sign in sign up
<<
Home , Continuant, Minimal pair, Phonemic contrast

Minimal Pair Approaches to Phonological Remediation

Jessica A. Barlow, Ph.D.,1 and Judith A. Gierut, Ph.D.2

ABSTRACT

This article considers linguistic approaches to phonological reme- diation that emphasize the role of the in . We discuss the structure and function of the phoneme by outlining procedures for de- termining contrastive properties of sound systems through evaluation of minimal pairs. We then illustrate how these may be applied to a case study of a child with phonological delay. The relative effectiveness of treat- ment approaches that facilitate phonemic acquisition by contrasting pairs of sounds in minimal pairs is described. A hierarchy of minimal pair treat- ment efficacy emerges, as based on the number of new sounds, the number of featural differences, and the type of featural differences being intro- duced. These variables are further applied to the case study, yielding a range of possible treatment recommendations that are predicted to vary in their effectiveness.

KEYWORDS: Phoneme, minimal pair, phonological remediation

Learning Outcomes: As a result of this activity, the reader will be able to (1) analyze and recognize the con- trastive function of in a phonological system, (2) develop minimal pair treatment programs that aim to introduce phonemic contrasts in a child’s phonological system, and (3) discriminate between different types of minimal pair treatment programs and their relative effectiveness.

Models of clinical treatment for children cognition given our need to understand how with functional phonological delays have been learning takes place in the course of interven- based on three general theoretical frameworks. tion. Still other approaches are grounded in Some models are founded on development linguistics because the problem at hand in- given that the population of concern involves volves the phonological system. In this article, children. Other models have their basis in we explore the linguistic bases of phonological

Updates in Phonological Intervention; Editors in Chief, Nancy Helm-Estabrooks, Sc.D., and Nan Bernstein Ratner, Ed.D.; Guest Editor, Shelley Velleman, Ph.D. Seminars in Speech and Language, volume 23, number 1, 2002. Address for correspondence and reprint requests: Jessica A. Barlow, Ph.D., Department of Communicative Disorders, San Diego State University, San Diego, CA 92182. E-mail: [email protected]. 1Department of Communicative Disorders, San Diego State University, San Diego, California and 2Department of Speech and Hearing Sciences, Indiana University, Bloomington, Indiana. Copyright © 2002 by Thieme Medical Publishers, Inc., 333 Seventh Avenue, New York, NY 10001, USA. Tel: +1(212) 584-4662. 0734–0478,p;2002,23;01,057,068,ftx,en;ssl00104x. 57 58 SEMINARS IN SPEECH AND LANGUAGE/VOLUME 23, NUMBER 1 2002

