Theoretical Considerations and Practical Applications

Loading...

4 Theoretical Considerations and Practical Applications LEARNING OBJECTIVES

When you have finished this chapter, you should be able to: • Trace how the development of the speech sound evolved into the phoneme concept. • Describe the major class features, the cavity features, and the manner of articulation features of the Chomsky and Halle (1968) distinctive feature classification. • Provide examples of marked and unmarked sounds or sound classes. • Define natural phonology. • List examples of the common phonological processes. • Distinguish the linear from the nonlinear (multilinear) phonologies. • Describe autosegmental phonology and its use of a tiered representation. • Explain the metrical trees in relationship to strong and weak stressing. • Describe the characteristics of feature geometry. • Understand the importance of optimality theory as a constraint-based approach.

M

any students and clinicians are skeptical about theories. They are viewed as unnecessarily difficult to understand, impractical, and not really relevant to diagnostic and therapeutic issues. Some theories may indeed be somewhat difficult to understand because they are attempts to explain highly complex behaviors. For example, Schwartz (1992a) states that a theoretical framework that addresses any aspect of language acquisition requires

1. a theory of abstract knowledge and representation about the subject; 2. the processing of input, how it is represented, and how it is produced; and 3. aspects of developmental change. It is easy to see why a theoretical framework that attempts to consider all these aspects might be complex. The second objection to theories is that they are impractical. Actually, theories can be 57

ch04.indd 57

7/12/2007 8:05:28 AM

58

CHAPTER 4

very practical. Theories are based on confirmed observations or systematic experiments. As such, they try to abstract from many practical experiences, attempting to find order and rules amid seemingly entangled details. Theories can also serve as blueprints for practical tasks. For example, phonological theories attempt to explain the structure and function of phonological systems. The structure and function outlined by a specific theory can in turn be used to analyze a particular phonological system. This analysis procedure can then be applied to both normal and disordered phonological systems. Various theories, such as natural phonology generating phonological processes, have resulted in analysis procedures that are used daily to evaluate the phonological systems of children. The third objection to theories is that they are not relevant to the diagnosis and treatment of individual clients. Because theories guide and direct clinical work, they are fundamentally important and relevant to the diagnostic and therapeutic process. For example, as stated earlier, many students and clinicians are currently using phonological processes to describe patterns of errors and to determine therapeutic goals. The concept of phonological processes evolved from the theory of natural phonology (Donegan and Stampe, 1979; Stampe, 1969, 1972, 1973). The theory of natural phonology applied certain principles of generative grammar, itself another theory that has revolutionized the way professionals view language. Both of these theories have resulted in major changes in diagnostic and therapeutic procedures. Different types of analyses are now employed diagnostically, and a major shift in therapy has occurred due to the impact of these theories. The relevance of theories to our profession should not be underestimated. Theories also offer a variety of clinical possibilities. Each theory provides a somewhat different perspective on the problem to be

ch04.indd 58

solved. Therefore, if one theory is used, assessment and treatment will vary from those suggested by a second theory. This gives clinicians several possible directions and approaches from which they can choose. Each theory and its application provide the clinician with unique problem-solving advantages. Without such problem-solving strategies, these features would go unnoticed; valuable diagnostic information would be lost. Thus, theories provide a means of maximizing diagnostic and therapeutic skills. They are significant to the work clinicians do professionally. Chapters 2 and 3 dealt primarily with production features of articulation—with speech sound forms. The focus in Chapter 4 shifts to phonology—to speech sound function. This shift is a consequence of the fact that contemporary theories in our field are phonological theories; they clearly emphasize the function of the phoneme as a meaning differentiating unit. The first goal of this chapter is to introduce the reader to some basic terminology and principles underlying many of the contemporary phonological theories. They are contained in the section on phonology. The second goal is to present several phonological theories that have been applied clinically within the discipline. Each phonological theory will be discussed in relationship to its theoretical framework, how it developed, and how it functions. Finally, clinical implications will be suggested for each of the presented theories.

PHONOLOGY What Is Phonology? Phonology can be defined as the description of the systems and patterns of phonemes that occur in a language. It involves determining the language-specific distinctive phonemes and the rules that describe the set of changes

7/12/2007 8:05:30 AM

THEORETICAL CONSIDERATIONS AND PRACTICAL APPLICATIONS

that take place when these phonemes occur in words (Ladefoged, 2006). Within this system, the smallest entity that can be distinguished by its contrasting function within words is called the phoneme. The phoneme is, thus, the central unit of phonology. Many different theoretical frameworks for phonological investigations exist. However, these various approaches all have one fundamentally important commonality, the differentiation between two levels of sound presentation: 1. the phonetic level, with sounds (phones, allophones) as central units, and 2. the phonemic level, represented by phonemes. How Does Phonology Work? To understand the concept of phonology, it is important to differentiate clearly between speech sounds and phonemes. Speech sounds (phones, allophones) are physical forms that are the result of physiological processes and that have objectively verifiable acoustic properties. Phonemes, on the other hand, are defined in terms of their linguistic function—that is, in terms of their ability to establish meaningful units in a language.

Phonology as a concept and discipline has undergone considerable changes. The original French and German terms phonologie (Baudouin de Courtenay, 1895) and Phonologie (Trubetzkoy, 1931) were, under the influence of structuralism, replaced by functional phonetics (Jakobson, 1962; Martinet, 1960). The term functional phonetics emphasized the functional aspect of speech sounds. Phonology has also been called phonemics (Sapir, 1925), underlining the linguistic function of the phoneme. The term phonology is presently preferred and used by most professionals within the field of communication disorders.

ch04.indd 59

59

How Did the Concept of the Phoneme Develop? The phoneme as concept and term was born toward the end of the nineteenth century at a point when linguists and A definition of the phoneticians found it nec- phoneme and its essary to expand the for- relationship to mer single sound concept phonology are found into a two-dimensional in Chapter 1. sound concept: 1. speech sounds as production realities 2. speech sounds in their meaning-establishing and meaning-distinguishing function, as “phonemes” In their works, the British phonetician Henry Sweet (1845–1912), the German Eduard Sievers (1850–1932), and the Swiss Jost Winteler (1846–1929) laid the foundation for the understanding of this duality. However, historically, Baudouin de Courtenay deserves the credit for introducing the concept of the phoneme in the year 1870. (The word phoneme existed prior to this time, but it was used as another label for speech sound.) N. H. Kruszewski (1881), a student of Baudouin de Courtenay’s, further popularized the term in his dissertation. Baudouin de Courtenay interpreted the proposed sound duality as differences between a physiologically concrete sound realization and its mental image. Influenced by the thinking of his time, Baudouin de Courtenay interpreted phonemes as primarily psychological sound units, as “psychic equivalents of the sound” (Lepschy, 1970, p. 60), as the sound “intended” by the speaker and “understood” by listeners. This was in contrast to the actually articulated sound, which was seen as a physiological fact. Similarly, the Russian linguist L. V. Scerba, who succeeded Baudouin de Courtenay, defined the phoneme as “the shortest general sound image of a given language which can be associated with

7/12/2007 8:05:30 AM

60

CHAPTER 4

meaning images, and can differentiate words” (Lepschy, 1970, p. 62). Although still psychologically oriented, the British phonetician Daniel Jones presented a more language-based phoneme concept in the first half of the twentieth century (Jones, 1938, 1950). Jones defined the phoneme as a “family of sounds in a given language which are related in character and are used in such a way that no one member ever occurs in a word in the same phonetic context as any other member” (Jones, 1950, p. 10). According to Jones’s definition, as long as speech sounds are understood as belonging to the same category, they constitute a phoneme of that language. For example, as long as [s]-productions, with all their verifiable phonetic differences (different speakers, various circumstances), are evaluated by listeners as being the same, as belonging to the s-category, these allophonic variations represent the single phoneme /s/ in that language. Today’s prevalent phoneme concept is still more functionally oriented. The specific use of the phoneme in a language is the primary emphasis. This strictly functional phoneme concept (strongly influenced by Ferdinand de Saussure’s [1916/1959] revolutionary new “structuralistic” way to look at language) was introduced by Nikolai S. Trubetzkoy and Roman Jakobson. Trubetzkoy, cofounder of the Prague School of linguistics, wrote that “the phoneme can be defined satisfactorily neither on the basis of its psychological nature nor on the basis of its relation to the phonetic variants, but purely and solely on the basis of its function in the system of language” (Trubetzkoy, 1939/1969, p. 41). When defining the phoneme, he added that it is “the sum total of the phonologically relevant properties of a sound” (p. 36). “References to psychology must be avoided in defining the phoneme, since the latter is a linguistic and not a psychological concept” (p. 38).

ch04.indd 60

One important aspect of a language’s phonological system is its phonemic inventory. However, this is not the only variable used in characterizing different phonological systems. Edward Sapir (1921) pointed out that two languages having the same phoneme inventory can, nevertheless, have very different phonologies. Even though inventories may be identical, the way these sound segments can and cannot be arranged to form words may be quite different. Consequently, the phonotactics, or “permissible” sound arrangements within a language, is an important aspect of phonemes’ “function” and is, therefore, an integral part of the phonology of a given language. Speech Sound versus Phoneme: Clinical Application Every utterance has two facets: an audible sequence of speech sounds and their specific meaning conveyed through this sequence. For example, if someone says, “Hey, Joe, over here,” there is an audible sequence of sounds [he doυ oυv h] that conveys a specific meaning. Both the physical form of the speech sound and its language-specific function need to be realized in order for the utterance to be meaningful. If only one aspect is realized, either speech sound form or function, a breakdown in the communicative process will occur. For example, although a child may have the correct speech sound form, in other words, be able to produce [p]–[b], [t]–[d], and [k]–[g], this child might leave out these sounds at the end of a word. Thus, form is accurate but the child’s realization of the function is inadequate. In this case, “beet” sounds like “bee” and “keep” becomes “key.” A breakdown in communication would probably occur. Adequate form and function of all segments are basic requirements for meaningful utterances in any language. Form is estab-

7/12/2007 8:05:30 AM

THEORETICAL CONSIDERATIONS AND PRACTICAL APPLICATIONS

lished by the way the segment in question is produced, by articulatory events. Segment function presupposes the observance of the language-specific rules regarding the arrangement of the speech sound segments. During an utterance, form and function become combined into meaning-conveying entities. Segment form and function are also largely dependent on one another. Without acceptable production features, sound segments cannot fulfill their functional task. If, for example, the word key is realized as tea, a frequent error made by children with t/k substitutions, elements of sound production have interfered with sound function. In this case, the phonological opposition between /t/ and /k/ has been destroyed. Segment function depends on normal segment form. Also, segment form depends on proper segment function. Without observance of the phonotactic rules governing the language, an acceptable sound production will not transmit the intended message. If, for example, a child produces a correct [s] but does not realize the phonotactic rules combining this [s] with other consonants in clusters, the meaning will be impaired. Stop might become top, or hats is realized as hat. For the purpose of effective verbal communication, regular segment form and function are indispensable. Articulation and Phonological Theories and Therapies: Separation or Unity? Historically, “correct” single sound realizations were often the central focus of articulation work. The mastering of how sound segments can and cannot be joined together to establish and convey meaning within the respective language was largely neglected. The underlying assumption was that speakers with defective articulation either “know” these rules already or will “learn” them through the various exercises that incorporated the sound in various contexts, for example. Articulation therapies fo-

ch04.indd 61

61

cused on the realization of acceptable speech sound forms. Today, it is often the other way around. The main orientation is the mastering of the phonological rules that govern the languagespecific utilization of the sound segments. Children with phonological disorders do not demonstrate difficulties primarily with the production of the sound segments but instead with their function, with the rule-governed arrangement of these segments. Thus, mastery of the phonological rules, not the speech sound realization, becomes the main goal. Phonological therapies focus on the realization of adequate segment function within a language system. Both intervention approaches have contributed substantially to the treatment of impaired articulation and phonology. They represent outgrowths of different theoretical viewpoints. However, their high degree of mutual dependency implies that, for successful articulation work, these two approaches are not clinically a matter of “either or” but of “as well as.” Of course, based on the specific clinical characteristics of an individual client, one approach may take precedence. If, for example, emphasis on speech sound form were the chosen approach, functional aspects would, nevertheless, also have to be taken into consideration. For example, if a child has just learned the speech sound [ ʃ ] (i.e., the form is learned), the child will practice this correct production in various syllable shapes according to phonotactic principles. In this example, function follows form. On the other hand, if speech sound function were the main goal, there might be a point in therapy when the clinician would need to consider aspects of speech sound form as well. For example, a child produces a speech sound that appears to be a correctly articulated [f ]. However, the child uses this [f ] as a substitution for [θ]. The word bath is articulated [bf ] and thing as [fŋ]. In words that normally

7/12/2007 8:05:31 AM

62

CHAPTER 4

are articulated with [f ], the child uses a [p]; “fan” becomes “pan” and “fig” is articulated as “pig.” The child is able to produce the form, but the function of the [f ] would need to be taught. Contrasting the phonemes /f/ and /p/ in minimal pairs might help with establishing the function of these two phonemes as meaning differentiating units. In summary, effective verbal communication always mirrors both aspects of speech sounds, acceptable form and function. Remediation must consider both sides of this duality; they represent two sides of the same coin. The next section of this chapter will address specific phonological theories. Each section will define, exemplify, and provide clinical examples to demonstrate the application of these theories to clinical assessment and treatment.

CLINICAL APPLICATION Phonological and Articulation Therapies Working Together Toby was 5;2 when he was seen by the new speech pathologist. Although he had previously received speech therapy, he was still considered very difficult to understand. A thorough assessment revealed that all fricative sounds were produced as stop-plosives. Thus, [f ] and [v] were articulated as [p] and [b], and the voiceless and voiced [s] and [z], [ ʃ ] and [], as well as [θ] and [ð] were articulated as [t] and [d]. Toby’s phonotactics for these sounds appeared intact. He did produce the noted substitutions consistently in all word positions. Toby often had difficulty discriminating words containing these phonemic oppositions. Thus, if the clinician asked the child to point to the picture of a “pin” versus a “fin” or of a “vase” versus a “base,” Toby would often be in error. After completing the evaluation, the clinician decided that Toby had a phonological disorder: Toby did not understand the function of these phonemes in the language system. The clinician began to work on differentiating and establishing these oppositions in meaningful contexts. Pictures and objects that contained these oppositions

ch04.indd 62

were used. The clinician noted that as Toby’s discrimination abilities improved, he attempted to produce [f ] and [v]; however, these realizations were consistently in error. As Toby struggled to correct the aberrant productions, the clinician realized that he was quickly becoming frustrated. The clinician used her knowledge of speech sound form to show Toby how to produce [f ] and [v] in an acceptable manner. Toby was interested, responded quickly to this instruction, and soon could produce regular [f ] and [v] sounds. He was very proud of his achievement and responded [naυ a kn te t wat].

