspeech sounds and the results of studies with nonhuman species that show perceptual ability similar to that of humans.

Front 18

(Samuel) What is the Direct Realist View?

Back 18

The Direct Realist View is a general theory of perception without a separate special processor for speech sounds, but states that listeners perceive the gestures that create the sounds that are heard.

Front 19

(Samuel) Describe the General Auditory Approach and specifically discuss how it differs from Motor Theory and Direct Realist View.

Back 19

The General Auditory Approach states that speech sounds are perceived with the same mechanisms used to perceive other general sounds. In this approach, the gestures of speech are not important because speech is treated as just another type of sound in the environment of the listener. This is drastically different from Motor Theory since there is no special processor. It is also different from the Direct Realist View. The Direct Realist View does not have a separate processor, but still differs from the General Auditory Approach because in the Direct Realist View the gestures of the speech sounds are perceived.

Front 20

(Samuel) Describe the following effects. Why are they relevant to this article?

Back 20

a. Ganong effect: The Ganong Effect provides evidence that listeners are lexically biased towards perceiving actual words rather than words that do not exist when presented with ambiguous stimuli, showing that lexical information is used to guide perception with a speech signal is underspecified. The actual study used sounds designed to be between two phonemes of a language (e.g., /d/ and /t/) and combined them with other sounds (e.g., “ash” and “ask”) that would create real words if one of the two phonemes were perceived (“task” and “dash”), but non-words if the other phoneme was perceived (“dask” and “tash”) .
b. McGurk effect: The McGurk Effect shows that visual information is also used during acoustic-phonetic processing. When presented with visual stimuli (e.g., video) of a speaker creating one sound (without any complementary audio) and an auditory stimuli of a speaker creating another sound (without any complementary video), the listener combines the information during percept

Front 21

(Samuel) Describe the Cohort Model.

Back 21

A cohort is a set of active representations and the Cohort Model suggests that this set of active representations becomes available to us from the moment we perceive stimuli and is dwindled down as we are given increasingly more information until we decide on the correct item.

Front 22

(Samuel) What are nonsegmental factors? Give examples

Back 22

Nonsegmental factors are factors that affect speech perception, but that do not apply directly to a given segment. Examples include coarticulation and phonotactic probabilities, prosodic influences, lexical stress, lexical tones, and indexical factors (e.g., pitch, loudness).

Front 23

(Samuel) Why is speech perception said to be dynamic?

Back 23

Speech perception must operate dynamically because we do not live in static, homogeneous speech environments. Instead, we are constantly exposed to new input (e.g., new speakers, new dialects, new accents, and new words). We must adjust in response to the new and variable input we receive. The more information that can be used to adjust perception the better.

Front 24

(Samuel) What is the conclusion of the author regarding the direction of future research?

Back 24

The authors believe that, at the time the review was published, speech perception research was rather divided with perception of speech being separated from research on word recognition. The authors believe that future research would benefit from the joining of these two areas of related research.

Front 25

(Pitt) How can the phonological knowledge of a language aid in word processing (941)?

Back 25

If a listener perceives certain phonotactically illegal sequences in a spoken word the listener can use phonological knowledge of the illegal sequence to identify the intended word. (Example: tomorrow → tmorrow, knowledge that /tm/ is illegal and that schwa (“o”) is often deleted when spoken can lead the listener to recognize the word anyway.)

Front 26

(Pitt) What is “phonotactics” (942)?

Back 26

Phonotactics are the permissible phoneme combinations of a language.

Front 27

(Pitt) What did the study by Massaro and Cohen (1983) show about the perception of steps along an /r/-/l/ continuum (942)?

Back 27

They found that listeners’ classification of responses showed a bias in favor of legal sequences, even when what was spoken was an illegal sequence.

Front 28

(Pitt) What two interpretations could be used to explain these results (942)?

Back 28

One interpretation is that knowledge of phonotactic constraints within English affected the processing of /l/ or /r/. Another interpretation is that the frequency with which the legal clusters occur in English either equally or additionally affected processing.

Front 29

(Pitt) What combinations of phonemes did Pitt use in his first experiment? Why did he include /b/ and /g/, which can legally combine with both /r/ and /l/ in English (943)?

Back 29

He used the same /t/, /d/, and /s/ obstruents with the liquids /l/ and /r/, but also added /b/ and /g/ as obstruents. He added /g/ because it could provide a means to dissociate phonotactic legality from cluster frequency – although both /gl/ and /gr/ are legal, /gr/ is much more frequent than /gl/. He added /b/ because it could serve as a baseline – both /br/ and /bl/ are similar in frequency.