intervention with a specific focus on the phonetic variants then are of the phoneme. We ask three questions: What are phoneme /t/. the structure and function of phonemes in lan- By definition, phonemes are abstract, men- guage generally? How are phonemes assessed tal concepts of sounds that reflect a speaker’s in children’s sound systems? And, how might internal or mental knowledge about the lan- treatment be designed to best facilitate - guage he or she speaks, and they function to mic acquisition? mark distinctions in meaning. Phonemic dis- tinctions are made most apparent in minimal pairs. A minimal pair is a set of that dif- fer by a single phoneme, whereby that differ- WHAT IS A PHONEME? ence is enough to signal a change in meaning. For instance, the words “map” [mp] and “mat” For the linguist interested in the study of [mt] form a minimal pair in English. These , a primary aim is to establish the two words are identical in terms of the first inventory of sounds and how these function in and the . They differ only by a given language. For the speech-language the last consonant—[p] versus [t]—and this pathologist, a key to establishing treatment difference signals a change in meaning (“map” goals is to assess the sounds of a child’s phono- versus “mat”). English speakers would agree logical system. As with the linguist studying a that “map” and “mat” are two very different newly discovered language or reanalyzing a words with two very different meanings. The well-known one, our clinical aim in analyzing “map”–“mat” example illustrates how minimal a developing phonological system is to deter- pairs reveal phonemic contrasts in final posi- mine which sounds are used as phonemes by a tion, but contrasts also occur in other contexts, child and which are used as allophones. To including word-initial position, such as “map” evaluate this requires a firm understanding of [mp]-“cap” [kp], as well as word-medial just what a phoneme is. A phoneme reflects a positions, such as “fashion” [fʃən]-“fasten” certain “ideal” concept of a sound, according to [fsən]. Cluster contexts, both initial and the perceptions of the speaker, even though final, can also be revealing of contrasts, as with that phoneme might actually be produced “spy” [spɑ]-“sky” [skɑ] or “cats” [kts]– phonetically in a variety of ways. Take the “caps” [kps]. too contrast within min- phoneme /t/, for example. As speakers of the imal pairs, as with “map” [mp]–“mop” [mɑp] , we all share a similar con- and “cap” [kp]—“keep” [kip]. Finally, near cept of /t/. This is reflected in our spelling as minimal pairs are found via cluster-singleton well as our perceptions about the sounds in our comparisons, such as “play” [ple]–“pay” [pe], language and in other . When we or vowel- versus consonant-final comparisons, hear words such as “tie,”“star,”“butter,” and such as “boat” [boυt]–“bow” [boυ]. “button,” we, as English speakers, recognize In comparison with these examples of that all of these words have the sound /t/ in phonemes revealed by minimal pairs, we con- common. This is the case despite the fact that sider another possible pair in English, [mæp] the relevant sounds in these words are actually and [mp]. These two forms are identical with quite different from one another in produc- respect to the first consonant and the vowel as tion. Whereas “tie” is pronounced with an as- with the “map”–“mat” example. Similarly, they pirated [th] as [thɑ], “star” is pronounced with also differ only by the final sound: released [p] an unaspirated [t] as [stɑr]. “Butter” is pro- versus unreleased [p ]. However, most impor- nounced with a flap in American English as tant, this phonetic difference does not signal a [bɾ], and “button” is produced with a glottal change in meaning because the two words mean stop as [bʔn ]. Nevertheless, speakers of the exactly the same thing, “map.” With no differ- English language assume that all four of ence in meaning, [mp] and [mp] do not these phones [t t ɾ ʔ] belong to one and the form a minimal pair. Consequently, [p] and [p ] same category—the phoneme /t/. These four are not contrastive, nor do they function as MINIMAL PAIR APPROACHES/BARLOW, GIERUT 59 phonemes in English; instead, they are allo- /f v/, /s z/, and /tʃ d/. Together, the features phones (phonetic variants) of the phoneme /p/. associated with place, manner, and are Phonemes are typically viewed as phonetic called nonmajor class distinctions. These are reflexes of a complex of smaller sized units differentiated from other major class properties. known as distinctive features. Consequently, it is The major class features serve to differ- not the phonemes per se that are the contrastive entiate among the main groupings of sounds in elements of a language; rather, it is their featural language, namely versus vowels, makeup. The features that contrast serve to cre- glides versus consonants, and (stops, ate an “opposition” between phonemes of a par- , and ) versus ticular language.1,2 Phonemes may contrast with (nasals, liquids, glides, vowels). The featural one another along one or more of a number of distinctions that mark these respective con- specific feature dimensions. In formalized lin- trasts are [syllabic], [consonantal], and [sono- guistic frameworks, these featural contrasts are rant]. For example, the contrast between /b/ described in terms of distinctive features as de- and /m/ in the pair “by”–“my” illustrates that a scribed in, for example, The Sound Pattern of major class distinction between obstruents and English.3 These can be interpreted in perhaps sonorants ([]) occurs in English. Simi- more familiar terminology associated with larly, the contrast between /m/ and /w/ in place, manner, and voice for the present pur- “my”–“why” shows that a major class distinc- pose, as illustrated for English phonemes in tion between consonants and glides ([conso- Table 1. nantal]) also occurs in English. features differentiate Based on their featural characteristics, then, labial, coronal, and dorsal contrasts, which are phonemes may contrast with one another either relevant to the phonemic distinctions among minimally or maximally. A minimal contrast is /p t k/ and /b d / in, for example, “pea,”“tea,” defined by a single or just a few feature differ- and “key” or “by,”“dye,” and “guy,” respectively. ences among phonemes. The difference between features differentiate “pie”–“by” involves a minimal contrast in voice, [continuant] contrasts, which are relevant to whereas the differences between “tea”–“key” or distinctions between stops, fricatives, and af- “toe”–“so” involve minimal contrasts in place and fricates. This feature would distinguish, for ex- manner, respectively. By comparison, a maximal ample, /p/ from /f/ in “pea”–“fee,” or /ʃ/ from contrast means that a phonemic difference cuts /tʃ/ in “shoe”–“chew.” Manner features also across many featural dimensions of place, man- differentiate [nasal] and [lateral] contrasts, ner, and voice. The phonemes /b/ and /s/ in the which relate to nasals, liquids, and glides. pair “be”–“see” differ along all three dimensions These would distinguish /n/ from /l/ in with a place contrast differentiating labial from “no”–“low,” or /l/ from // in “lake”–“rake,” re- coronal, a manner contrast differentiating stop spectively. Voice features differentiate cognate from , and a voice contrast differentiat- pairs, such as /p b/ in “pie”–“by” as well as /t d/, ing voiced from voiceless.