DISTINCTIVE FEATURE THEORIES Distinctive feature theories are an attempt to determine the specific properties of a sound that serve to signal meaning differences in a language. The task is to determine which features are decisive for the identification of the various phonemes within a given language. Phonetic constituents that distinguish between phonemes are referred to as distinctive features. What Are Distinctive Features? How does one differentiate between apparent likenesses? For example, how do we distinguish between similar cars, houses, or streets? We look for discernible marks that might set the particular object apart from similar objects. A tree on the corner of a particular street, a brightly colored door on a house, for example, may serve as distinctive features that discriminate between streets or houses. “A distinctive feature is any property that separates a subset of elements from a group” (Blache, 1978, p. 56). A sound component is said to be distinctive if it serves to distinguish one phoneme from another. These units, which are smaller than sound segments, are considered to be “atomic” constituents of sound segments that cannot

7/12/2007 8:05:31 AM

63

THEORETICAL CONSIDERATIONS AND PRACTICAL APPLICATIONS

be broken down any further ( Jakobson, 1949). Theoretically, an inventory of these properties would allow the analysis of phonemes not only of General American English but also of all languages. Thus, distinctive features are considered to be universal properties of speech segments. How Do Distinctive Features Work? Distinctive features are the smallest indivisible sound properties that establish phonemes. An inventory of distinctive sound features would demonstrate similarities and dissimilarities between phonemes. These similarities and differences are marked by the presence of certain properties in some phonemes and the absence of these properties in others. The term binary is used in most distinctive feature analyses to indicate these similarities and differences. A binary system uses a plus (+) and minus (–) system to signal the presence (+) or absence (–) of certain features. Many different distinctive features must be considered in order to arrive at those that distinguish between phonemes. For example, consonants must be distinguished from vowels, voiced consonants from voiceless consonants, nasals from nonnasals, to mention just a few. If /k/ and // are considered, the following binary oppositions could be established: /k/ is a consonant = + consonantal

// is a consonant = + consonantal

is not a vowel = – vocalic is not a vowel = – vocalic is not voiced = – voice

is voiced = + voice

In this representation of similarities and dissimilarities, voicing is the only feature that distinguishes /k/ from //. Two sound segments are considered distinct and can, therefore, serve as phonemes if at least one of their features is different.

ch04.indd 63

The concept of binarity goes back to Jakobson’s influence on the evolution of distinctive feature theories. Jakobson, Fant, and Halle (1952) formulated that “any minimal distinction carried by the message confronts the listener with a two-choice situation” (paragraph 1.1). It follows that distinctive features are two-valued and require a yes/no decision concerning their presence or absence within sound segments. The concept of binarity has essentially been accepted by later distinctive feature systems. Ladefoged’s “Prime Features” (1971) are the clear exception. This system uses multivalued features in its description.

To expand slightly upon this feature system, consider the phonemes /k/, //, and /ŋ/. As previously noted, /k/ and // are distinguished from one another by the feature of voicing. How could this feature system be expanded to include the distinctive features that distinguish between /k/, //, and /ŋ/? /k/

//

/ŋ/

+ consonantal – vocalic – voice – nasal

+ consonantal – vocalic + voice – nasal

+ consonantal – vocalic + voice + nasal

Although voice distinguishes /k/ from // and /ŋ/, nasality is the feature that differentiates // and /ŋ/, all their other features being the same. In this example, nasality is the distinctive feature that creates an opposition between the phonemes // and /ŋ/. Presence or absence of the sound segments’ distinctive features can be displayed conveniently in a matrix form. Often, the Chomsky– Halle (1968) distinctive feature system is noted in textbooks for speech-language pathologists. However, this is not the only distinctive feature system. Over the years, many different distinctive feature systems have been developed (e.g., Jakobson, 1949; Jakobson, Fant, and Halle, 1952; Jakobson and Halle, 1956; Ladefoged,

7/12/2007 8:05:31 AM

64

CHAPTER 4

1971; Singh and Polen, 1972). Each of these authors had a somewhat different idea about which distinctive features were important when distinguishing between phonemes. Most of the feature systems were binary; however, Peter Ladefoged (1971, 2006) used multivalued features. In addition, most distinctive feature systems used articulatory dimensions to classify the phonemes, although acoustic parameters have been utilized as well (Jakobson, Fant, and Halle, 1952). One distinctive feature system is not necessarily superior to another. They were all developed to address somewhat different aspects of feature distinctions. In addition, many distinctive feature systems originated as a means of analyzing universal similarities and differences observed in phoneme systems of many different languages. This goal would of necessity incorporate feature modalities that are not necessary when analyzing General American English speech sounds. To summarize, distinctive feature systems are an attempt to document specific speech sound constituents that establish phonemes. Distinctive feature theories organize sound constituents according to some productional (or in some cases acoustic) properties that might be utilized in languages to establish meaning differences. The result is a system of contrastive, linguistically relevant sound elements. Historically, many different feature systems exist and many of the newer phonological theories, such as feature geometry, utilize their own somewhat different distinctive features. No one feature system has clear advantages over another. All distinctive feature systems reflect the authors’ concept of those characteristics that most aptly define the phoneme. How Did Distinctive Feature Theories Develop? The original distinctive feature theories grew out of the phoneme concept, which was de-

ch04.indd 64

veloped by the members of the Prague School in the 1930s. Very early in his work, Roman Jakobson, cofounder of the Prague School, hypothesized that the ultimate constituent of language was not the phoneme itself but its smaller components, its distinctive features. Jakobson stressed that these minimal differences serve the function of distinguishing between words that are different in meaning. It is these distinctive features that are functioning to distinguish between bat and pat, for example. The Jakobson, Fant, and Halle (1952) system used twelve acoustic features based on the sound segments’ spectrographic display. Such descriptions soon proved unsatisfactory for linguistic use primarily because similar acoustic representations can be the result of a number of different articulatory gestures. This led to a revision of the original system. In 1956, Jakobson and Halle published a new distinctive feature system that included articulatory production features. Many of the later distinctive feature systems (Chomsky and Halle, 1968; Halle, 1962; Ladefoged, 1971, 2006; Miller and Nicely, 1955; Singh, 1968; Singh and Black, 1966; Voiers, 1967; Wickelgren, 1966) were defined primarily according to ar-

An example of multivalued features includes using 1, 2, 3, and 4 to distinguish between differences in vowel height. Thus, [] is considered [1 height], while [i] is [4 height] (Ladefoged, 2006). The use of acoustic parameters to specify distinctive features resulted in features such as “compact” and “diffuse.” Based on acoustic displays of vowels, (+) compact was defined as a concentration of acoustic energy in the midfrequency region of the spectrum. Low vowels were considered (+) compact. The distinctive feature (+) diffuse was defined by a spread of acoustic energy over a wider frequency range. High vowels and labial, dental, and alveolar consonants were (+) diffuse (Jakobson and Halle, 1956).

7/12/2007 8:05:31 AM

THEORETICAL CONSIDERATIONS AND PRACTICAL APPLICATIONS

ticulatory features (or to a combination of articulatory and acoustic parameters). Distinctive Feature Theories: Clinical Application Distinctive feature systems were developed as a means of analyzing phonemes and entire phoneme systems of languages. Each phoneme of the particular system was assessed to determine if the distinctive feature was present (+) or absent (–). Although originally devised to analyze the regular realization of phonemes within and across languages, its analysis potential for disordered speech could not be overlooked. When sound substitution features were compared to target sound features, similarities and differences could be noted. Distinctive feature systems offered several advantages over the previous analysis systems of classifying errors according to substitutions, deletions, and distortions. First, they provide a more complete analysis. For example, sound substitutions can be broken down into several different feature components, which can then be compared and analyzed. Second, and perhaps more important, distinctive feature systems concentrate on the features that distinguish phonemes within a language. Previous analysis procedures had, at best, focused on phonetic production aspects of speech sounds. With the impact of phonology on the field of communication disorders, this emphasis now shifted to the phoneme and its function within the language system. Distinctive feature analysis contrasted the features of the target sound to the substitution, resulting in a list of distinctive features that differentiated between the two. This analysis could show whether (1) error and target sounds shared common features and (2) specific error patterns existed.

ch04.indd 65

65

Therapeutic implications follow logically. If the child can be taught to differentiate between the presence and absence of these differentiating distinctive features, the aberrant sound productions should be easily remediated. However, can children really understand and differentiate between distinctive features? Jakobson’s (1942/1968) hypothesis that children acquire features rather than sounds seems to support this assumption. If this is the case, therapy could facilitate this developmental process. In addition, if children acquire features rather than sounds, a certain amount of generalization should occur. Consequently, children should be able to generalize features from sounds they can realize to others they cannot. This could be therapeutically utilized. A child who can produce, for example, + voicing in one phonemic context should be able to generalize this + voicing to other phonemic contexts. Therefore, treatment of one phonemic opposition with specific distinctive features should lead to the norm production of other phonemic oppositions with the same distinctive feature oppositions. This would be a means of treating more than one phoneme in a time-efficient manner. Over time, several distinctive feature therapy programs were developed (Blache, 1989; Compton, 1970, 1975, 1976; McReynolds and Engmann, 1975; Weiner and Bankson, 1978). However, both the analysis procedures and the clinical applicability of distinctive features for speech-disordered children have been difficult to use and questioned by several authors (Carney, 1979; Foster, Riley, and Parker, 1985; Parker, 1976; Walsh, 1974). Some critical comments have focused on the fact that distinctive feature theory and distinctive features are abstract concepts: Distinctive features are theoretical concepts that were formulated to account for the sound patterns of languages. Carney (1979) further argued that a distinctive feature analysis, based on the phoneme concept, compels the clinician to ignore

7/12/2007 8:05:32 AM

66

CHAPTER 4

Are distinctive features dated? Although distinctive feature therapy does seem to be “out,” newer nonlinear (multilinear) phonological theories still rely heavily on distinctive features. Markedness is also an important aspect of one of the more contemporary phonological theories—“optimality theory.”

phonetic information. This phonetic information, exemplified by [s ] or [ ʃ ] , is not classifiable ↔ according to distinctive features and may lead to classifying errors inappropriately or not at all. For example, if the child produces a dentalized s-sound, [s ], how is this classified? There is no distinctive feature for dentalized [s]. The clinician might ignore the distortion, declaring it a norm production, or could perhaps classify it as a [θ]. In both cases, valuable diagnostic and therapeutic information would be lost.

GENERATIVE PHONOLOGY What Is Generative Phonology? Generative phonology is an outgrowth of distinctive feature theory, and it represents a substantial departure from previous phonological theories. Pregenerative theories of phonology— that is, those occurring prior to generative phonology (e.g., Jakobson, Fant, and Halle, 1952; Jakobson and Halle, 1956)—distinguished between phonetic and phonemic levels of realization. However, in pregenerative theories, both the phonetic and phonemic levels were analyzed by means of the actual productions, or the concrete realizations, of speech—for example, by using tape recordings of different language samples to assess the systems. Thus, pregenerative theories were developed around the surface forms, the actualities of speech production. On the other hand, generative phonologies expanded this concept decisively to include what has been called the underlying form. The underlying form is a purely theo-

ch04.indd 66

retical concept that is thought to represent a mental reality behind the way people use language (Crystal, 1987). Underlying forms exemplify the person’s language competency as one aspect of his or her cognitive capacity. The underlying forms also serve as points of orientation to describe regularities of speech reality as they relate to other areas of language, notably morphology and syntax. Generative phonology, then, assumes two levels of sound representation, an abstract underlying form called phonological representation and its modified surface form, the phonetic representation. Phonological rules are used to demonstrate the relationship between the underlying (phonological) and the surface (phonetic) forms. How Does Generative Phonology Work? Distinctive features and phonological rules are central to the concept of generative phonology. Phonological rules explain the differences between phonological and phonetic representations. These rules are usually stated in a formalized notation system. This notation system was important, but its practical clinical use was cumbersome and it is no longer used. Although often considered outdated, these features have continued to be utilized in more current phonological therapies, including maximal oppositions (see Chapter 9 for a detailed account of this therapy). This section will introduce the distinctive features system that has been most widely used (Chomsky and Halle, 1968). However, generative phonological rules, specifically their notation, will not be discussed. Grunwell (1987) gives excellent coverage of the notational rules and their application to disordered speech. Generative Distinctive Features. The first accounts of a generative distinctive feature theory were presented by Noam Chomsky (1957) and

7/12/2007 8:05:32 AM

THEORETICAL CONSIDERATIONS AND PRACTICAL APPLICATIONS

Chomsky and Morris Halle (1968). Chomsky and Halle’s (1968) The Sound Pattern of English is often cited as the major work in this area. They developed a new set of distinctive features that were different from those proposed by Jakobson and Halle (1956). In The Sound Pattern of English, the authors describe five features that are able to establish and distinguish between phonemes: (1) major class features, (2) cavity features, (3) manner of articulation features, (4) source features, and (5) prosodic features. The major class features characterize, and distinguish between, three sound production possibilities that result in different basic sound classes: 1. Sonorant. “Open” vocal tract configuration promoting voicing. American English vowels, glides, nasals, and liquids belong to this category. 2. Consonantal. Sounds produced with a high degree of oral obstruction, such as stops, fricatives, affricates, liquids, and nasals. 3. Vocalic. Sounds produced with a low degree of oral obstruction (not higher than required for the high vowels [i] and [u]), such as vowels and liquids. Cavity features refer to organ and/or place of articulation: 1. Coronal. Sounds produced with the apical/ predorsal portion of the tongue (“the blade of the tongue raised from its neutral position,” Chomsky and Halle, 1968, p. 304). This cavity feature marks several consonants, for example, [t], [d], [s], [z], [n], and [l]. See Table 4.1 for additional consonants. 2. Anterior. Sounds produced in the frontal region of the oral cavity with the alveolar ridge being the posterior border, that is, labial, dental, and alveolar consonants. [m], [n], [b], [p], [d], and [t] are examples.

ch04.indd 67

67

3. Distributed. Sounds with a relatively long oral-sagittal constriction, such as [ ʃ ], [s], and [z]. 4. Nasal. Sounds produced with an open nasal passageway—exemplified by the nasals [m], [n], and [ŋ]. 5. Lateral. Sounds produced with lowered lateral rim portions of the tongue (uni- or bilateral). The only American English example is [l]. 6. High. Sounds produced with a high tongue position, vowels as well as consonants. Thus, [i], [u], [k], and [ŋ] would be [+ high]. 7. Low. Vowels produced with a low tongue position—[ɑ], for example. The only consonants qualifying for this category are [h], [ʔ], and pharyngeal sounds. The latter are produced with the root of the tongue as organ of articulation. 8. Back. Vowels and consonants produced with a retracted body of the tongue, that is, back vowels and velar and pharyngeal consonants. 9. Round. Refers to the rounding of the lips for the production of vowels and consonants. [u] and [w] are [+ round]. Manner of articulation features specify the way organ and place of articulation cooperate to produce sound classes, signaling production differences between stops and fricatives, for example: 1. Continuant. “Incessant” sounds produced without hindering the airstream by any blockages within the oral cavity. Vowels, fricatives, glides, and liquids are [+ continuant]; stops, nasals, and affricates are [– continuant]. 2. Delayed release. Refers to sounds produced with a slow release of a total obstruction within the oral cavity. Affricates such as [ ] and [] are [+ delayed release].