Front 30

(Pitt) The results of the first experiment are graphed in Figure 2. Explain and interpret this graph (944).

Back 30

When /s/ preceded /r/ at the very end of the continuum, this produced the lowest response rate out of the 5 contexts (demonstrating an /l/ bias following /s/), while the other 4 were all fairly close and within 15% of complete accuracy. All responses for /l/ were not quite as accurate as for /r/, but notably /gl/ was one of the most accurate, showing that perhaps frequency of occurrence in English did not influence the processing. /t/ showed an /r/ bias, opposite to /s/.

Front 31

(Pitt) Which of the two interpretations discussed above do the results appear to support more? Why (945)?

Back 31

The results appear to better support the theory of phonotactic permissibility because it shows that the phonotactic bias is only found with illegal clusters. Frequency does not seem to be behind this processing as evidenced by the lack of a bias toward /gr/ and against /gl/ along the continuum as well as the lack of a high correlation between the size of the phonotactic effect in each context and the size of the difference between /r/ and /l/ frequency counts in each context.

Front 32

(Pitt) How is phonological illegality avoided word medially (946)?

Back 32

It is avoided by syllabifying the words so that the initial consonant is placed in the coda of the first syllable and the liquid is placed in the onset of the following syllable.

Front 33

(Pitt) How would the order of syllabification and phonotactic context rules in speech processing affect the perception of a word-medial illegal cluster (946)?

Back 33

If syllabification took precedent over phonotactic context rules then the consonants of a word medial illegal cluster would be perceived as belonging to different syllables before knowledge of phonotactic illegality would affect liquid processing, meaning that we would not see any bias in labeling of /r/ or /l/. However, if phonotactic rules take priority we would still expect to see large labeling biases.

Front 34

(Pitt) Were phonotactic context effects found in the word-medial clusters? What does this suggest about the order of syllabification and phonotactic permissibility (947-8)?

Back 34

Phonotactic effects appeared weaker when the clusters occurred word medially, however large phonotactic effects were still found in /s/ and /t/ contexts.

Front 35

(Pitt) In the third experiment, Pitt explores another method of legalizing impermissible clusters: schwa insertion. What is his hypothesis? Is it supported (948-9)?

Back 35

Pitt asks the question: “Is vowel epenthesis another method by which phonotactic processes affect the perception of illegal clusters such as /sr/ and /tl/?”. His hypothesis is that it is and that we should see two-syllable responses occurring more often with these illegal clusters (in an effort to sort of legalize them) than with their legal counterparts.

Front 36

(Pitt) Pitt notes that his study suggests “that the domain of influence of phonotactic and syllabification processes are not the same” (949). What does he mean by this? What “domains” do the two processes operate in?

Back 36

In Experiment 2 there were no syllabification effects found on liquid processing. This, along with data from other studies has led him to believe that syllabification, a global process, requires larger chunks of the signal on which to operate, thereby occurring later in recognition. Meanwhile, phonotactic processes of permissibility that are local occur earlier because they can operate on smaller segments of speech. In other words – the two “domains are (1) global for syllabification processes and (2) local for phonotactic processes.

Front 37

(Levelt) List four characteristics of the elements that are involved in the process of word production.

Back 37

The steps involved include lexical selection (semantically and syntactically appropriate), phonological encoding, syllabification, and articulation (preparation of gestures).

Front 38

(Levelt) What is the crucial difference between the ‘speech error’ tradition and the ‘chronometric’ tradition?

Back 38

The chronometric approach measures naming latencies while the speech error approach records and subsequently analyzes errors made in casual or induced speech.

Front 39

(Levelt) Why is Dell’s computer-implemented model referred to as a two-step interactive activation model?

Back 39

It is referred to as a two-step model because there are two steps from the semantic to the phonological level. Semantic feature nodes spread their activation to the corresponding word or lemma nodes, which in turn spread their activation to phoneme nodes.

Front 40

(Levelt) What is a ‘mixed error’? Give an example.

Back 40

Mixed errors are both semantic and phonological in character. For example, if your target word is bear and you say hare there is a semantic similarity (they are both animals) and phonological similarities (they rhyme).

Front 41

(Levelt) How is Weaver’s model different from other computational models in the chronometric tradition? (Briefly describe both)

Back 41

Most implemented computational models in the chronometric tradition can only access a word’s name in a semantic or conceptual form – there is no phonological encoding. However, the WEAVER model has a fully developed phonological component.

Front 42

(Levelt) Describe Dell’s interactive lexical network.