Table 1 English Phonemes Labial Coronal Dorsal Labio Inter Palato Bilabial dental dental Alveolar alveolar Palatal Velar Glotta

Obstruent Stop p b t d k g Fricative f v θð s z ʃ tʃ d Sonorant Nasal m n ŋ Liquid l r Glide w j h 60 SEMINARS IN SPEECH AND LANGUAGE/VOLUME 23, NUMBER 1 2002

Thus, we have seen that the features that longitudinal study of children with phono- make up phonemes may vary in terms of the logical delay of unknown origin.9–12 Joseph type and the number of contrastive differences. presented with normal hearing, intelligence, Significantly, phonemic distinctions that in- oral-motor functioning, and receptive and ex- volve a maximal number of feature differences pressive language skills as determined by for- and that involve major class features are gener- mal testing procedures. His phonetic inventory ally the most salient contrasts of languages of is shown in Table 2, as derived from a two-time the world.4,5 These also tend to emerge first in occurrence of sounds, independent of context , and the more specific and accuracy, from a 306-item single-word distinctions are elaborated later.1 Therefore, speech sample.13 Only the phones that Joseph more fine-grained phonemic contrasts are typ- actually produced are shown (whether or not ically indicative of increased featural complex- target appropriate); all other (target) sounds ity in the phonological system. were not produced or used by the child. Hence, these sounds were excluded from the phonetic repertoire. As can be seen, Joseph produced stops, nasals, and glides of target English but ASSESSING PHONEMES IN A lacked most fricatives (excepting [f v]) and all CHILD’S SYSTEM affricates and liquids. The data in Table 3 provide examples of In assessing the phonemes of a child’s phono- Joseph’s productions that can be used to iden- logical system, the speech-language patholo- tify which of these phones actually functioned gist, like the linguist, must evaluate both their as phonemes in his system. (These data are structure (what sounds occur?) and their func- provided in alphabetic order by column ac- tion (is their role contrastive?). To do this cording to English spelling of the target word.) requires examining the child’s phonemic sys- Recall that phonemes are revealed by minimal tem independently of the target phonology pairs; therefore, it is necessary to identify mini- and then comparing how the child’s use of mal pairs in Joseph’s sample. One example phonemes may diverge from what is expected. from Table 3 is the child’s production of the In this way, independent and relational analy- words “drive” [ɑ] and “bite” [bɑ]. The pro- ses are employed.6–8 ductions were identical with the exception of To illustrate the assessment process, we the sounds [] and [b], and they differed se- consider the clinical case of a child we will call mantically. These two words then formed a “Joseph.” This boy (aged 4;1) was diagnosed minimal pair, which means that // and /b/ with a functional phonological delay. His functioned as phonemes in Joseph’s sound sys- speech samples were drawn from a large-scale tem. Moreover, the featural contrast involved a

Table 2 Phonetic/Phonemic Inventory of Joseph (Male, Aged 4;1)* Labial Coronal Dorsal Labio Inter Palato Bilabial dental dental Alveolar alveolar Palatal Velar Glottal

Obstruent Stop /p/ /b/ /t/ /d/ /k/ /g/ /ʔ/ Fricative f v Affricate Sonorant Nasal /m/ /n/ /ŋ/ Liquid Glide /w/ /j/ /h/ *All sounds are included in phonetic inventory, while sounds in phonemic inventory are presented in slashes. MINIMAL PAIR APPROACHES/BARLOW, GIERUT 61