7/12/2007 8:05:32 AM

68

CHAPTER 4

3. Tense. Consonants and vowels produced with a relatively greater articulatory effort (muscle tension, expiratory air pressure). [p], [t], [k], [i], and [u], for example, are [+ tense]. [b], [d], [g], [], and [υ], by comparison, are [– tense].

glish vowels, glides, liquids, nasals, and voiced stops, fricatives, and affricates are [+ voiced]. [p], [t], [k], [f ], [s], and [ ʃ ], by contrast, are [– voiced]. 3. Strident. The term strident (making a loud or harsh sound) is a feature of American English voiceless and voiced fricatives and affricates. However, the interdental fricatives [θ] and [ð] are [– strident].

Source features refer to subglottal air pressure, voicing, and stridency: 1. Heightened subglottal pressure. American English voiceless aspirated stops ([p], [t], [k]) are [+ HSP] because their production requires an added amount of expiratory airflow that, after freely passing the glottis, accumulates behind the occlusion within the oral cavity. 2. Voiced. Produced with simultaneous vocal fold vibration. All American En-

Prosodic features are named but not discussed in Chomsky and Halle (1968). To see how several of these distinctive features apply to General American English consonants and vowels, see Tables 4.1 and 4.2. Generative Naturalness and Markedness. One aspect of distinctive feature theory that seems to have more direct clinical applicability and

Table 4.1

General American English Consonant Matrix According to the Chomsky and Halle (1968) Distinctive Features p

b

t

d

k

g

θ

ð

f

v

s

z

ʃ







m

n

ŋ

r

l

w

j

h

Sonorant

































+

+

+

+

+

+

+

+

Consonantal

+

+

+

+

+

+

+

+

+

+

+

+

+

+

+

+

+

+

+

+

+







Vocalic







































+

+







Coronal





+

+





+

+





+

+

+

+

+

+



+



+

+







Anterior

+

+

+

+





+

+

+

+

+

+









+

+





+







Nasal

































+

+

+











Lateral









































+







High









+

+













+

+

+

+





+





+

+



Low















































+

Back









+

+

























+





+





Round











































+





Continuant













+

+

+

+

+

+

+

+











+

+

+

+

+

Del. Release





























+

+

















Voiced



+



+



+



+



+



+



+



+

+

+

+

+

+

+

+



Strident

















+

+

+

+

+

+

+

+

















ch04.indd 68

7/12/2007 8:05:32 AM

THEORETICAL CONSIDERATIONS AND PRACTICAL APPLICATIONS

69

4.2

Table General American English Vowel Matrix According to the Chomsky and Halle (1968) Distinctive Features

Consonantal Vocalic Coronal Anterior High Low Back Round Tense

i



e

ε



ɑ

ɔ

o

υ

u



– + – – + – – – +

– + – – + – – – –

– + – – – – – – +

– + – – – – – – –

– + – – – + – – –

– + – – – + + – +

– + – – – + + + –

– + – – – – + + +

– + – – + – + + –

– + – – + – + + +

– + – – – – + – –

can be found in later theoretical constructs is the concept of naturalness and markedness. Naturalness and markedness can be seen as two ends of a continuum. The term naturalness designates two sound aspects: (1) the relative simplicity of a sound production and (2) its high frequency of occurrence in languages. In other words, more natural sounds are those that are considered easier to produce and occur in many languages. Markedness, on the other hand, refers to sounds that are relatively more difficult to produce and are found less frequently in languages (Hyman, 1975). For example, [p] is considered a natural sound (= unmarked). It is easy to produce and occurs in many languages around the world. The affricate [tʃ ], though, is a marked sound: It is relatively more difficult to produce and is found infrequently in other languages. Marked and unmarked features are typically used when referring to cognate pairs, such as /t/ and /d/, and sound classes, such as nasals. Sloat, Taylor, and Hoard (1978) describe the following sounds and sound classes according to markedness parameters: Voiceless obstruents are more natural (unmarked) than voiced obstruents.

ch04.indd 69

Obstruents are more natural (unmarked) than sonorants. Stops are more natural (unmarked) than fricatives.

Obstruents include the stops, fricatives, and affricates. See Chapter 2 for a more complete definition.

Fricatives are more natural (unmarked) than affricates. Low-front vowels appear to be the most natural (unmarked) vowels. Close-tense vowels are more natural (unmarked) than open-lax vowels. Anterior consonants are more natural (unmarked) than nonanterior consonants. Consonants without secondary articulation are more natural (unmarked) than those with secondary articulation (such as simultaneous lip rounding). The concept of naturalness versus markedness became a relevant clinical issue when it was observed that children with phonological disorders have a tendency to substitute more unmarked, natural classes of segments for marked ones. For example, children substituted stops for fricatives and deleted the more

7/12/2007 8:05:34 AM

70

CHAPTER 4

marked member of a consonant cluster (Ingram, 1989b). Although the results of at least one investigation demonstrated contrary findings (McReynolds, Engmann, and Dimmitt, 1974), most investigations supported the notion that children and adults with speech disorders more frequently showed a change from marked segments to unmarked substitutions (Blumstein, 1973; Klich, Ireland, and Weidner, 1979; Marquardt, Reinhart, and Peterson, 1979; Toombs, Singh, and Hayden, 1981; Williams, Cairns, Cairns, and Blosser, 1970; Wolk, 1986). Markedness is also an important variable in newer theoretical models such as optimality theory (Prince and Smolensky, 1993). How Did Generative Phonology Develop? Generative phonology represents the applications of principles of generative (or transformational) grammar to phonology. The concept of generative grammar was first introduced by Noam Chomsky in 1957 in a book titled Syntactic Structures. Generative grammar departed radically from structuralistic and behavioristic approaches to grammar, which had dominated linguistic thought during the decades before Chomsky’s work. Prior to the introduction of generative grammar, linguists had analyzed the surface forms of sentences into their constituent parts. This type of analysis was found to be inadequate in various respects. An often-used example illustrates this point: John is eager to please. John is easy to please. If the two sentences are analyzed according to a structuralistic point of view, the results will indicate that both sentences have exactly the same structure. However, this analysis does not reveal that the two sentences have drastically different meanings. In the first sentence, John wants to please someone else—John is the subject of pleasing. In the second sen-

ch04.indd 70

tence, someone else is involved in pleasing John—John is the object of pleasing. One aim of generative grammar was to provide a way to analyze sentences that would account for such differences. To do this, a concept was developed that postulated not only a surface level of realization but also a deep level of representation. Competence and performance were also terms that distinguished between surface and deep levels of representation. Language competence was viewed as the individual’s knowledge of the rules of a language, whereas performance was actual language use in real situations. Structuralists and behaviorists had focused on an individual’s performance; generative grammar shifted this focus to include the concept of an individual’s language competence. The formulation of rules governing the events between the deep-level competence and surface-level performance was an important goal for generative grammarians. Distinctive Features and Generative Phonology: Clinical Application Generative phonology was originally developed to analyze the phonological systems of languages. Its application to phonological development in children has its foundation in Smith’s (1973) case study of his son Amahl. Other authors (e.g., Compton, 1975, 1976; Grunwell, 1975; Lorentz, 1976; Oller, 1973) extended these analysis principles to children with disordered phonological systems. Generative phonology, applied in this manner, compares the child’s phonological system to the adult’s. To do this using the distinctive feature system, the target sound is compared to the child’s substitution, noting the distinctive features that are different between the target and substitution. Distinctive Feature Analysis. A distinctive feature analysis compares the phonetic features

7/12/2007 8:05:35 AM

THEORETICAL CONSIDERATIONS AND PRACTICAL APPLICATIONS

of the target sound with the phonetic features of its substitution. Because the distinctive feature system is binary, (+) and (–), similarities and differences between target and substitution can be clearly ascertained. One of the advantages of this analysis is that it allows for a comparison of several sound substitutions to the target phoneme. For example, if a client substitutes [t] for [d], [z], and [ ʃ ], similarities and differences between all sound features can be compared. In addition, correctly and incorrectly realized features across several phonemes can be examined to see whether patterns exist. A pattern is characterized by frequent use of one or more identical distinctive features when the target sound and the sound substitution are compared. Most clinical applications use a version of the Chomsky and Halle (1968) distinctive feature system (Elbert and Gierut, 1986; Gierut, 1992; Grunwell, 1987; Lowe, 1994; McReynolds and Engmann, 1975). These distinctive features can be found in Tables 4.1 and 4.2. Any other distinctive feature system could be substituted; the principles of analysis would remain the same. Distinctive feature analysis is also used in feature geometry. (See pages 87–90.) What to Do? To describe patterns of errors, the distinctive features of the target phoneme and its substitution(s) are analyzed. Figure 4.1 depicts an example of a worksheet that can be used to identify them. 1. List the target phoneme and the substitution at the top of one of the boxes. If there are several substitutions for one target phoneme, each substitution should be listed in a separate box. 2. List the features that differ between the target sound and the substitution in the blank spaces under Feature Differences. Record their (+) or (–) values. These features are taken

ch04.indd 71

71

from the distinctive feature table you are using. 3. Transfer the information from the completed worksheet to the Summary Sheet for Distinctive Feature Analysis (Table 4.3). Table 4.3 provides the number of phonemes affected by each of the specific distinctive features. In Table 4.4, the results of an articulation test from H. H. are transcribed. A distinctive feature worksheet and summary form are completed for H. H. in Figure 4.2 and Table 4.5. By looking at Table 4.5 we can see that there are four distinctive features that each impact six different phonemes: – anterior is changed to + anterior (tʃ → t, d → d, k → t,  → d, ʃ → d, ʃ → s), + high is changed to – high (tʃ → t, d → d, k → t,  → d, ʃ → d, ʃ → s), + continuant is changed to – continuant (f → b, θ → b, ð → d, ʃ → d, s → t, z → t), and + strident is changed to – strident (tʃ → t, d → d, f → b, ʃ → d, s → t, z → t). In summary, distinctive feature systems attempt to capture those phonetic features that distinguish between phonemes of a language. Although these distinctive features are primarily productionally based, they represent an important aspect of a phonemic analysis. Error patterns can clearly be seen when frequently occurring distinctive features are summarized. Distinctive feature analyses cannot account for deletions, assimilations, or changes of the syllable structure.

NATURAL PHONOLOGY What Is Natural Phonology? “[Natural phonology] is a natural theory . . . in that it presents language as a natural reflection

7/12/2007 8:05:35 AM

72

CHAPTER 4

Figure 4.1

ch04.indd 72

Worksheet for Distinctive Feature Analysis

7/12/2007 8:05:35 AM

73

THEORETICAL CONSIDERATIONS AND PRACTICAL APPLICATIONS

4.3

Table Summary Sheet for Distinctive Feature Analysis Using the Chomsky–Halle Distinctive Feature System Feature Change

No. of Phonemes Affected

Feature Change

No. of Phonemes Affected

Sonorant Consonantal

+ to – + to –

________________ ________________

– to + – to +

________________ ________________

Vocalic

+ to –

________________

– to +

________________

Coronal

+ to –

________________

– to +

________________

Anterior

+ to –

________________

– to +

________________

Nasal

+ to –

________________

– to +

________________

Lateral

+ to –

________________

– to +

________________

High

+ to –

________________

– to +

________________

Low

+ to –

________________

– to +

________________

Back

+ to –

________________

– to +

________________

Round

+ to –

________________

– to +

________________

Feature

Continuant

+ to –

________________

– to +

________________

Delayed Release

+ to –

________________

– to +

________________

Voiced

+ to –

________________

– to +

________________

Strident

+ to –

________________

– to +

________________

Summary: ________________________________________________________________________________ ___________________________________________________________________________________________ ___________________________________________________________________________________________ ___________________________________________________________________________________________ ___________________________________________________________________________________________

of the needs, capacities, and world of its users, rather than as a merely conventional institution” (Donegan and Stampe, 1979, p. 127). Natural phonology incorporates features of naturalness theories and was specifically designed to explain the development of the child’s phonological system. The theory of natural phonology postulates that patterns of speech are governed by an innate, universal set of phonological processes. Phonological processes are innate and universal; therefore, all children are born with the capacity to use the same system of processes. Phonological processes, as natural processes, are easier for a child to produce and are substituted

ch04.indd 73

for sounds, sound classes, or sound sequences when the child’s motor capacities do not yet allow their norm realization. These innate, universal natural phonological processes are operating as all children attempt to use and organize their phonological systems. Therefore, all children begin with innate speech patterns but must progress to the language-specific system that characterizes their native language. Phonological processes are used to constantly revise existing differences between the innate patterns and the adult norm production. The theory points out prominent developmental steps children go through until the goal of adult phonology is reached in the child’s early

7/12/2007 8:05:37 AM

74

CHAPTER 4

Table 4.4

Single-Word Responses to Goldman-Fristoe Test of Articulation for Child H. H.

Target Word

Child’s Production

Target Word

Child’s Production

1. house

[hɑυ]

22. carrot

[tεwə]

2. telephone

[tεfoυ]

3. cup

[t p]

4. gun

[d n]

5. knife

[nɑ]

6. window

[wnoυ]

finger

[bnə]

7. wagon

[wdən]

ring

[wŋ]

wheel

orange 23. bathtub bath 24. thumb

[oυwn] [bt ] [b] [b m]

[wi]

25. jump

[d mp]

8. chicken

[ttə]

26. pajamas

[dmi]

9. zipper

[tpə]

27. plane

[ben]

10. scissors

[ttə]

11. duck yellow

blue

[bu]

[d t]

28. brush

[b s]

[jεwoυ]

29. drum

[d m]

12. vacuum

[tu]

30. flag

[b]

13. matches

[mtət]

31. Santa Claus

[tnə dɑ]

14. lamp

[wmp]

32. Christmas

15. shovel

[d və]

tree

[ttmə] [ti]

16. car

[tɑə]

33. squirrel

[tw ə]

17. rabbit

[wb]

34. sleeping

[twipn]

18. fishing

[bdn]

19. church

[t ]

20. feather

[bεdə]

21. pencils

[pεntə]

this

[bεd] [doυ]

child would not say

years. Disordered phonology is seen as an inability to realize this “natural” process of goaloriented adaptive change. How Does Natural Phonology Work? The theory of natural phonology assumes that a child’s innate phonological system is continuously revised in the direction of the adult phonological system. Stampe (1969) proposed three mechanisms to account for these changes: (1) limitation, (2) ordering, and (3) suppres-

ch04.indd 74

bed 35. stove

sion. These mechanisms reflect properties of the innate phonological system as well as the universal difficulties children display in the acquisition of the adult sound system. Limitation occurs when differences between the child’s and the adult’s systems become limited to only specific sounds, sound classes, or sound sequences. Limitation can be exemplified by the following: A child might first use a more “natural” sound for a more marked one. For example, all fricatives might be replaced by homorganic stops (e.g., [f ] → [p],

7/12/2007 8:05:38 AM

THEORETICAL CONSIDERATIONS AND PRACTICAL APPLICATIONS

Figure 4.2

ch04.indd 75

75

Distinctive Feature Worksheet for Child H. H.