Back 42

The three layers are semantic, word, and phoneme layers. The phoneme layer consists of onsets, vowels, and codas. All connections are bidirectional and facilitatory. There are further versions not pictured that utilize a phonological layers of features, such as voicing and nasalization, and syllable nodes. Activation spreads throughout the networks and it is fully cascading.

Front 43

(Levelt) Describe Weaver’s model.

Back 43

The first layer is conceptual, where there is bidirectional flow among the semantic relationships holding concepts together. The next layer is syntactic, the lemma layer. The nodes here include the frame, inflectional features, gender, etc. Here, the connection between lemma node and feature nodes is unidirectional. With more unidirectional spreading, the final layer is the form layer. Here morpheme nodes can connect with metrical and phoneme nodes. Phoneme nodes connect with all phonetic syllables in which they participate, unlike in Dell’s model. All connections in Weaver’s model are also facilitatory.

Front 44

(Levelt) What is the first step in conceptual preparation?

Back 44

The first step is the activation of the lexical concept. This can be any concept for which you have a morpheme or word in your lexicon.

Front 45

(Levelt) What influences word choice?

Back 45

Word choice is influenced by context, the perspective you take, and the shared knowledge of others involved in the dialogue.

Front 46

(Levelt) What has been the major discovery with implicit priming?

Back 46

A word’s form is built up incrementally, meaning that beginning with the first segment a word’s phonological form is generated segment by segment each time it is used (as opposed to being activated as an entire template).

Front 47

(Levelt) What is the basic difference between the homogeneous and the heterogeneous conditions in implicit priming?

Back 47

In the homogeneous condition the words in the set share an ‘implicit prime’ (a phonological property). The heterogeneous condition has no implicit prime.

Front 48

(Levelt) Give the two processes involved in morpho-phonological encoding and briefly explain each one.

Back 48

Code retrieval is a process by which the morphological makeup, metrical shape, and segmental makeup of a word are retrieved, following activation/selection of the lemma. Prosodification primarily consists of incremental syllabification, meaning that a word is broken down into syllables one syllable at a time and in context. This is evident because syllables can change depending on the morphemes surrounding a morpheme.

Front 49

(Levelt) Describe the term syllabary.

Back 49

A syllabary is believed to be a repository of overused, high frequency syllabic gestures where each is tied to a syllabic score. The syllabic score in turn indicates the appropriate motor tasks to generate the syllable.

Front 50

(Fromkin) What does Fromkin hope to show through her analysis of speech errors (p.27)?

Back 50

She wishes to demonstrate the probable laws of the formation of speech, or more specifically she proposes a specific ordering of the rules that govern language production.

Front 51

(Fromkin) What are the data for the study (p.28)?

Back 51

The data that Fromkin uses is a collection of six hundred speech errors made by friends and colleagues that she compiled over the course of three years.

Front 52

(Fromkin) What linguistic unit is most often involved in speech errors?

Back 52

The phone is the most often involved in speech errors.

Front 53

(Fromkin) How does Fromkin’s data show that clusters are sequences of discrete phones (p. 31)?

Back 53

Most specifically, her data show that in many errors where the intended utterance included consonant clusters only one segment of the cluster is involved in the error, thereby demonstrating that the separate consonants (or separate phones) are discrete.

Front 54

(Fromkin) How does her data show the reality of phonetic features (nasality, voicelessness, etc.) (pp. 35-6)?

Back 54

It show that phonetic features are likely independently extractable as performance units because of the evidence in many speech errors that involve errors only in an individual feature. Furthermore, segmental errors may also be explained in terms of features but because they may also be explained as segmental errors they cannot reliably demonstrate the possible feature errors.

Front 55

(Fromkin) Why does Fromkin suggest that there is a hierarchy and interdependence among phonetic features (p. 37)?

Back 55

She uses the example that you may find an example of addition or subtraction of nasality you will never find a nasality switch resulting in a voiceless nasal. They are therefore interdependent features and must occur in a hierarchical order. If a voiceless phoneme becomes nasalized, it must also become voiced, etc.

Front 56

(Fromkin) How does Fromkin’s data provide evidence for the existence of syllables (p.39)?

Back 56

One form of evidence she uses is that the distance between the origin and target segments generally does not exceed 7 syllables and we know that the human short term memory can accommodate about 7 units therefore this division may exist in terms of the way we process phonemes. Another form of evidence is that typically initial segments of a syllable will replace initial segments of another syllable in a speech error, nuclear syllables replace nuclear syllables and final syllables replace final syllables.