Table 3 Data from Joseph noted in his phonetic inventory (Table 2). This Target Production Target Production collapse of the contrast resulted in homonymy, whereby words with different meanings were ɑ [b ] “bite” [pi] “pig” pronounced in the same way.  [b ] “bus” [kin] “queen” Continuing in the fashion of identifying [ti] “cheese” [ho] “robe” minimal pairs and homonymy in the speech  υ [k p] “cup” [ta p] “sharp” sample, Joseph’s phonemic inventory was ʔ [k ] “cut” [to] “soap” thereby constructed. The sounds that func- ʔ] [do] “dress” [tu “soup” tioned as phonemes for the child are presented ɑ  [g ] “drive” [t ni] “sunny” between slashes in Table 2. From this table, it is ɑ  [p ] “five” [t mi] “thumby” apparent that there are a number of gaps in the  [g p] “gift” [to] “toes” child’s phonemic repertoire relative to target ʔ [du] “juice” [tu ] “tooth” English; in particular, fricatives, affricates, and  [k p] “kids” [bæn] “van” liquids did not function to signal meaning dif- ŋ υ [ki ] “king” [wa p] “yard” ferences in this child’s use of language. These  [m ] “mud” [ju] “you” sound classes then would be potential targets for clinical treatment, with the goal being to introduce the phonemic contrasts of English minimal and nonmajor class difference be- that are absent from the child’s system. tween dorsal and labial places of articulation. Notice, however, that this minimal pair in Joseph’s system was not target appropriate. STRUCTURING PHONEMIC This notwithstanding, the phonemes // and INTERVENTION: CONTRAST /b/ that Joseph used to signal meaning differ- TREATMENT ences are also used as phonemes in the English language, as with “guy”–“buy”. Thus, although Minimal pair treatment is one model of inter- the minimal pair evidence was incorrect rela- vention aimed at introducing new phonemic dis- tive to English, the phonemic status of // and tinctions in language.14–24 This model is an out- /b/ was the same in Joseph’s system and Eng- growth of its original applications to second lish. Thus, there was a match between the language learning,25 and its underlying assump- phonemic repertoires of the target language tions follow directly from linguistic constructs as and Joseph’s phonology. outlined earlier. In essence, minimal pair treat- The same was not true of some of Joseph’s ment teaches featural contrasts through the use other productions. The words “toes” and of pairs of words that differ by a single phoneme. “soap,” for example, were both produced as The featural differences between the phonemes [to]. Similarly, “soup” and “tooth” were both are the focus of treatment. The premise is that, produced as [tuʔ]. These pairs of words clearly once a featural difference is introduced, that have different meanings in English, but they same difference will be applied by a child to did not form a minimal pair in Joseph’s system other relevant phonemic pairs. For example, the because they were produced in exactly the phonemes /k/–/t/ differ in terms of place, dorsal same way. This means that the target English versus coronal. If the place distinction is learned phonemes /s/ and /t/ did not function the in treatment of /k/–/t/ pairs, this same dorsal- same way in Joseph’s system. There was a mis- coronal contrast should be carried over to other match between the phonemes of English and dorsal-coronal pairs such as //–/d/ and /ŋ/– the phonemes of Joseph’s phonology. Joseph /n/. The clinical advantage then is that general- collapsed the intended phonemic contrast, ization to untreated phonemes is expected to with all target /s/ and target /t/ words being occur. Minimal pair treatment can be set up in a produced identically as [t]. In fact, [s] never variety of ways that affect the extent of general- occurred in any of Joseph’s productions as was ization for children with phonological delays. 62 SEMINARS IN SPEECH AND LANGUAGE/VOLUME 23, NUMBER 1 2002