7/12/2007 8:05:39 AM

76

CHAPTER 4

Table 4.5

Summary Sheet for Distinctive Feature Analysis: Application H. H.

Feature

Feature Change

No. of Phonemes Affected

Feature Change

No. of Phonemes Affected

Sonorant

+ to –

________________

– to +

________________

Consonantal

+ to –

________________ 2

– to +

________________ 0

Vocalic

+ to –

2 ________________

– to +

0 ________________

Coronal

+ to –

3 ________________

– to +

2 ________________

Anterior

+ to –

1 ________________

– to +

6 ________________

Nasal

+ to –

________________

– to +

________________

Lateral

+ to –

1 ________________

– to +

0 ________________

High

+ to –

6 ________________

– to +

2 ________________

Low

+ to –

________________

– to +

________________

Back

+ to –

2 ________________

– to +

2 ________________

Round

+ to –

0 ________________

– to +

2 ________________

Continuant

+ to –

6 ________________

– to +

0 ________________

Delayed Release

+ to –

2 ________________

– to +

0 ________________

Voiced

+ to –

________________ 1

– to +

________________ 3

Strident

+ to –

________________ 6

– to +

________________ 0

Summary:

–ant. to +ant:  → t,  → d, k → t, g → d, ʃ → d, ʃ → s +high to –high:  → t,  → d, g → d, k → t, ʃ → d, ʃ → s +cont to –cont: f → b, θ → b, ð → d, ʃ → d, s → t, z → t +strident to –strident:  → t,  → d, f → b, ʃ → d, s → t, z → t

[θ] → [t], [s] → [t]). Later, this global substitution of all fricatives by stops might become limited to sibilant fricatives only. Ordering occurs when substitutions that appeared unordered and random become more organized. Ordering can be exemplified by the following: A child’s first revisions may appear unordered. To stay with the stop for fricative example, a child might at first also devoice the voiced stops of the substitution,

ch04.indd 76

thus, ([s] → [t] and [z] → [t]). Thus, Sue is pronounced as [tu], but zoo is also articulated as [tu]. Later, the child might begin to “order” the revisions by voicing initial voiced stops but still retaining the stop substitution. Now Sue is [tu] and zoo is [du]. The term suppression refers to the abolishment of one or more phonological processes as children move from the innate speech patterns to the adult patterns. Suppression occurs when

7/12/2007 8:05:41 AM

THEORETICAL CONSIDERATIONS AND PRACTICAL APPLICATIONS

a previously used phonological process is not used any longer. According to Stampe (1979), all children embark on the development of their phonological systems from the same beginnings. Stampe sees children as possessing a full understanding of the underlying representation of the adult phoneme system: that is, from the very beginning, the child’s perceptual understanding of the phonemic system mirrors the adult’s. Children just have difficulties with the peripheral, motor realization of the phonetic surface form. Many authors have questioned the validity of this idea (e.g., Fey, 1992; Oller, Jensen, and Lafayette, 1978; Stoel-Gammon and Dunn, 1985). In addition, Stampe’s account of phonological development presents children as passively suppressing these phonological processes. Other contemporary authors, notably Kiparsky and Menn (1977), see children as being far more actively involved in the development of their phonological systems. In spite of such shortcomings, Edwards (1992) states that “it is not necessary to totally discard the notion of phonological processes just because we may not agree with all aspects of Stampe’s theory of Natural Phonology, such as his view that phonological processes are ‘innate’ and his assumption that children’s underlying representations are basically equivalent to the broad adult surface forms” (p. 234). Phonological process analysis has found widespread clinical utilization, although it is used not so much to explain developmental speech events, as was the original intent of natural phonologists, as to describe the deviations noted in the speech of children. Because phonological processes are so central to the workings of natural phonology, and to its clinical application, some of the more common processes are listed here with some explanatory remarks.

ch04.indd 77

77

Phonological Processes Although many different processes have been identified in the speech of normally developing children and those with phonological disorders, only a few occur with any regularity. Those processes that are common in the speech development of children across languages are called natural processes. Phonological processes are categorized as syllable structure processes, substitution processes, or assimilatory processes. Syllable structure processes describe those sound changes that affect the structure of the syllable. Substitution processes describe those sound changes in which one sound class is replaced by another. Assimilatory processes describe changes in which a sound becomes similar to, or is influenced by, a neighboring sound of an utterance. Syllable Structure Processes Cluster reduction. The articulatory simplification of consonant clusters into a single consonant, typically the more “natural” member of the cluster. Example: [pun] for spoon. Reduplication. This process is considered a syllable structure process because the syllable structure is “simplified”; that is, the second syllable becomes merely a repetition of the first. Total reduplication refers to the exact reduplication of the first syllable. In partial reduplication, the vowel in the second syllable is varied (Ingram, 1976). Examples: Total reduplication: [wɑwɑ] for water. Partial reduplication: [babi] for blanket. Weak syllable deletion. an unstressed syllable.

The omission of

 Example: [nnə] for ba nana.

7/12/2007 8:05:42 AM

78

CHAPTER 4

Final consonant deletion. The omission of a syllable-arresting consonant. Example: [hε] for head. Substitution Processes Consonant cluster substitution. The replacement of one member of a cluster. Example: [stwit] for street. Note: This is additionally referred to as gliding to indicate the specific type of substitution. Changes in Organ or Place of Articulation Fronting. Sound substitutions in which the organ and/or place of articulation is more anteriorly located than the intended sound. Prominent types include velar fronting (t/k substitution) and palatal fronting (s/ʃ substitution). Examples: [ti] for key; [su] for shoe. Labialization. The replacement of a nonlabial sound by a labial one. Example: [fm] for thumb. Alveolarization. The change of nonalveolar sounds, mostly interdental and labiodental sounds, into alveolar ones. Example: [sm] for thumb. Changes in Manner of Articulation Stopping. The substitution of stops for fricatives or the omission of the fricative portion of affricates. Examples: [tn] for sun; [dus] for juice. Affrication. The replacement of fricatives by homorganic affricates. Example: [tʃu] for shoe. Deaffrication. The production of affricates as homorganic fricatives. Example: [ʃiz] for cheese.

ch04.indd 78

Denasalization. The replacement of nasals by homorganic stops. Example: [dud] for noon. Gliding of liquids/fricatives. The replacement of liquids or fricatives by glides. Examples: [wεd] for red; [ ju] for shoe. Vowelization (vocalization). The replacement of syllabic liquids and nasals, foremost [l], [], and [n], by vowels. Examples: [tebo] for table; [ldυ] for ladder. Derhotacization. The loss of r-coloring in rhotics [r] and central vowels with r-coloring, [] and []. Examples: [bd] for bird, [ldə] for ladder. Changes in Voicing Voicing. The replacement of a voiced for a voiceless sound. Example: [du] for two. Devoicing. The replacement of a voiceless for a voiced sound. Example: [pit] for beet. Assimilatory Processes (Harmony Processes) Labial assimilation. The change of a nonlabial into a labial sound under the influence of a neighboring labial sound.

Assimilation processes can also be classified according to the type and degree of the assimilatory changes. For definitions and examples, see Sounds in Context: Coarticulation and Assimilation in Chapter 2.

Example: [fwŋ] for swing. Velar assimilation. The change of a nonvelar into a velar sound under the influence of a neighboring velar sound. Example: [ɑ] for dog. Nasal assimilation. The influence of a nasal on a nonnasal sound.

7/12/2007 8:05:42 AM

THEORETICAL CONSIDERATIONS AND PRACTICAL APPLICATIONS

Example: [mni] for bunny. Note: The place of articulation is retained; only the manner is changed. Liquid assimilation. The influence of a liquid on a nonliquid sound. Example: [lεloυ] for yellow. According to natural phonology, phonological processes are recognizable steps in the gradual articulatory adjustment of children’s speech to the adult norm. Ages of suppression This implies a chronology of the various proof phonological processes, cesses are discussed specific ages at which the in Chapter 5. process could be operating and when the process should be suppressed (Grunwell, 1981, 1987; Vihman, 1984). As useful as a chronology of normative data might seem for clinical purposes, tables of established age norms can easily be misleading. Individual variation and contextual conditions may play a large role in the use and suppression of phonological processes. To summarize, natural phonologists assume an innate phonological system that is progressively revised during childhood until it corresponds with the adult phonological output. Limitation, ordering, and suppression are the mechanisms for the revisions that manifest themselves in phonological processes. Phonological processes are developmentally conditioned simplifications in the realization of the phonological system in question. As these simplifications are gradually overcome, the phonological processes become suppressed. How Did Natural Phonology Develop? David Stampe introduced natural phonology in 1969. However, several of its basic concepts had been established considerably earlier, most prominent among them being the concepts of naturalness (markedness) and under-

ch04.indd 79

79

lying and surface forms, which are important aspects of generative phonology. Jakobson (1942/1968) extended the concept of naturalness and markedness to implied universals, which could be found in different languages, children’s acquisition of speech, and the deterioration of speech in aphasics. These universals were even used as a predictive device. Some examples include “fricatives imply stops” and “voiced stops imply voiceless stops.” These examples would mean that if a language has fricatives, that language will have stops as well, and if a language has voiced stops in its inventory, the language will also have voiceless stops. Applying these two examples to children’s acquisition of speech, children will acquire stops before (homorganic) fricatives. Also, voiceless stops are acquired before their voiced cognates. In an aphasic condition, the breakdown of speech would be characterized by the loss of the lateracquired sounds before the earlier-acquired ones. Thus, aphasics would lose fricatives before (homorganic) stops and voiced stops before voiceless ones. Whether these universal “laws” are generally valid under all of the previously mentioned conditions has been repeatedly questioned. However, they clearly exemplify the concepts of naturalness and markedness as universal phenomena. Markedness theory also plays a central role in generative phonology and optimality theory (Prince and Smolensky, 1993; McCarthy and Prince, 1995). According to generative phonologists, markedness values are considered to be universal and innate. Thus, Jakobson, with his concept of universal naturalness, and Chomsky and Halle, with their understanding of universal and innate naturalness, set the stage for Stampe’s natural phonology. Stampe incorporated the conceptual framework of naturalness into his theory of natural phonology.

7/12/2007 8:05:42 AM

80

CHAPTER 4

At the same time, the meaning and use of the term underlying form changed drastically as it was incorporated into natural phonology. Within generative grammar, underlying forms, lexical as well as phonological, are highly abstract entities. They represent assumed points of reference that are necessary for the explanation of the many possible surface forms. In contrast, within the context of natural phonology, underlying forms as “models” for surface realizations suddenly gain some concrete reality. The underlying form is the adult norm that is the intended goal for children’s production efforts. Natural Phonology: Clinical Application The concept of phonological processes within natural phonology has impacted both the assessment and the treatment of disordered phonological systems. Assessment procedures using phonological processes consist of contrasting the target word to the child’s production. Aberrant productions are identified and labeled according to the phonological process that most closely matches the sound change. Typically, the processes are listed and the frequency of occurrence of individual processes is noted. Frequency of occurrence and the relative age of suppression play a role in targeting a process or processes for therapy. Depending on the age of the child, more frequent processes that should have been suppressed are commonly targeted for therapy. Some authors (Hodson and Paden, 1991; McReynolds and Elbert, 1981) suggest that a process should occur a certain number of times in order for it to be considered a possibility for therapy. Unlike other analysis procedures, phonological processes can account for changes in syllable or word structures and those due to assimilations. Although phonological processes are not commonly used to identify sound distortions, they could be. For example, [ s ] could be labeled fronting and [s] backing.

ch04.indd 80

Phonological Process Analysis. A phonological process analysis is a means of identifying substitutions, syllable structure, and assimilatory changes that occur in the speech of clients. Each error is identified and classified as one or more of the phonological processes. Patterns of error are described according to the most frequent phonological processes present and/or to those that affect a class of sounds or sound sequences. The processes utilized to identify substitutions are again primarily productionally based; however, they do account for sound and syllable deletions as well as several assimilation processes. There are several assessment protocols that analyse phonological processes in articulation tests or in spontaneous speech (Bankson and Bernthal, 1990; Dean, Howell, Hill, and Waters, 1990; Grunwell, 1985a; Hodson, 2004; Ingram, 1981; Khan and Lewis, 2002; Lowe, 1996; Shriberg and Kwiatkowski, 1980). All of them identify each aberrant production according to the phonological process or processes that best represent the changes that have occurred. Most protocols also summarize the phonological processes by counting the total number of specific processes. Table 4.6 represents a protocol for summarizing the established phonological processes from the articulation test and the spontaneous speech sample. To analyze a speech sample according to phonological processes: 1. Identify the phonological process that best describes the change. More than one phonological process might apply to a given misarticulation. For example, if a child substitutes [d] for [s] ([s] → [d]), this needs to be identified as stopping and voicing. 2. Tally the number of times the child used each process. On the summary form, list the processes and their frequency of occurrence. The phonological processes and their frequency of occurrence for H. H. are contained in Table 4.7. A word-by-word analyis of his

7/12/2007 8:05:42 AM

81

THEORETICAL CONSIDERATIONS AND PRACTICAL APPLICATIONS

4.6

Table Phonological Process Analysis Summary Sheet

Summary of Phonological Processes for Child H. H.

Processes

Number of Occurrences

Processes

Syllable Structure Changes Cluster reduction

_______________

Syllable Structure Changes Cluster reduction

Number of Occurrences 16

Cluster deletion

_______________

Cluster deletion

1

Reduplication

_______________

Reduplication

0

Weak syllable deletion

_______________

Weak syllable deletion

2

Final consonant deletion

_______________

Final consonant deletion

17

Initial consonant deletion

_______________

Initial consonant deletion

1

Other __________________

_______________

Other __________________

Substitution Processes Consonant cluster substitution Fronting

_______________ _______________

Substitution Processes Consonant cluster substitution Fronting

Labialization

_______________

Labialization

Alveolarization

_______________

Alveolarization

9 15 5 1

Stopping

_______________

Stopping

20

Affrication

_______________

Affrication

0

Deaffrication

_______________

Deaffrication

0

Denasalization

_______________

Denasalization

0

Gliding of liquids

_______________

Gliding of liquids

7

Gliding of fricatives

_______________

Gliding of fricatives

0

Vowelization

_______________

Vowelization

0

Derhotacization

_______________

Derhotacization

7

Voicing

_______________

Voicing

Devoicing

_______________

Devoicing

Other __________________

_______________

Other __________________

__________________

_______________

__________________

__________________

_______________

__________________

Labial assimilation

_______________

Assimilation Processes Labial assimilation

Velar assimilation

_______________

Velar assimilation

Nasal assimilation

_______________

Nasal assimilation

Liquid assimilation

_______________

Liquid assimilation

Other __________________

_______________

Other _________________

_________________

_______________

_________________

_________________

_______________

_________________

Assimilation Processes

ch04.indd 81

Table 4.7

10 3

7/12/2007 8:05:43 AM

82

CHAPTER 4

phonological processes are contained in Appendix 4.1 at the end of this chapter. As can be seen from Table 4.7, H. H. demonstrates only five different processes 10 or more times: voicing (= 10 times), fronting (= 15 times), cluster reduction (= 16 times), final consonant deletion (= 17 times), and stopping (= 20 times). If the articulation test results are examined (Table 4.4 or Appendix 4.1), one can note that final consonant deletion impacts some of the fricatives, the stop-plosives, the nasals, one of the affricates, and the lateral [l], while stopping affects the fricatives and affricates. On the other hand, fronting is limited to [k], [], and [ ʃ ]. Phonological processes can be used to analyze substitutions and deletions, something that distinctive feature analysis was not able to do. In addition, phonological process analysis can generate patterns, by noting the most frequent processes, and allows you to examine the sounds or sound classes that are most frequently included in the various phonological processes. The next section will introduce the more recent developments in phonological theories, the so-called nonlinear or multilinear phonological theories. They represent a radical departure from the conceptual framework that preceded them.