Front 57

(Fromkin) Fromkin states that transpositions must adhere to “the phonological rules of the language” (p. 40). How do her examples illustrate this?

Back 57

One example she uses is a spoonerism: “sphinx in moonlight” becoming “minx in spoonlight” because the /sf/ sound becomes /sp/ due to the fact that /sf/ is an exception and typically only voiceless stop obstruents can follow an initial /s/. In order to be permissible /sf/ becomes /sp/.

Front 58

(Fromkin) What happens to stress (both word stress and the contour of the sentence) in speech errors (p. 42-3)?

Back 58

Word stress moves with the word itself, meaning the syllable that would receive stress if the word were isolated will be the one to receive the primary stress when the word is moved. On the other hand, the contour of the phrase is fixed by the syntactic structure of the phrase, meaning that if the word receiving the primary stress is transposed then it will no longer receive the primary stress and instead the word taking its place will receive the primary stress.

Front 59

(Fromkin) What syntactic segments are exchanged in speech errors (pp. 44-45)?

Back 59

Syntactic phrases can be exchanged (groups of morphemes or words) as well as smaller stems and affixes (beginnings, endings, or roots of words). Also words will always exchange places or be replaced by words of the same class (a noun for a noun or verb for a verb, etc.).

Front 60

(Fromkin) What types of semantic errors appear in the data (p. 46)?

Back 60

Semantic errors can occur in the form of semantic blends (where some combination of two words with similar meanings comes out), the use of antonyms, the switching of a word relating to space to a word relating to time, and words that are a part of the same semantic class (though they may share nothing phonologically or phonetically).

Front 61

(Libben) What are the “first possible parse” and “last possible parse” hypotheses?

Back 61

The first possible parse hypothesis suggests that in ambiguous compounds the first possible constituent would be isolated, while the last possible parse hypothesis suggests that the last possible constituent would be isolated. For example, the first possible parse hypothesis would break “clamprod” down into “clam” and “prod”, while the last possible parse hypothesis would break it down into “clamp” and “rod”.

Front 62

(Libben) What other alternative do the experiments test?

Back 62

They test Libben’s notion that ambiguous novel compounds are actually processed through a “recursive parsing procedure that results in the creation of all possible morphologically-legal representations”, as opposed to only being processed according to the first or last constituent.

Front 63

(Libben) Explain the experimental procedure. What were the four components?

Back 63

In the morpheme recall task of the first experiment four components were used: (1) an alert component, (2) a stimulus component, (3) a focus component, and (4) a response component:
(1) An auditory beep is followed by the presentation of a “+” sign on the screen.
(2) Three stimuli are presented in succession, a monomorphemic word at the top of the screen, a compound presented in the center of the screen, and another monomorphemic word at the bottom of the screen.
(3) An arrow pointing in one of four directions.
(4) Participants write the morpheme that corresponds with the arrow direction (left and right being critical).

Front 64

(Libben) What hypothesis did the results of the first experiment support? What was the role of semantic plausibility?

Back 64

The first experiment does not support either the first or last possible parse hypotheses. Instead, it supports Libben’s recursive parsing hypothesis. Semantic plausibility highly correlated with parsing choices, showing that all possible constituents are activated during recognition of a novel compound and subsequently the most semantically plausible is chosen.

Front 65

(Libben) What prediction was the second experiment designed to address? How was the procedure altered?

Back 65

It was designed to address whether or not a compound prime would prime semantic associates of each constituent in the compound. The morpheme recall task was modified so that the monomorphemic words at the top and bottom of the screen were used as pre and post primes by either preceding or following critical compounds.

Front 66

(Libben) What priming effect was seen in the results of this experiment?

Back 66

Ambiguous compounds primed semantic associates of both their parses in both conditions, suggesting that both parses were conducted during processing. This is evidenced by the fact that monomorphemic recall was improved when items had some semantic relation to the compound, whether it was to the first or second parse. However, they also found that semantic priming likely does not affect prelexical parsing itself.

Front 67

(Libben) According to the authors, what is the “primary function of the prelexical parser? Why do their results lead toward a “re-analysis of the role of morphological parsing in the lexical processing system”?

Back 67

The primary function of the prelexical parser seems to be to supply all possible parses of a string, as opposed to determining the parsing choice. Because they found that extended semantic priming effects result as an outcome of prelexical parsing, but that semantic priming likely does not affect the prelexical parsing procedure itself they believe the role of morphological parsing need be re-analyzed. It has been a primary assumption that the goal of morphological parsing is efficiency, however this study reveals that efficiency is not the result.