Conventional Minimal Pair Treatment tended “sail.” In this way, outputs that elimi- nate ambiguity associated with homonymy The conventional minimal pair model of treat- would be deemed “correct.” Homonymy pre- ment targets a phonemic contrast by pairing a sumably motivates phonemic change. Thus, sound that is produced in error by a child with the foundation of the conventional minimal its corresponding substitute. Presumably, the pair approach is to (1) modify groups of sounds substitute will be a sound that differs mini- that are produced in error in a patterned way, mally from the target phoneme.22–24 In Joseph’s (2) highlight featural contrasts rather than ac- case, a clinician might select /s/ and contrast it curate sound production, and (3) emphasize with /t/. Recall that the phoneme /s/ is absent sound use for communicative purposes.26 from Joseph’s phonemic repertoire and [t] is Two spin-offs of the conventional mini- the corresponding substitute. As noted, these mal pair model have been proposed, both of two sounds differ minimally in manner of ar- which focus on relationships between target ticulation (Table 1). With acquisition of the phonemes and substitutes. Under a multiple op- fricative-stop contrast between /s/ and /t/, a position approach,27,28 sound pairs are selected clinician might expect that other related frica- based on every one of a child’s substitutes for a tive-stop distinctions would also emerge in target sound.29 For a child who exhibits a great Joseph’s sound system. deal of variability, for example, producing tar- The conventional minimal pair model get /s/ as [t d θ l], four sets of minimal pairs makes two further assumptions that are best would be introduced in treatment: /s/–/t/, exemplified by their application. Continuing /s/–/d/, /s/–/θ/, and /s/–/l/. the example, the two phonemes /s/–/t/ would Under an alternate metaphon approach, be presented in a treatment session as pairs of one target-substitute sound pair is selected words, perhaps represented pictorially and dis- for treatment, but the way it is taught includes played in an array. In a game format, Joseph a metalinguistic component.30–32 Metaphon would be instructed to name one of the pic- emphasizes real-world opposites (e.g., long- tures, with the clinician guessing which picture short, back-front) and relates them to opposi- was named. For example, using the minimal tions of the sound system in language. The pair “sail”–“tail,” it is likely that Joseph would concept of long-short might be associated with produce both words as [tel] given the collapse a fricative-stop error or back-front with a dor- of contrast between /s/ and /t/. Consequently, sal-coronal substitution pattern. the clinician would be likely to point to the The conventional minimal pair treatment picture of “tail” even when Joseph had in- model has been reported as clinically effective, tended the alternate “sail.” Typically, under but a number of questions have been raised the conventional format, explicit instructions that have prompted further research in this about the place, manner, or voice of /s/ produc- area. The conventional approach has been tion would not be provided. The clinician viewed as a conceptual form of intervention, would only provide feedback about the mis- but some have speculated about whether in- communication, showing confusions about structions about place, manner, or voice of pro- which one of the pair to choose. To correct the duction might also be necessary.33 This may be ambiguity, Joseph must learn to modify his ho- of relevance when children use nonambient mophonous productions, and this may take sounds as substitutes, such as the velar fricative different forms. For example, Joseph might [x] for /ʃ/, or when distortions are produced, produce another fricative in place of /s/, per- such as [s] or [$] for /s/.34 In cases such as haps [f ] or [ʃ]. Because Joseph’s productions these, it would not be appropriate to pair the would no longer result in homonymy, and target with its nonambient or distorted substi- the target fricative-stop contrast was being tute because these are neither phonetic nor preserved, the clinician would reward a suc- phonemic in English. Moreover, a child’s out- cessful communication by selecting the in- puts would not result in the homonymy that is MINIMAL PAIR APPROACHES/BARLOW, GIERUT 63 apparently needed to promote change. In addi- manipulate the number of new phonemes intro- tion to these concerns, some have considered duced in treatment, the nature of the compari- whether homonymy is indeed relevant to son sound, and the number and type of featural change in development.20,33,35–39 If homonymy differences among phonemes being treated. does trigger change, then one might ask why Consider that the conventional model intro- fully developed languages allow homophonous duces only one target phoneme, and this is forms without breakdowns in communication paired with the substitute. It is also possible to (e.g., English “to”–“two”–“too,”“buy”–“bye”– teach one target phoneme but to pair it with an- “by”).40 With these questions in mind, we turn other known (and correct) phoneme unrelated to two studies that have examined some of the to the substitute. Further, two target phonemes, assumptions of the conventional model. both absent from a child’s inventory, may be Saben and Ingham33 evaluated the effects paired with each other. The effectiveness of of minimal pair treatment with and without these alternate pairings was evaluated relative imitation and articulatory placement cues. Re- to the conventional format and to each other. sults indicated that some children may require Results indicated that contrasting one new these steps before improvements in target phoneme with an unrelated known phoneme sound production are observed. This suggests induced greater generalization than the conven- that addressing the function of the sounds in tional target-substitute format.38,39 In turn, tar- communication, as recommended by conven- geting two new phonemes led to greater gener- tional minimal pair approaches, may not be the alization than treatment of a new phoneme single most important variable that is crucial to contrasted with another known sound. treatment effectiveness. Gierut20,37–39 further The number of feature differences in a examined the role of homonymy in promoting contrast has also been examined. Recall that sound change in a series of studies that evalu- the conventional model recommends treating ated the relative effects of two different treat- phonemes that differ minimally, thus allowing ment conditions on children’s sound systems. the emergence and generalization of a specific In one treatment condition, a target sound was featural contrast. But it is also possible to paired with its substitute for expected ho- manipulate maximal feature differences in mini- mophony in production. In another treatment mal pairs by differentiating along the dimen- condition, two target sounds, both of which sions of place, manner, and voice as well as were excluded from the child’s inventory, were major class properties.36 Targeting maximal dif- paired with each other. Here, the children’s ferences does not necessarily constrain which outputs were expected to be nonhomonymous. features will emerge or generalize but leaves this Results indicated that the nonhomonymous to a child’s own phonological problem solving.41 condition resulted in greater improvements in Research that has tested minimal versus maxi- the children’s sound systems. Although the mal feature differences in treatment has found homonymous condition did lead to generali- in fact that targeting a maximally opposed zation, this occurred to a lesser extent. These sound pair resulted in greater generalization findings lend further support to the hypothesis than did a minimally opposed pair.20,37 that homonymy may not be a necessary com- Finally, the nature of the feature differ- ponent in . ences presented in minimal pairs has been ex- perimentally evaluated. The distinction is be- tween treatment of salient featural contrasts Minimal Pair Variants associated with major class versus more fine- grained distinctions associated with nonmajor In a related vein, researchers have also consid- class properties of place, manner, or voice.20,36 ered the substance of the contrast that is pre- These studies determined that sound pairs dif- sented in minimal pair treatment. This has re- fering by major class features result in greater sulted in several new minimal pair models that generalization than sound pairs that differ only 64 SEMINARS IN SPEECH AND LANGUAGE/VOLUME 23, NUMBER 1 2002