LINEAR VERSUS NONLINEAR PHONOLOGIES What Are Linear and Nonlinear Phonologies? Phonological theories, theories of generative phonology included, were based on the understanding that all speech segments are arranged in a sequential order. Consequently, underlying phonological representations and surface phonetic realizations, too, consisted of a string of discrete elements. For example:

ch04.indd 82

Wow, what a test. [waυ wt ə tεst] The sequence of segments in this phrase begins with [w] and ends with [st]. All segments in between follow each other in a specific order to convey a particular message. Such an assumption that all meaning-distinguishing sound segments are serially arranged characterizes all linear phonologies. Linear phonologies, exemplified by distinctive feature theories and early generative phonology, can be characterized as follows: 1. emphasis on the linear, sequential arrangement of sound segments, 2. assumption that each discrete segment of this string of sound elements consists of a bundle of distinctive features, 3. assumption that a common set of distinctive features is attributable to all sound segments according to a binary + and – system, 4. assumption that all sound segments have equal value and all distinctive features are equal; thus, no one sound segment has control over other units, 5. the phonological rules generated apply only to the segmental level (as opposed to the suprasegmental level) and to those changes that occur in the distinctive features (Dinnsen, 1997). Linear phonologies with sound segments (and their smaller distinguishing distinctive features) as central analytical units fail to recognize and describe larger Hierarchy refers to linguistic units. Linear any system in which phonologies also do not elements are ranked account for the possibility one above another. that there could be a hierarchical interaction between segments and other linguistic units. Nonlinear or nonsegmental phonologies attempt to account for these factors. Nonlinear (or what have been termed multilinear) phonologies are a group of phonological theories understanding segments

7/12/2007 8:05:44 AM

THEORETICAL CONSIDERATIONS AND PRACTICAL APPLICATIONS

as governed by more complex linguistic dimensions. The linear representation of phonemes plays a subordinate role. More complex linguistic dimensions—for example, stress, intonation, and metrical and rhythmical linguistic factors—may control segmental conditions. These theories explore the relationships among units of various sizes, specifically the influence of larger linguistic entities on sound segments. Therefore, rather than a linear view of equal-valued segments (in a left-to-right horizontal sequence), a hierarchy of factors is hypothesized to affect segmental units. Rather than a static sequence of segments of equal value (as in linear phonology), a dynamic system of features, ranked one above the other, is proposed. For example, syllable structure could affect the segmental level. A child may demonstrate the following pattern: “man” [mn] “window” [w doυ] “dog” [dɑ] “jumping” [d p] “ball” [bɑl] “Christmas tree” [kr mə tri] This child deletes the final consonant of each syllable in a multisyllabic word; however, no final consonant deletion occurs in onesyllable words. In this example, the number of syllables in a word interacts with and affects the segmental level. The number of syllables has priority over the segmental level: It determines segmental features. Nonlinear phonologies would rank syllable structure above the level of sound segments. Another factor that may affect the segmental level is stress. Children have a tendency to delete segments in unstressed syllables. The following transcriptions demonstrate this: banana potato telephone

→ → →

[nnə] [te toυ] [tε foυn]

In these examples, the syllable stress clearly affects segmental realization; word stress has priority over the segmental level. “Instead of

ch04.indd 83

83

a single, linear representation (one unit followed by another with none having any superiority or control over other units), they [nonlinear phonologies] allow a description of underlying relationships that would permit one level of unit to be governed by another” (Schwartz, 1992b, p. 271). How Do Nonlinear Phonologies Work? There are many different types of nonlinear/ multilinear phonologies. Several new theories have been advanced and others have been modified. All nonlinear phonologies are based on a belief in the overriding importance of larger linguistic units influencing, even controlling, the realization of smaller ones. Nonlinear phonologies also attempt to incorporate this hierarchical order of linguistic elements into analytical procedures, using so-called tiered representations of features. To describe the many different nonlinear phonologies is beyond the scope of this book. The New Phonologies: Developments in Clinical Linguistics (Ball and Kent, 1997) is an excellent source of more detailed information on autosegmental phonology, feature geometry, underspecification theory, dependency phonology, government phonology, grounded phonology, optimality theory, and gestural phonology. This section will be restricted to an introduction to nonlinear phonologies exemplified by autosegmental, metrical, feature geometry, and optimality theories. These theories are in no way superior to other nonlinear phonologies. Autosegmental Phonology. Autosegmental phonology was proposed by John Goldsmith in 1976. Originally, Goldsmith presented this theory to account for tone phenomena in languages in which segmental features interact with varying tones. Parker (1994) illustrates the essential problem in the following manner:

7/12/2007 8:05:44 AM

84

CHAPTER 4

According to the concept of linear (generative) phonology, features extend throughout a segment. Therefore, a segment such as /p/ is considered to be [– voice] throughout its entire segment, while /u/ is [+ voice] throughout its entirety. However, consider the problem posed by affricates. By definition, an affricate begins like a stop and ends like a fricative. The features that differentiate stops and fricatives are + and – continuant. This posed a problem for the linear phonologists because one segment cannot be designated as both + and – one distinctive feature. To solve this problem, the linear phonologists constructed the feature of “delayed release” to designate affricates. However, the feature of delayed release violates the construct of distinctive feature theory in that this property does not extend throughout the entire segment. Autosegmental phonology proposed that changes within the boundary of a segment could be factored out and put onto another “tier.” Thus + and – continuant could be placed on another level to indicate the change within the segment boundary. A diagram of an affricate such as /tʃ/ would look accordingly:

Tone languages, which represent a large number of the languages of the world, are distinguished by changes in the meaning of a word simply by changing the pitch level at which it is spoken. Thus, phonemic differences can be signaled by distinctive pitch levels known as tones or tonemes (Crystal, 1987). For example, in Mandarin Chinese, four different tones with the identical sound segments [ma] result in words that mean “mother,” “hemp,” “horse,” and “scold.” Autosegmental phonology placed these tones on a tier above the sound segments, demonstrating the overriding importance of these tones for the meaning of the word.

explain one-to-many mappings (one tone associated with more than one segment) and manyto-one mappings (more than one tone associated with one segment). However, this “tiered” organization can demonstrate many characteristics of children’s speech as well—relationships between certain syllable types and production of sound segments, for example. For an understanding of autosegmental phonology, specific terms need to be defined: Tiers

Separable and independent levels that represent a sequence of gestures or a unified set of acoustic features.

Association lines

Indicators for connections between autosegments on different tiers. Association lines cannot cross.

Linkage condition

Any condition governing the association of units on each tier. A linkage condition states, for instance, that if a segment is not linked to a position on another tier, it will not be phonetically realized.

Skeletal (or CV) tier

A representation of a syllable and its hierarchically related components’ onset and rhyme.

Onset

Onset of a syllable. Includes all segments before the nucleus.

Continuant-tier [– continuant] [+ continuant] Other features

+ consonantal – vocalic • • •

= tʃ

As can be seen, a single segment on one tier can be associated with more than one segment on another tier. Using the example of /tʃ/, the + consonantal segment can be associated with + and – continuant on another tier. In fact, the term autosegmental refers to the concept that certain segments are autonomous—they do not have a one-for-one match on another level. As mentioned earlier, Goldsmith’s (1976) dissertation addressed tone phenomena in socalled tone languages. This concept was used to

ch04.indd 84

7/12/2007 8:05:44 AM

85

THEORETICAL CONSIDERATIONS AND PRACTICAL APPLICATIONS

Rhyme

Cover term for nucleus (vowel) and coda (the arrest of the syllable).

1. [+ round] spreads from a vowel to adjacent consonant. soy sauce

The following diagram depicts the skeletal tier of a CVC syllable:

[s ɔ s ɑ s]

Skeletal or CVC Tier

[+ round] [– round] 2. [+ round] spreads from a consonant to an adjacent consonant up to a vowel.

Onset

twenty

Rhyme

[t w ε n t i] Nucleus

Coda

V

C

C

[+ round] [– round]

The following is an example of a child who deleted final consonants in two-syllable words. The following diagram illustrates this relationship: Word (CVCCVC)

Onset

Rhyme

Onset

Nucleus Coda C

V



Rhyme

CLINICAL APPLICATION

Nucleus Coda C

V



Autosegmental phonology also accounted for feature spreading. Certain features such as + and – rounding, for example, can spread to other vowels and consonants. There are two rules for spreading. First, + and – round spreads from a vowel to adjacent consonants. Second, + and – round spreads from a consonant to an adjacent consonant up to a vowel. The following examples demonstrate the two types of feature spreading. The solid line is an inherent feature, while the dotted line represents a spread feature specification:

ch04.indd 85

Feature spreading also occurs with such features as [+ voice] and [+ nasal]. To summarize, autosegmental phonology was originally conceived to account for cases in which a single segment is associated with two mutually exclusive features. It has since been expanded to demonstrate relationships between certain syllable types and consonant realizations. Feature spreading accounts for examples in which the feature or property of one segment spreads to adjacent segments.

Using Autosegmental Phonology to Analyze an Error Pattern The following autosegmental chart is for a child who produces all initial consonant clusters (two- and threesound clusters) as [d]: Word (CCVC or CCCVC)

Onset

[d]

Rhyme

Nucleus

Coda

V

C

7/12/2007 8:05:45 AM

86

CHAPTER 4

Metrical Phonology. Metrical phonology (Liberman, 1975; Liberman and Prince, 1977) extended a hierarchical-based analysis to stress. In linear phonology, for example, stress was not handled in a binary + and – way; rather, there were an infinite number of prominence values that could be assigned to stress. The stress assignment rules of linear phonology produced a relative ordering within any given string of sound segments. This relative ordering can be used (1) to analyze the relative stressing of individual words within a sentence (sentence stress) as well as (2) to analyze the relative stressing of syllables within a word (word stress). The following example demonstrates the linear phonology stress assignment of individual words (word stress) and when these words are placed within a sentence. The numeral 1 indicates the primary stress: Word stress

stress. “Metrical trees” are used to reflect the syntactic structure of an utterance. To show the relative prominence of each constituent in an utterance, stress patterns are represented by a binary branching of these metrical trees. One branch is labeled S for “stronger” stress and the other W for “weaker” stress. Applying this principle to an example, the following metrical tree can be drawn:

W

S

big

brother

Thus, every tree in metrical phonology must have either a W S or an S W branching. This renders a binary stress representation. If the phrase is expanded, the following stress pattern emerges:

a. customer 1

3 2

b. services 1

W

Sentence stress c. customer services 1

2

d. He is the supervisor of customer services. 4

6 5

1

7

2

3

This system of assigning stress to words within phrases appeared inadequate to Liberman and Prince. For example, the words customer services are assigned two different values in examples c and d even though the same words are used. Stress assignment rules in linear (generative) phonology were relational and changed depending on the prominence given to the words within a phrase (Hogg and McCully, 1989). Metrical phonologists proposed another concept for understanding and analyzing

ch04.indd 86

S

3 2

John’s

W

S

big

brother

This pattern indicates that brother has more prominence than big and that the phrase big brother is more stressed than John’s. The same relationship, then, is maintained between big and brother in both metrical representations. The second basic concept in metrical phonology pertains to the syllable. Although word boundaries are indicated in generative phonology, the syllable structure was not considered. Metrical phonologists indicate not only the number of syllables within a word but also which consonants belong (or are hypothesized to belong) to each syllable.

7/12/2007 8:05:45 AM

THEORETICAL CONSIDERATIONS AND PRACTICAL APPLICATIONS

87

The notation uses the Greek letter sigma (σ) to indicate the individual syllables: σ

1

σ

2

σ

W

S

W

S

3 S

W

ε k s a t ə d This hierarchical arrangement can also be used to include the morphological representation of the word together with its syllabic divisions. The Greek symbol mu (µ) denotes the morphemes within this word: σ

σ

σ

1

2

3

ε k s a t ə d µ1

µ2

µ3

Such an analysis clearly indicates the difference between syllabic and morphological boundaries. To summarize, metrical phonology is a theoretical construct that extends hierarchical analysis procedures to stress and syllable boundaries. Stress is analyzed according to a binary “strong” and “weak” system rather than to a relational numbering system that was used by earlier phonologists, including the linear (generative) phonologists. This hierarchical analysis has also been used when dividing words into syllables. Syllabic analyses allow for comparisons between syllable and morpheme boundaries. CLINICAL APPLICATION Using Metrical Phonology to Analyze an Error Pattern The following metrical tree demonstrates a child’s deletion of unstressed syllables in the two-syllable word above and the three-syllable word umbrella:

ch04.indd 87

ə

b v

əm

brε



Ø

b v

Ø

brε



Feature Geometry. Feature geometry represents a group of theories that have adopted the tiered representation of features used in autosegmental phonology. However, feature geometry theories have added a number of other hierarchically ordered feature tiers. Feature geometry attempts to explain why some features (and not others) are affected by assimilation processes (known as spreading or linking of features) while others are affected by neutralization or deletion processes (known as delinking) (Dinnsen, 1997). There are several different tier representations in feature geometry. Figure 4.3 is a feature geometry representation that was provided by Bernhardt and Stemberger (1998) based on the proposals of Bernhardt (1992a, 1992b), Clements (1985), McCarthy (1988), and Sagey (1986). In accordance with Note that distinctive principles of nonlinear features also play a phonologies, feature ge- central role in the ometry also utilizes hierar- newer nonlinear/ chically organized levels of multilinear phonologies. According to representation, so-called Bernhardt and Stemtiers. These tiers interact berger (1998), the with one another. Some distinctive features features are designated as for feature geometry are based on those of nodes, which means that Chomsky and Halle they may dominate more (1968), except for than one other feature and the features for place serve as a link between the of articulation, which follow Sagey (1986). dominated feature and

7/12/2007 8:05:45 AM

88

CHAPTER 4

Figure 4.3

Feature Geometry of the English Consonant System

Source: From Handbook of Phonological Development from the Perspective of Constraint-Based Nonlinear Phonology (p. 92), by B. H. Bernhardt and J. P. Stemberger, 1998, San Diego, CA: Academic Press. Copyright 1998, reprinted with permission from Elsevier Science.

higher levels of representation. For example, in Figure 4.3, the Place node serves as a link between the Labial, Coronal, Dorsal, and Radical nodes and the Root node. Features at a higher level of representation are said to dominate other features. The Place node, for example, dominates the different places of articulation (Labial, Coronal, and Dorsal nodes). The Place node must be activated, so to speak, before a specific place of articulation can be chosen. Or, the Laryngeal node as a higher level of representation must be functioning before [+ voice] can be designated. Features that are dominated are considered to be subordinate or at a lower level of representation. The following is a brief explanation of the different nodes and features, summarized from Bernhardt and Stemberger (1998).

ch04.indd 88

Laryngeal features 1. [+ voiced] sounds produced with vocal fold vibration (e.g., [d], [i]). 2. [+ spread glottis] the vocal cords are spread wide, leading to low-amplitude voice at the glottis (e.g., voiceless aspirated stops [thim] are + s.g. [h], as well as [f], [θ], [s], and [ ʃ ]). 3. [+ constricted glottis] the vocal cords are pulled together tightly, so that regular periodic vibration is impossible (e.g., all + c.g. segments are – voiced, glottal stops are + c.g.). Manner features 1. [+ sonorant] sounds in which the pressure above the larynx allows the vocal cords to vibrate continuously, without any rise

7/12/2007 8:05:46 AM

THEORETICAL CONSIDERATIONS AND PRACTICAL APPLICATIONS

2.