Figure 1 Hierarchy of maximally opposed minimal pair treatment alternatives and their relative effectiveness in inducing phonological change in delayed sound systems. Adapted from Gierut.20

by nonmajor class. This includes comparisons research as most effective, there may be certain of obstruents versus sonorants and consonants instances when a clinician selects a treatment versus glides. pair that is expected to be less than optimal. Taken together, research on minimal pair These may be cases in which the goal is to variants20,37,38 suggests that the most effective elaborate upon properties of a child’s sound conditions of contrast include pairing two new system that are already in place, as in building phonemes that differ maximally and by major the fricative repertoire if at least one fricative is class features. Although this may be the opti- already being used phonemically. Here, it may mal structure of minimal pairs, there is a hier- be appropriate to select one new fricative and archy of effectiveness as depicted in Figure 1. to pair it with another phoneme to create non- Notice in particular that there is a trade-off be- major and maximal distinctions. tween number of new sounds (two versus one) To illustrate the application of these poten- and type of distinctions (major versus nonma- tial minimal pair variants, we return again to the jor). That is, two new sounds contrasting a case of Joseph. For continuity with the prior dis- nonmajor place, manner, or voice distinction cussion, we selected /s/ as one target phoneme are likely to be just as effective as teaching a to be treated. Following Figure 1 and with refer- major class distinction by contrasting one new ence to Joseph’s inventory in Table 2, it is possi- phoneme paired with a known sound. Also, al- ble to make predictions about which phonemes though this hierarchy describes minimal pairs might induce the greatest generalization using only involving maximal feature differences, the minimal pairs in treatment. Table 4 shows sev- same relative relationships hold true for mini- eral possible contrasts that may be targeted in mal feature differences.37,38 Finally, it is also treatment for Joseph, assuming that the target important to keep in mind that although one sound was /s/. These include homonymous minimal pair structure has emerged from this versus nonhomonymous, one versus two new MINIMAL PAIR APPROACHES/BARLOW, GIERUT 65

Table 4 Possible Contrastive Sound Pairs for Joseph:Treatment Target /s/ Place, Manner, Target Sound No. of New Homonymous Voice Major Class Pair Sounds? Pair? Differences? Distinction?

/s/–/t/ 1 Yes M No /s/–/f/ 1 No P No /s/–/b/ 1 No PMV No /s/–/w/ 1 No PMV Yes /s/–/θ/2 NoP No /s/–/d/2 NoPMVNo /s/–/r/ 2 No PMV Yes

sounds, maximal versus minimal opposition, same considerations, including the pairs that and major versus nonmajor class distinction. target a sound other than /s/. In all cases, the Predictably, the greatest generalization potential contrasts for treatment are identified will occur if /s/ is paired with another new inreferencetoJoseph’s phonemic inventory in phoneme involving maximal and major class Table 2. feature differences. Given the phonemic in- Throughout this article we have discussed ventory of this child, we see that phonemes /f minimal pair treatment programs that focus on v θðszʃtʃdlr/ are all excluded. Thus, production. However, some minimal pair ap- for the most effective treatment involving the proaches have implemented a perceptual com- structure of two new sounds with maximal and ponent prior to or perhaps even in lieu of pro- major class feature differences, /s/–/r/ would duction treatment.15,42,43 Such research has be a likely target sound pair. These two sounds established that perceptual training involving differ from one another in terms of place, minimal pairs can facilitate production accu- manner, and voice features and by the major racy. Furthermore, the effectiveness of percep- class feature [sonorant] (Table 1). For some- tually based minimal pair treatment is based on what less effective treatment for Joseph, we the pairing of sounds that differ by a maximal might consider pairing two new sounds that number of distinctive features, including the are maximally opposed with nonmajor class major class features.42,43 Thus, productively distinctions. In this case, /s/–/d/ would be based and perceptually based minimal pair ap- one appropriate target pair. Although these proaches converge in their format. sounds contrast along only nonmajor class fea- In conclusion, minimal pair treatment tures, they differ maximally across the three models, no matter their form, maintain a com- domains of place, manner, and voice, as can be mon goal that derives from constructs of lin- observed from Table 1. An equally effective guistic theory; that is, to teach the phonemic contrast would pair one new sound with one distinctions of language. Phonemic contrasts, known sound that is maximally opposed by a however, constitute just one of the formal major class feature, such as /s/–/w/. These two properties of a phonological system. Hierar- sounds differ by two major class features chical elements of phonology include such ([sonorant, consonantal]) and furthermore dif- structures as the word itself, , and sylla- fer along all three dimensions of place, man- bles, in addition to phonemes and features. ner, and voice. The relatively least effective Models of intervention that target these com- treatment would involve pairing one new plementary structures have likewise benefited sound and one known sound that differ maxi- from linguistic theory.44–52 This underscores mally and by nonmajor class features, such as the integral influence of linguistics on inter- /s/-/b/. Contrasts other than those listed in vention for children with functional phonolog- Table 4 may be selected on the basis of the ical delays. 66 SEMINARS IN SPEECH AND LANGUAGE/VOLUME 23, NUMBER 1 2002

ACKNOWLEDGMENTS significance of properties. J We would like to thank Wendy Wilton for Speech Hear Disord 1981;46:291–296 help with preparation of the manuscript. This 15. Blache SE. A distinctive-feature approach. In: Creag- research was supported by a grant from the head NA, Newman PA, Secord WA, eds. Assessment and Remediation of Articulatory and Phonological National Institutes of Health to Indiana Uni- Disorders. New York: Macmillan; 1989:361–382 versity, Bloomington (DC01694). 16. Cooper R. The method of meaningful minimal contrasts in functional articulation problems. J Speech Hear Assoc Va 1968;10:17–22 REFERENCES 17. Costello J, Onstine J. The modification of multiple articulation errors based on distinctive feature the- 1. Jakobson R. Child Language, Aphasia and Phono- ory. J Speech Hear Disord 1976;41:199–215 logical Universals. The Hague, Netherlands: Mou- 18. Elbert M, Rockman B, Saltzman D. Contrasts: ton; 1968 The Use of Minimal Pairs in Articulation Train- 2. Trubetzkoy NS. Principles of Phonology. Los An- ing. Austin, TX: Exceptional Resources; 1980 geles: University of California Press (original work 19. Ferrier E, Davis M. A lexical approach to the re- published 1958); 1969 mediation of sound omissions. J Speech Hear Dis- 3. Chomsky N, Halle M. The Sound Pattern of Eng- ord 1973;38:126–130 lish. New York: Harper & Row; 1968 20. Gierut JA. The conditions and course of clinically- 4. Graham LW, House AS. Phonological oppositions induced phonological change. J Speech Hear Res in children: a perceptual study. J Acoust Soc Am 1992;35:1049–1063 1971;49:559–566 21. Haas W. Phonological analysis of a case of dyslalia. 5. Stevens KN, Keyser SJ. Primary features and their J Speech Hear Disord 1963;28:239–246 enhancement in consonants. Language 1989;65: 22. McReynolds LV, Bennett S. Distinctive feature 81–106 generalization in articulation training. J Speech 6. Dinnsen DA, Chin SB. Independent and relational Hear Disord 1972;37:462–470 accounts of phonological disorders. In: Yavas M, 23. McReynolds LV, Huston K. A distinctive feature ed. First and Phonology. San analysis of children’s misarticulations. J Speech Diego, CA: Singular; 1994:135–148 Hear Disord 1971;36:156–166 7. Dinnsen DA. Methods and empirical issues in an- 24. Weiner FF. Treatment of phonological disability alyzing functional misarticulation. In: Elbert M, using the method of meaningful minimal contrast: Dinnsen DA, Weismer G, eds. Phonological The- two case studies. J Speech Hear Disord 1981;46: ory and the Misarticulating Child (ASHA Mono- 97–103 graphs No 22). Rockville, MD: ASHA; 1984:5–17 25. Lado R. Linguistics across Cultures. Ann Arbor, 8. Stoel-Gammon C, Dunn C. Normal and Disor- MI: University of Michigan Press; 1957 dered Phonology in Children. Austin, TX: Pro- 26. Fey ME. Articulation and phonology: inextricable Ed; 1985 constructs in speech pathology. Lang Speech Hear 9. Dinnsen DA, Chin SB, Elbert M, Powell TW. Ser 1992;23:225–232 Some constraints on functionally disordered 27. Williams AL. Multiple oppositions: theoretical : phonetic inventories and phonotac- foundations for an alternative contrastive interven- tics. J Speech Hear Res 1990;33:28–37 tion framework. Am J Speech Lang Pathol 2000; 10. Dinnsen DA, Chin SB. Individual differences in 9:282–288 phonological disorders and implications for a the- 28. Williams AL. Multiple oppositions: case studies ory of acquisition. In: Eckman FR, ed. Confluence: of variables in phonological intervention. Am J Linguistics, L2 Acquisition and Speech Pathology. Speech Lang Pat 2000;9:289–299 Amsterdam: John Benjamins; 1993:139–154 29. McCabe R, Bradley D. Systematic multiple 11. Barlow JA. A Constraint-Based Account of Sylla- phonemic approach to articulation therapy. Acta ble Onsets: Evidence from Developing Systems. Symb 1975;6:1–18 Bloomington, IN: Indiana University; 1997 30. Dean E, Howell J, Hill A, Waters D. Metaphon 12. Chin SB. The Organization and Specification of Resource Pack. Windsor: NFER-Nelson; 1990 Features in Functionally Disordered Phonologies. 31. Dean EC, Howell J, Waters D, Reid J. Metaphon: Bloomington, IN: Indiana University; 1993 a metalinguistic approach to the treatment of 13. Stoel-Gammon C. Phonetic inventories, 15–24 phonological disorder in children. Clin Linguist months: a longitudinal study. J Speech Hear Res Phonet 1995;9:1–19 1985;28:505–512 32. Howell J, Dean E. Treating Phonological Disor- 14. Blache SE, Parsons CL, Humphreys JM. A mini- ders in Children: Metaphon—Theory to Practice. mal-word-pair model for teaching the linguistic London: Whurr; 1994 MINIMAL PAIR APPROACHES/BARLOW, GIERUT 67