3.

4.

5.

6.

in pressure above the larynx (e.g., voiced vowels, glides, liquids [r] and [l], [h], and nasals are + sonorant). [+ consonantal] sounds with a narrow constriction in the oral and/or pharyngeal cavities that significantly impede the flow of air (e.g., stops, affricates, nasals, fricatives, laterals, taps, and trills are + consonantal). [+ continuant] sounds in which air continues to move through the oral cavity (e.g., vowels, glides, liquids, and fricatives). [+ nasal] sounds with the velum lowered so that air moves through the nasal cavity (e.g., nasals). [+ lateral] sounds in which central airflow is blocked in the oral cavity, but in which air is directed over at least one side of the tongue (e.g., laterals). [+ tense] sounds produced with relatively greater “muscular tension” (e.g., tense vowels, voiceless obstruents). Place features

Lips 1. [Labial] sounds made with more involvement of one or both lips (e.g., bilabials [p, b, m], labiodentals [f, v], and [r], [w] are + labial). 2. [+ round] sounds involving protrusion of the lips with narrowing at the corners of the mouth (e.g., all rounded vowels and labialized consonant [kw], [r], and [w] are + round. Bilabials and labiodentals are – round). 3. [+ labiodental] labial sounds that are made with only one lip (e.g., [f ] and [v]). The tip of the tongue 1. [Coronal] sounds made with raising of the tip or blade of the tongue (e.g., [t, d, s, z, ʃ, , θ, ð, n, r, l, and j] plus high-front vowels are included). 2. [+ anterior] coronal sounds made at the alveolar ridge or further forward (e.g., [t, d,

ch04.indd 89

89

θ, ð, n, l] are + anterior, – anterior includes [ ʃ, , tʃ, d, r, and j], and front vowels). 3. [+ distributed] coronal sounds made with a wide area of contact between the tip/ blade of the tongue and the roof of the mouth or teeth (e.g., [tʃ, d, f, v, ʃ, , r, and j] are + distributed, – distributed sounds include [t, d, s, z, n, and l]). 4. [+ grooved] coronal sounds made with a grooved tongue, a narrow channel at or near the midline (e.g., alveolar fricatives [s, z, ʃ, ] and affricates). The tongue body 1. [Dorsal] sounds made with the back of the tongue (e.g., [k, , ŋ], back vowels, [w], and [ j ] also dark [l]). 2. [+ back] sounds with the back of the tongue body raised or lowered (e.g., velar sounds [k, , ŋ, w], including the dark [l], back and central vowels). 3. [+ high] sounds where the tongue body is raised (e.g., high vowels, [k, , ŋ, w, and j]. 4. [+ low] sounds where the tongue body is lowered (e.g., low vowels). The tongue root 1. [Radical] sounds in which the root of the tongue is advanced or retracted (e.g., pharyngeal and pharyngealized consonants, not typical for American English speech sounds). 2. [Advanced Tongue Root (ATR)] sounds in which the tongue root is advanced (e.g., high vowels, [i], [e], [u], and [o] are + ATR, consonants are blank for this feature). As can be seen from the preceding explanation, the use of features and the definition of certain features are different from those proposed by Chomsky and Halle (1968). Another nonlinear theory, the theory of radical underspecification (Archangeli, 1988; Archangeli and Pulleyblank, 1994; Bernhardt, 1992b; Kiparsky, 1982; Pulleyblank, 1986),

7/12/2007 8:05:46 AM

90

CHAPTER 4

suggests that underlying representations contain only “unpredictable” features. A predictable feature is one that would be commonly associated with that particular segment or class of sounds. For example, nasals are typically voiced (although there are unvoiced nasals in some African languages). Voicing for nasals is, then, predictable and would not be contained in the underlying representation. Or, because all sonorants are voiced, this is again a predictable feature and is not contained in the underlying representation. On the other hand, obstruents can be [+ voice] or [– voice]; therefore, the unpredictable nature of this feature is contained in the underlying representation. Rules in nonlinear analysis are restricted to two basic operations: spreading (known as linking) and deletion (known as delinking) of phonological information from one tier to another. Spreading of features could be exemplified by the production of [k] for duck. The coronal place node for /d/ is subject to linking or assimilation from the dorsal place node feature of /k/. Thus, the dorsal place node of the final [k] in duck affects the initial [d]. The end result is that the initial [d] is produced as []. The place of articulation is moved from coronal [d] to dorsal []. Delinking could be exemplified by the production of [d] for duck. Under the assumption that the underlying representation is intact, the final consonant slot for that production is delinked from the representation along with the actual features of /k/. Linking and delinking result from, and are constrained by, principles of association between tiers. These principles are outlined in Bernhardt and Stemberger (1998) and could be used as a reference for more detailed analysis procedures. To summarize, one nonlinear phonology, feature geometry, theorizes that segments are composed of multitiered hierarchically organized features. Specific nodes that can dominate other features and link various levels of

ch04.indd 90

representation are designated. According to this theory, features can link (assimilate) or delink, causing neutralization or deletion. Principles of association are used to explain occurrences between tiers. Optimality Theory. Optimality theory, first formalized by Prince and Smolensky (1993) and McCarthy and Prince (1995), is considered a constraint-based approach. Constraints are a limit to what constitutes a possible pronunciation of a word (Stemberger and Bernhardt, 1997). When constraints are applied linguistically, a set of grammatical universals is said to exist that includes the fact that all languages have syllables and that certain syllable patterns seem to be more (or less) common. For example, in General American English, there are words that begin with three consonants, such as street, but not any that begin with four consonants in a row. Therefore, we could say that American English has a constraint on how many consonants can occur at the beginning of a word; three consonants are acceptable, but four are not. Languages will demonstrate certain constraints if compared to one another. For example, Hawaiian allows no more than one consonant in a row, resulting in words such as kanaka for “man.” When comparing this to English, which allows several consonants in a row, in such words as street and sixths, we could say that Hawaiian has a constraint against more than one consonant as an onset or as a syllable coda. Constraints characterize patterns that are and are not possible within or across languages. Applying this principle generally to children with articulatoryphonological disorders, it could be stated that a child who does not produce syllable codas, thus evidences final-consonant deletion, has a constraint against producing final codas. Constraints are based on principles of markedness. Thus, each constraint violation indicates markedness in that respect. Con-

7/12/2007 8:05:46 AM

91

THEORETICAL CONSIDERATIONS AND PRACTICAL APPLICATIONS

straints are a means of Markedness is (1) characterizing uni- discussed on pages versal patterns that occur 69–70 and 74 of this chapter. across languages, (2) demonstrating variations of patterns that occur between languages, and (3) determining markedness indicated by constraint violations (Archangeli and Langendoen, 1997). Optimality theory, as a constraint-based approach, was originally developed to explain the differences that occur between languages. Optimality theory presupposes a Universal Grammar and states that constraints characterize universals; however, constraints can be violated. Some constraints are very important (within and across languages) and are rarely violated, whereas others are not as important and can be violated. In this sense, constraints are violable. If we examine constraints in this manner, we will find that the following universal trends are considered typical (unmarked) properties of syllables. To the right of the constraint is the name given to it according to Archangeli and Langendoen (1997): Syllables begin with a consonant. ONSET Syllables have one vowel.

PEAK

Syllables end with a vowel.

NOCODA

Syllables have at most one consonant at an edge.

*COMPLEX

In examining this list, we can see whether there are constraint violations in American English. 1. Syllables begin with a consonant: ONSET. Not all syllables begin with a consonant, as demonstrated by words such as away and eat. Probably in General American English most syllables do, however, begin with a consonant. This is a violable constraint (although it is maintained most of the time) in American English.

ch04.indd 91

2. Syllables have one vowel: PEAK. In American English this seems to be the case all the time. Some syllables consist of syllabic consonants such as [bi.tl ] or [fʃ.n ]; however, no syllables contain two vowels. In American English this is a constraint that is rarely, if at all, violated. 3. Syllables end with a vowel: NOCODA. Not all syllables in American English end with a vowel. Many syllables end with a consonant in words such as hat, clock, and antique. This constraint is violated in American English. 4. Syllables have at most one consonant at an edge: *COMPLEX. This is also violated in American English. Words such as clocks and streets demonstrate a violation of this constraint. In summarizing, we could state that some of the previously mentioned constraints are violated while others are not. This could lead to a rank ordering of constraints from those constraints that are never or rarely violated → to those that are sometimes violated → to those that are often violated. Those constraints that are rarely violated are considered higher-order constraints and are separated from others by a double arrow >>. Those violable constraints are separated from each other by a comma. Based on the previous discussion, the following rank ordering could be made: PEAK>>ONSET, NOCODA, *COMPLEX Thus, in American English the constraint PEAK (syllables have one vowel) is not violated. Therefore, it is separated from the others ONSET, NOCODA, *COMPLEX by >>. The others, which can be violated, are separated by commas. Therefore, one important concept within optimality theory is the rank ordering of the constraints.

7/12/2007 8:05:46 AM

92

CHAPTER 4

Optimality theory, like other linguistic theories, proposes an input (an underlying representation), an output (the surface representation), and a relation between the two. The only specification of the input is that it is linguistically well formed; it does not contain variables that are not grammatical. The output is the actual production. Optimality theory does not account for differences between the input and output in terms of rules (as in generative grammar) or processes (as in natural phonology), but in terms of constraints. In optimality theory the relation between the input and output is mediated by two formal mechanisms: the generator (GEN) and the evaluator (EVAL). The generator links the input with potential outputs. It can add, delete, or rearrange, for example. The evaluator judges the outputs to determine which one is the optimal output. For any given input, such as [p], which is the mental representation of the word pig, the GEN can generate an infinite number of possible phonetic outputs for that form. All these output forms compete with one another, but one output must be chosen as the optimal one. The EVAL evaluates all these different outputs and chooses the output that is the optimal response for that particular language. These output forms are evaluated through the constraints and their ranking within that language. The constraints and their relative rankings, thus, restrict the possible output forms (Ball, 2002; Barlow, 2001). There are two types of constraints functioning within this mechanism: faithfulness and markedness. Faithfulness constraints require that input and output forms be identical to one another. If segments between the input and output are deleted, inserted, or rearranged, the faithfulness constraint is violated. If a child produces the word skip as [sp], then the faithfulness constraint has been violated. Markedness constraints require outputs to be unmarked or

ch04.indd 92

simplified in structure. Unmarked features are those that are easier to perceive or produce or those that occur frequently across languages. Consonant clusters are considered to be marked (see *COMPLEX mentioned previously). Thus, the child who produces skip as [skp] violates the markedness constraint. However, a child who says the word skip as [sp] has not violated this constraint; the output is unmarked or simplified. As can be seen, faithfulness and markedness constraints are conflicting; there is an antagonistic relationship between the two. The conflict between faithfulness and markedness leads to violation of constraints, or what is termed constraint violability. Every utterance violates some constraint; if faithfulness is maintained, then markedness is violated. (The most unmarked syllable would be something like [bɑ], so any more complex syllable structure would be in some violation of markedness.) So how does the EVAL judge which one is the most optimal form? At this point the theory postulates that the rank ordering of the constraints becomes the deciding factor. Lower-ranked constraints can be violated to satisfy higher-ranking constraints. In our previous example, ONSET, NOCODA, or *COMPLEX could be violated to satisfy PEAK. If this theory is applied to phonological development, the hypothesis is that children acquire the correct ranking of the constraints as they develop. Immature patterns demonstrate that this ranking, according to the language in question, has not yet been mastered. Individual patterns of normal development are seen as products of the individual’s idiosyncratic constraint rankings. Applying this to children with phonological disorders, these children also have their own unique constraint rankings. Our job is to find out the rankings that would then account for their error patterns. The next step is to try to rerank

7/12/2007 8:05:47 AM

93

THEORETICAL CONSIDERATIONS AND PRACTICAL APPLICATIONS

the constraints so that they are more in line with the input. It is assumed that markedness constraints (thus, the typical simplification that occurs in relationship to the production features) must be demoted. Demotion is a process where higher-ranking constraints that do not match the adult rankings become lower— that is, they become more easily violated. If they are more easily violated, then the rankings will eventually match the adult ones. The names of constraints and how they are abbreviated from text to text varies. Table 4.8 is a list of possible constraints that is summarized from Barlow (2001). Optimality theory uses tableaus to demonstrate the rank order of constraints. Tableaus are boxes with the word listed on the far left followed by the rank-ordered constraints.

Table 4.8

Higher-ranking constraints are to the left. The following tableau demonstrates the ranking of constraints if a child were to say [p] for pig. /pg/ pig

*CODA

a.

*!

[pg]

b. ☞ [p]

MAX

*

☞ = optimal output, optimal candidate, or the one with the fewest, lowest violations; * = constraint violation; *! = fatal violation (a violation that eliminates a candidate completely).