33. Saben CB, Ingham JC. The effects of minimal 44. Barlow JA. The structure of /s/-sequences: evi- pairs treatment on the speech-sound production of dence from a disordered system. J Child Lang two children with phonologic disorders. J Speech 2001;28:291–324 Hear Res 1991;34:1023–1040 45. Gierut JA. Syllable onsets: clusters and adjuncts in 34. Smit AB. Phonologic error distributions in the acquisition. J Speech Lang Hear Res 1999;42: Iowa-Nebraska Articulation Norms Project: con- 708–726 sonant singletons. J Speech Hear Res 1993;36: 46. Gierut JA, Champion AH. Learning and the rep- 533–547 resentation of complex onsets. In: Greenhill A, 35. Smith MD, Brunette D. Early rampant homonymy: Littlefield H, Tano C, eds. Proceedings of the 23rd problem or strategy? Pap Rep Child Lang Dev Annual Boston University Conference on Lan- 1981;20:133–139 guage Development. Somerville, MA: Cascadilla 36. Gierut JA. Maximal opposition approach to Press; 1999:196–203 phonological treatment. J Speech Hear Disord 47. Gerken L, McGregor K. An overview of prosody 1989;54:9–19 and its role in normal and disordered child lan- 37. Gierut JA. Differential learning of phonological guage. Am J Speech Lang Pat 1998;7:38–48 oppositions. J Speech Hear Res 1990;33:540–549 48. Kehoe MM, Stoel Gammon C. Truncation pat- 38. Gierut JA. Homonymy in phonological change. terns in English speaking children’s word produc- Clin Linguist Phonet 1991;5:119–137 tions. J Speech Lang Hear Res 1997;40:526–541 39. Gierut JA, Neumann HJ. Teaching and learning 49. Kehoe M. Prosodic patterns in children’s multisyl- //: a nonconfound. Clin Linguist Phonet 1992;6: labic word productions. Lang Speech Hear Ser 191–200 2001;32:284–294 40. Labov W. The overestimation of functionalism. In: 50. Tyler AA. Examining phonological-morphological Dirven R, Fried V, eds. Functionalism in Linguis- interactions with converging sources of evidence. tics. Philadelphia: John Benjamins; 1987:311–332 Clin Linguist Phonet 1999;13:131–156 41. Ferguson CA, Menn L, Stoel-Gammon C. 51. Tyler AA, Watterson K. Effects of phonological : Models, Research, versus language intervention in preschoolers with Implications. Monkton, MD: York Press; 1992 both phonological and language impairment. 42. Jamieson DG, Rvachew S. Remediating speech Child Lang Teach Ther 1991;7:141–160 production errors with sound identification train- 52. Fey ME, Cleave PL, Ravida AI, Long SH, Dejmal ing. J Speech Lang Pathol Audiol 1992;16:201–210 AE, Easton DL. Effects of grammar facilitation on 43. Rvachew S. Speech perception training can facili- the phonological performance of children with tate sound production learning. J Speech Hear Res speech and language impairments. J Speech Hear 1994;37:347–357 Res 1994;37:594–607