Here, the optimal output refers to the child’s output, not the adult form. As can be seen from this tableau, the markedness constraint CODA (no final consonants, no coda) is ranked higher than the faithfulness constraint

Markedness and Faithfulness Constraints, with Examples of Violations and Nonviolations

Constraint

Definition

Violation

Nonviolation

Markedness *COMPLEX *CODA *FRICATIVES *LIQUIDS

No clusters No final consonants (no codas) No fricatives No liquids

*LIQUID-[l] *LIQUID-[r]

No liquid [l] No liquid [r]

sweep → [swip] cat → [kt] sun → [s n] lake → [lek] rain → [ren] lake → [lek] rain → [ren]

sweep → [sip] cat → [k] sun → [t n] lake → [wek] rain → [wen] lake → [wek] rain → [wen]

Faithfulness MAX

No deletion

DEP IDENT-FEATURE

No insertion Don’t change features

IDENT-[cons] IDENT-[cont]

Don’t change [consonantal] Don’t change [continuant]

cat → [k] sweep → [sip] sweep → [səwip] lake → [wek] sun → [t n] lake → [wek] sun → [t n]

cat → [kt] sweep → [swip] sweep → [swip] lake → [lek] sun → [s n] lake → [lek] sun → [s n]

Source: From “Case Study: Optimality Theory and the Assessment and Treatment of Phonological Disorders,” by J. Barlow, 2001, Language, Speech, and Hearing Services in Schools, 32, p. 245. Copyright © 2001 by the American SpeechLanguage-Hearing Association. Reprinted by permission.

ch04.indd 93

7/12/2007 8:05:47 AM

94

CHAPTER 4

of MAX (no deletions) as the child has violated the faithfulness constraint MAX. On the other hand, if one examines the norm adult pattern, MAX would be ranked higher than CODA. In this rather simplified example, the CODA constraint must be demoted to obtain final consonants. Optimality theory offers a new way of viewing both the acquisition of phonological patterns and the categorizing of disordered phonological systems. The concept of constraints and demoting constraints reminds one of phonological process suppression. However, the theoretical model and the information gained are far more detailed and give the clinician valuable information about what the child can do and not just what the child is incapable of doing. How Did Nonlinear Phonology Develop? John Firth, professor of general linguistics at the University of London, was a key figure in the development of modern linguistics in the United Kingdom. In a way, nonlinear phonology, too, can be traced back to Firth’s (1948) so-called prosodic analysis. For the first time, Firth challenged the one-sided linguistic importance of the phonemic units in their consecutive linearity. He advocated the necessity for additional nonsegmental analyses, “prosodies,” which represent larger linguistic entities, such as syllables, words, and phrases. He postulated that speech is a manifestation of consecutively ordered units as well as a manifestation of larger prosodic units that bind phonemes together into linguistically more comprehensive units. Different analytical systems may need to be set up in order to explain the range of contrasts involved. With this approach, known as polysystemicism, the concept of nonlinear phonology was born. Contemporary nonlinear/multilinear phonologies are an outgrowth of generative pho-

ch04.indd 94

nology. Chomsky and Halle’s (1968) major contribution, The Sound Pattern of English, was innovative in its description of two levels of representation, a surface phonetic representation and an underlying phonemic representation. Although the idea of distinctive features was taken from the Prague School of linguistics, Chomsky and Halle understood the distinctive feature concept in a different way and modified it accordingly. Nonlinear phonologies adopt the generative concepts of distinctive features and surface-level and underlying representation. However, these new phonologies understand the surface-level representation in a very different way. Chomsky and Halle’s generative phonology described speech components in a linear manner: It was segment based. The components of any utterance were arranged in a sequence, with one discrete segment following the next. A common set of distinctive features is attributed to all segments, and each feature is specified by the assignment of a binary value. This limited the generation of phonological rules in several respects. First, only whole segments could be deleted or added. The only other modifications that could occur in the segment were achieved by changing the + or – values of one or more distinctive features. (Thus, this system analyzes only additions, deletions, or substitutions; analysis of nonphonemic distortions is not possible.) Second, because all segments are equally complex and all distinctive features are equal within this system, there is no reason to expect that any one segment, or any one distinctive feature, might be affected by any given phonological rule. However, many observations and investigations have reported, for example, that certain sounds and sound classes appear to be especially vulnerable to assimilation, whereas others cause assimilation (Dinnsen, 1997). Third, early generative phonology adopted the division between the segmentals and the supra-

7/12/2007 8:05:48 AM

THEORETICAL CONSIDERATIONS AND PRACTICAL APPLICATIONS

segmentals that the structural linguists had used to describe and analyze speech events. However, such a division does not allow a vertical, hierarchical understanding of the interaction between segmental units and prosodic features. The nonlinear phonologies represent a challenge to the earlier segment-based approaches. “Nonlinear phonological theory is another step in the evolution of our understanding of phonological systems” (Bernhardt and Stoel-Gammon, 1994, p. 126). Nonlinear Phonologies: Clinical Implications Although contemporary nonlinear phonologies began with Goldsmith’s (1976) dissertation on autosegmental phonology, many different nonlinear phonological theories have since been proposed. This section has attempted to briefly introduce four nonlinear approaches: autosegmental phonology, metrical phonology, feature geometry, and optimality theory. However, it should be kept in mind that many other nonlinear phonologies exist. Several of these nonlinear approaches have been applied to the analysis of disordered phonological systems. In the 1980s, Leonard and Brown (1984) and Pollock and Schwartz (1988) demonstrated how disordered phonological systems could be analyzed utilizing autosegmental principles. Bernhardt and

95

Stoel-Gammon (1994) applied principles of grounded phonology (Archangeli and Pulleyblank, 1994) to the analysis of phonologically disordered speech. Stemberger and Bernhardt (1997) used optimality theory (McCarthy and Prince, 1993; Prince and Smolensky, 1993) to develop an intervention plan in a case study of a child. Dinnsen (1997) examined how underspecification theory, more specifically radical underspecification theory (Archangeli, 1988; Kiparsky, 1982; Pulleyblank, 1986), can explain both the normal acquisition of phonological systems and disordered phonology. More recently, many different articles have featured application of principles of optimality theory to the assessment process and to the selection of treatment targets. Based on a case study, Barlow (2001) and Dinnsen and O’Connor (2001) demonstrated how optimality theory could be utilized to analyze a child’s speech and to arrive at treatment targets. Bernhardt and Holdgrafer (2001a, 2000b) have outlined in detail how an in-depth phonological analysis can be generated that incorporates several nonlinear principles. Feature geometry has been widely applied to the analysis of phonological disorders in children (Bernhardt, 1990, 1992a, 1992b, 1994; Bernhardt and Gilbert, 1992; Bernhardt and Stemberger, 1998; Bernhardt and StoelGammon, 1994; Chin and Dinnsen, 1991; Von Bremen, 1990).

CLINICAL APPLICATION More Information—Feature Geometry versus Phonological Processes Let’s look at the difference between how feature geometry versus phonological processes would explain an example of a child who says [g ] for duck and [d ]

ch04.indd 95

for dumb. If phonological processes are assigned to these substitutions and deletions, the following results are noted:

“duck”

[d k]



[g ]

backing [d] → [g] final consonant deletion [k] → Ø

“dumb”

[d m]



[d ]

final consonant deletion [m] → Ø

7/12/2007 8:05:48 AM

96

CHAPTER 4

Although these phonological processes are easily identifiable, they give no information about the child’s underlying representation. Where to begin in therapy would be a relatively arbitrary choice that would be based on the number of times the processes were observed and the age at which they should be suppressed. Feature geometry demonstrates that the underlying representation for this child includes information about the dorsal place node, that is, about /k/ and /g/. This is evidenced by the dorsal production of [g] in [g ] for duck. Articulatory constraints, how-

ever, prevent realization of final consonants. If this was just a case of final consonant deletion, both duck and dumb should have been realized as [d ]. In the underlying representation, the child might be trying to differentiate between duck and dumb. This suggests that if the articulatory constraints could be eliminated, the child’s g/d substitution (backing) might also be eliminated. The concept of feature geometry and underlying representation provides us with more insight into reasons for the child’s output patterns.

Autosegmental and metrical phonologies have not received as much application as other nonlinear theories. Autosegmental phonology was conceived to account for cases in which a single segment is associated with two mutually exclusive features. Although it was originally applied to tone languages, it might also offer some applicability to disordered phonological systems. One important factor that is absent from many phonological theories is the child’s production of nonphonemic distortions, for example, a dentalized s-production ([s ] for /s/). According to generative phonology, this distortion must be classified either as /s/, therefore [+ strident], or as /θ/, [– strident]. It is possible, however, that this single segment [s ] could share these two seemingly mutually exclusive features of [+ strident] and [– strident]. If so, autosegmental phonology may offer a means of analyzing these types of distortions. Feature spreading within autosegmental phonology is structured to analyze assimilation processes but also offers a means of specifying coarticulatory conditions that might be useful in therapy. Although not used extensively in clinical application, metrical phonology, which extends hierarchical analysis procedures to stress and syllable boundaries, suggests several possi-

bilities in its application to disordered phonological systems. Kehoe (2001) has documented how a metrical approach can form the basis for an analysis of multisyllabic word productions, and Velleman and Shriberg (1999) have used metrical phonology to analyze the speech of children with suspected developmental apraxia of speech. Stress, both within a word and within a sentence, often plays a decisive role in the production of segments. Metrical phonology offers analysis procedures that emphasize the interrelationships that might exist between stress and segment production. The comparison of syllable and morphological boundaries within metrical phonology provides the potential to link phonological segmental production to other language areas. To summarize, many different nonlinear phonologies have been developed within the last decade or so. Some of them have been applied to case studies of children with disordered phonological systems. The results seem to indicate that these phonologies promise new insights into, and a deeper understanding of, the phonological system. Future research should document which of these theories can provide the clinician with new possibilities for the assessment and treatment of individuals with impaired phonological systems.

ch04.indd 96

7/12/2007 8:05:48 AM

THEORETICAL CONSIDERATIONS AND PRACTICAL APPLICATIONS

97

SUMMARY This chapter first introduced some of the basic terminology and principles underlying contemporary phonological theories. The relationship between the sound form and the sound function (as phoneme) was established as a basis for the understanding of phonological theories. The development of the phoneme concept was traced historically to provide a foundation for the understanding of how phonological theories could evolve from this “new” concept. Clinical application of these basic principles stressed the interrelationship between sound– form and sound–function. The remainder of this chapter was a summary of several different phonological theories that impact the assessment and treatment of phonological disorders. These theories were enumerated in a historical sequence. The lin-

ear phonologies were represented by distinctive feature theory, generative phonology, and natural phonology. The nonlinear phonologies included autosegmental, metrical, feature geometry, and optimality theory. Each phonological theory was discussed in respect to what the theoretical framework stands for, how it developed, how it functions, and its clinical implications. The field of phonology is constantly evolving. Current phonological theories are an attempt to describe the phonological system with all its complexity in a different manner. Although some of the newer models have yet to stand the test of time and research, all offer new insights into the intricate nature of normal and impaired phonological systems.

CASE STUDY The distinctive feature analysis procedure can be demonstrated using a slightly modified clinical example from Chapter 2, pages 33–34. The following sample is from Tina, age 3;8.

ch04.indd 97

The following errors are noted: [s] → [θ]

house, yes

[k] → [t]

duck, cat, cup, key

[θ] → [t]

bath, thumb

dig

[dε]

boat

[bot]

red, ring

house

[haυθ]

[r] → [w]

cup

[tυp]

[ ʃ ] → [s]

ship

knife

[nɑf]

lamp

[wmp]

[l] → [w]

lamp

duck

[dυt]

goat

[dot]

[] → [d]

goat

cat

[tt]

ring

[wŋ]

[ð] → [d]

that

bath

[bt]

thumb

[tm]

[z] → [ð]

zip

red

[wed]

that

[dt]

ship

[sp]

zip

[ðp]

fan

[fεn]

key

[ti]

yes

[jεθ]

win

[wn]

These target sounds and substitutions can be inventoried using the Distinctive Feature Worksheet. The following patterns emerge: The most frequent distinctive features include

7/12/2007 8:05:48 AM

98

CHAPTER 4

high (5 times), anterior (4 times), coronal (4 times), and back (4 times). A phonological process analysis procedure can be demonstrated using the same child, Tina, age 3;8. Target→Error [s]→[θ] house, yes [k]→[t] duck, cat, cup, key [θ]→[t] bath, thumb [r]→[w] red, ring [ ʃ ]→[s] ship [l]→[w] lamp []→[d] goat [ð]→[d] that [z]→[ð] zip

Phonological Process fronting velar fronting stopping + backing* gliding palatal fronting gliding velar fronting stopping + backing* fronting

Summarizing the phonological processes, we see that fronting (including both velar and palatal fronting) affected five sounds (s → θ, k → t, ʃ → s,  → d, and z → ð). Both stopping + backing (θ → t and ð → d) and gliding (l, r → w) were noted on two different sounds.

*Although backing was not covered, it is considered to be an idiosyncratic process that can be found in the speech of children with phonological disorders. Backing refers to a substitution in which the organ and/or place of articulation is more posteriorly located than the intended sound.

THINK CRITICALLY The following are the results of an articulation test from Ryan, age 6;6: horse wagon monkey comb fork knife cow cake baby bathtub nine train gun dog yellow

[hoυθ] [wən] [mŋki] [koυm] [fok] [nɑf] [kɑυ] [kek] [bebi] [bftəb] [nɑn] [twen] [n] [dɑ] [wεloυ]

pig cup swinging table cat ladder ball plane cold jumping TV stove ring tree green

[pk] [kp] [s wŋŋ] [tebəl] [kt] [lɾ] [bɑl] [pwen] [koυd] [dmpən] [tivi] [θtoυv] [wŋ] [twi] [win]

chair watch thumb mouth shoe fish zipper nose sun house steps nest books bird whistle

[ʃε] [wɑʃ] [fm] [mɑυf] [su] [fs] [ðp] [noυθ] [θn] [hɑυθ] [stεp] [nεt] [bυkθ] [bd] [wθəl]

doll

[dɑl]

this

[dθ]

carrots

[kεət]

1. Summarize the substitutions according to distinctive features. Which distinctive features occur most frequently?

ch04.indd 98

2. Summarize the errors according to phonological processes. Which phonological processes occur most frequently?

7/12/2007 8:05:49 AM

THEORETICAL CONSIDERATIONS AND PRACTICAL APPLICATIONS

99

TEST YOURSELF 1. Which one of the following does not belong to the phoneme/phonology concept? a. meaning-establishing and meaningdifferentiating function of sound units b. underlying form or representation c. production realities d. sound unit function within a particular language system 2. Which one of the following is a major class feature that distinguishes sounds produced with a high degree of oral obstruction? a. sonorant c. vocalic b. consonantal d. coronal 3. Which one of the following statements concerning phonological processes is not true? a. they are innate b. they are universal c. children with different language backgrounds begin with different sets of phonological processes d. they are used to simplify productions for the child in the developmental period 4. If a child says [wɑʃ ] for watch, this is an example of which phonological process? a. stopping c. deaffrication b. affrication d. labialization 5. Which one of the following is true about nonlinear/multilinear phonologies? a. segments are governed by more complex linguistic dimensions such as stress b. emphasis is on the sequential arrangement of sound segments c. all sound segments have equal value d. no one sound segment has control over the other units 6. Which one of the following terms is not representative of autosegmental phonology?

7.

8.

9.

10.

a. tiers are separable and independent levels b. certain segments are autonomous and do not have a one-for-one match on another level c. strong and weak stress are emphasized d. feature spreading is also a portion of this concept According to metrical phonology, the word potato has which one of the following stress patterns? a. weak branching to “po,” strong branching to “tato”; further divided into strong branching on “ta,” weak branching on “to” b. strong branching on “po,” weak branching to “tato”; further divided into strong branching on “ta,” weak branching on “to” c. weak branching to “po,” strong branching to “tato”; further divided into weak branching on “ta,” strong branching on “to” Which one of the following terms is not associated with feature geometry? a. spreading b. distinctive features c. faithfulness d. delinking In optimality theory, the constraint “markedness” requires outputs to be a. the same as the input b. simplified in structure c. marked d. demoted If a child produces [tɑ] for stop, then which one of the following constraints is violated? a. *COMPLEX c. *FRICATIVES b. *CODA d. MAX

WEBSITES www.phonologicaldisorders.com This website, created by the author of this textbook, contains basic definitions and examples of

ch04.indd 99

phonological processes. It also gives examples of articulation test results that are analyzed according to phonological processes. Links are given to other websites and resources.

7/12/2007 8:05:49 AM

100

CHAPTER 4

www.speech-language-therapy.com/Table2.htm This website from Caroline Bowen summarizes some more common phonological processes and gives examples of each process. For a beginning review, this website could be helpful. www.essex.ac.uk/speech/teaching-01/documents/ df-theory.html This website from Mark Tatham (1999 copyright) gives a brief discussion of distinctive feature theory. It provides references and a discussion on redundancy. It also has “clips” scattered throughout the website, which give you a more in-depth discussion of the various terms. www.chass.utoronto.ca/~contrast/#Summary This website reports on a research project on markedness that was conducted by Elan Dresher and Keren Rice (funded by the Social Science and Humanities Research Council of Canada). It out-

lines the goals of the project and provides an overview of markedness and how the topic could be extended to other areas such as second language acquisition. http://egg.auf.net/99/docs/abstracts/polgardi.html This relatively compact website, created by Krisztina Polgardi, discusses government phonology and optimality theory. There is a link to the Plovdiv website, which gives more broad-based information about phonology. http://camba.ucsd.edu/phonoloblog/index .php/2006/12/04/review-phonologicaldevelopment-and-disorders-in-children/ This website provides a book review of Phonological Development and Disorders in Children (Hua and Dodd, 2006, published by Multilingual Matters). It is a good read and gives a lot of interesting information on the multilingual perspective, which is the focus of the book.

FURTHER READINGS Archangeli, D., & Langendoen, T. (1998). Optimality theory: An overview. Malden, MA: Blackwell. Ball, M., & Kent, R. (1997). The new phonologies: Developments in clinical linguistics. San Diego: Singular. Bernhardt, B., & Stemberger, J. (1998). Handbook of phonological development: From the perspective of constrain-based nonlinear phonology. San Diego: Academic Press.

Bernhardt, B., & Stemberger, J. (2000). Workbook in nonlinear phonology for clinical application. Austin, TX: ProEd. Lombardi, L. (2001). Segmental phonology in optimality theory: Constraints and representations. New York: Cambridge University Press.

APPENDIX 4.1 1. Transcription of H. H. According to Pre-, Inter-, and Postvocalic Positions Child’s Production

Position

Description

1. house

[hɑυ]

postvocalic

[s] deletion

2. telephone

[tεfoυ]

[unstressed syllable deletion—noted but not counted on matrices] postvocalic [n] deletion

3. cup

[tp]

prevocalic

[k]



[t]

4. gun

[dn]

prevocalic

[]



[d]

Target Word

ch04.indd 100

7/12/2007 8:05:50 AM

THEORETICAL CONSIDERATIONS AND PRACTICAL APPLICATIONS

101

Child’s Production

Position

Description

5. knife

[nɑ]

postvocalic

[f ] deletion

6. window

[wnoυ]

intervocalic

[nd] →

7. wagon wheel

[wdən] [wi]

intervocalic postvocalic

[] → [d] [l] deletion

8. chicken

[ttə]

prevocalic intervocalic postvocalic

[tʃ ] → [t] [k] → [t] [n] deletion

9. zipper

[tpə]

prevocalic vowel nucleus

[z] []

→ →

[t] [ə]

10. scissors

[ttə]

prevocalic intervocalic nucleus + postvocalic

[s] → [z] → [z] →

[t] [t] [ə]

11. duck yellow

[dt] [jεwoυ]

postvocalic intervocalic

[k] [l]

→ →

[t] [w]

12. vacuum

[tu]

prevocalic intervocalic postvocalic

[v] deletion [kj] → [t] [m] deletion

13. matches

[mtət]

intervocalic postvocalic

[tʃ ] [z]

→ →

[t] [t]

14. lamp

[wmp]

prevocalic

[l]



[w]

15. shovel

[dvə]

prevocalic postvocalic

[ ʃ ] → [d] [l] deletion

16. car

[tɑə]

prevocalic vowel nucleus

[k] []

17. rabbit

[wb]

prevocalic postvocalic

[r] → [w] [t] deletion

18. fishing

[bdn]

prevocalic intervocalic postvocalic

[f ] → [b] [ ʃ ] → [d] [ŋ] → [n], this is considered a variation in regular pronunciation and not an error, not counted

Target Word

→ →

[n]

[t] [ə]*

* [ɑ] is considered to be a centering diphthong; therefore, it is the nucleus of the syllable.

ch04.indd 101

7/12/2007 8:05:50 AM

102

CHAPTER 4

Target Word

Child’s Production

19. church

Position

Description

[t]

prevocalic vowel nucleus postvocalic

[tʃ ] → [t] [] → [] [tʃ ] deletion

20. feather

[bεdə]

prevocalic intervocalic vowel nucleus

[f] [ð] []

21. pencils

[pεntə]

intervocalic

[ns] → [nt] [lz] deletion

child would not say

postvocalic

this 22. carrot

orange 23. bathtub bath

[tεwə]

[oυwn] [bt] [b]

→ → →

[b] [d] [ə]

prevocalic intervocalic postvocalic intervocalic postvocalic

[k] → [t] [r] → [w] [t] deletion [r] → [w] [nd] → [n]

intervocalic postvocalic postvocalic

[θt] → [t] [b] deletion [θ] deletion [θ] [f] [ŋ] [] [r]

→ → → → →

[b] [b] [n] [ə] [w]

[d] →

[d]

24. thumb finger

[bm] [bnə]

ring

[wŋ]

prevocalic prevocalic intervocalic vowel nucleus prevocalic

25. jump

[dmp]

prevocalic

26. pajamas

[dmi]

[unstressed syllable deletion—noted but not counted in matrices] prevocalic [d] → [d] [i] in end noted as a diminutive, final consonant deletion not counted in matrices

27. plane blue

[ben] [bu]

prevocalic prevocalic

[pl] [bl]

→ →

[b] [b]

28. brush

[bs]

prevocalic postvocalic

[br] → [ʃ] →

[b] [s]

29. drum

[dm]

prevocalic

[dr] →

[d]

30. flag

[b]

prevocalic postvocalic

[fl] → [b] [] deletion

ch04.indd 102

7/12/2007 8:05:51 AM

THEORETICAL CONSIDERATIONS AND PRACTICAL APPLICATIONS

Target Word

Child’s Production

31. Santa Claus

[tnə dɑ]

103

Position

Description

prevocalic intervocalic intervocalic postvocalic

[s] → [t] [nt] → [n], considered a normal assimilation, counted as correct [kl] → [d] [z] deletion

32. Christmas tree

[ttmə ti]

prevocalic intervocalic intervocalic

[kr] → [sm] → [str] →

33. squirrel

[twə]

prevocalic vowel nucleus postvocalic

[skw] → [tw] [] → [] [l] deletion

34. sleeping

[twipn]

prevocalic postvocalic

[sl] → [tw] [ŋ] → [n] this is considered a variation in regular pronunciation and not an error, not counted

prevocalic postvocalic

[st] → [d] [v] deletion

bed 35. stove

[t] [tm] [t]

[bεd] [doυ]

2. Phonological Processes for H. H. Child’s Production

Position

Description

1. house

[hɑυ]

postvocalic

final consonant deletion

2. telephone

[tεfoυ] postvocalic

weak syllable deletion final consonant deletion

Target Word

ch04.indd 103

3. cup

[tp]

prevocalic

velar fronting

4. gun

[dn]

prevocalic

velar fronting

5. knife

[nɑ]

postvocalic

final consonant deletion

6. window

[wnoυ]

intervocalic

cluster reduction

7. wagon wheel

[wdən] [wi]

intervocalic postvocalic

velar fronting final consonant deletion

8. chicken

[ttə]

prevocalic intervocalic postvocalic

stopping velar fronting final consonant deletion

7/12/2007 8:05:51 AM

104

CHAPTER 4

Target Word

Child’s Production

Position

Description

9. zipper

[tpə]

prevocalic vowel nucleus

stopping + devoicing derhotacization

10. scissors

[ttə]

prevocalic intervocalic nucleus postvocalic

stopping stopping + devoicing derhotacization final consonant deletion

11. duck yellow

[dt] [jεwoυ]

postvocalic intervocalic

velar fronting gliding

12. vacuum

[tu]

prevocalic intervocalic postvocalic

initial consonant deletion cluster reduction + cluster substitution (velar fronting) final consonant deletion

13. matches

[mtət]

intervocalic postvocalic

stopping stopping + devoicing

14. lamp

[wmp]

prevocalic

gliding

15. shovel

[dvə]

prevocalic postvocalic

stopping + fronting + voicing final consonant deletion

16. car

[tɑə]

prevocalic nucleus

velar fronting derhotacization

17. rabbit

[wb]

prevocalic postvocalic

gliding final consonant deletion

18. fishing

[bdn]

prevocalic intervocalic postvocalic

stopping + labialization + voicing stopping + fronting + voicing not counted, normal variation

19. church

[t]

prevocalic nucleus postvocalic

stopping derhotacization final consonant deletion

20. feather

[bεdə]

prevocalic intervocalic nucleus

stopping + labialization + voicing alveolarization + stopping derhotacization

21. pencils

[pεntə]

intervocalic postvocalic

cluster substitution (stopping) cluster deletion

22. carrot

[tεwə]

prevocalic intervocalic postvocalic intervocalic postvocalic

velar fronting gliding final consonant deletion gliding cluster reduction

orange

ch04.indd 104

[oυwn]

7/12/2007 8:05:51 AM

105

THEORETICAL CONSIDERATIONS AND PRACTICAL APPLICATIONS

Target Word

Child’s Production

23. bathtub

[bt]

bath

[b]

Position

Description

intervocalic postvocalic postvocalic

cluster reduction final consonant deletion final consonant deletion stopping + labialization + voicing stopping + labialization + voicing cluster reduction + velar fronting derhotacization gliding

24. thumb finger

[bm] [bnə]

ring

[wŋ]

prevocalic prevocalic intervocalic nucleus prevocalic

25. jump

[dmp]

prevocalic

stopping

26. pajamas

[dmi] prevocalic postvocalic

weak syllable deletion stopping diminutive—use of [i]

27. plane blue

[ben] [bu]

prevocalic prevocalic

cluster reduction + voicing cluster reduction

28. brush

[bs]

prevocalic postvocalic

cluster reduction palatal fronting

29. drum

[dm]

prevocalic

cluster reduction

30. flag

[b]

prevocalic

cluster reduction, cluster substitution (stopping + labialization + voicing) final consonant deletion

postvocalic 31. Santa Claus

[tnə dɑ]

prevocalic intervocalic intervocalic

postvocalic 32. Christmas tree

[ttmə ti]

prevocalic intervocalic intervocalic

33. squirrel

[twə]

prevocalic nucleus postvocalic

ch04.indd 105

stopping [nt] → [n] considered normal assimilation, not counted cluster reduction, cluster substitution (velar fronting + voicing) final consonant deletion cluster reduction, cluster substitution (velar fronting) [sm] → [t] cluster reduction, cluster substitution (stopping) [str] → [t] cluster reduction cluster reduction, cluster substitution (velar fronting) derhotacization final consonant deletion

7/12/2007 8:05:51 AM

106

CHAPTER 4

Target Word

Child’s Production

Position

Description

34. sleeping

[twipn]

prevocalic

cluster substitution (stopping, gliding) not counted, normal variation

postvocalic bed 35. stove

[bεd] [doυ]

prevocalic postvocalic

cluster reduction, cluster substitution (voicing) final consonant deletion

3. Spontaneous Speech Sample for H. H. Looking at pictures: [d ə ptə əv ə tɑ] That a picture of a dog.

[oυ d t ə tti] Oh, that is a kitty.

[hi ə b dɑ] He a big dog.

[wi hf ə tti] We have a kitty.

[hi baυ ən h ə tɑwə] He brown and has a collar.

[wi dɑt aυ tti ə wɑ:ŋ tam] We got our kitty a long time.

Conversation with Mom: [tn wi do tu mədɑnoυ] Can we go to McDonald?

[hi tm tu mədɑnə wt t] He come to McDonald with us?

[a w ə tibdə] I want a cheeseburger.

[xxxxx mɑ haυ] xxxx My house.

[a w fεnfɑθ] I want french fries.

[mɑmi lε do] Mommy let go.

[wε t bwi] Where is Billy?

[lε do naυ] Let go now.

Talking about summer vacation: [wi doυf tu dmɑt] We drove to Grandma.

[si ht wɑtə taυt] She has lot’a cows.

[si wf n oυ +hao] She live in Ohio.

[taυt ju noυ mu taυ] Cows, you know, moo cow.

[si ht ə fɑm] She has a farm.

[de it ə ho wɑt] They eat a whole lot.

ch04.indd 106

7/12/2007 8:05:51 AM

Loading...

Theoretical Considerations and Practical Applications

4 Theoretical Considerations and Practical Applications LEARNING OBJECTIVES When you have finished this chapter, you should be able to: • Trace how t...

676KB Sizes 0 Downloads 0 Views

Recommend Documents

theoretical and practical considerations regarding the cost calculation
system, process cost system), and the partial cost accounting (e.g. direct costing, variable costing). This article has

Some Theoretical Considerations on Caste
1960; Sinha 1967; Berreman 1968; Appadurai 1986; Gould 1990). Whether caste is best considered as a cultural or a struct

Theoretical and Experimental Considerations - UCF Stormwater
The kinetics of reductive dissolution of hematite ( 01-Fe203 ) by the dissimilatory iron- reducing bacterium Shewanella

Numerical optimization theoretical and practical aspects pdf
... textbook for the health sciences pdf creeps. Clancy damped reversal assignments vary engrandecer professional tax fo

Practical Handbook of Photovoltaics: Fundamentals and Applications
The Handbook covers the principles of solar cell function, the raw materials, photovoltaic systems, standards, calibrati

Practical Handbook of Photovoltaics: Fundamentals and Applications
Practical Handbook of Photovoltaics: Fundamentals and Applications eBook: Augustin McEvoy, Tom Markvart, Luis Castaner,

Practical Applications for Effective Teaching and Learning
The key principles of effective teaching: 1 Using ... your students and their learning and, most important of all, provi

Practical Handbook of Photovoltaics: Fundamentals and Applications
Low cost industrial manufacture of crystalline silicon solar cells. J. Szlufcik, S. Sivoththaman, J. Nijs, R.P. Mertens

Practical Handbook of Photovoltaics: Fundamentals and Applications
Fundamentals and Applications Augustin Joseph McEvoy, T. Markvart, Luis Castañer ... [1] [2] [3] [4] [5] [6] [7] [8] [9

Lysogeny: Practical applications and new - Scholar Commons
Mar 29, 2005 - Scholar Commons Citation. McDaniel, Lauren, "Lysogeny: Practical applications and new discoveries" (2005)