Linguistic approaches to treating children's speech disorders are motivated by the fact that a phonology is a communication system (Stoel-Gammon and Dunn, 1985; Ingram, 1990; Grunwell, 1997). Within this system, patterns are detectable within and among various subcomponents: (1) syllable and word shapes (phonotactic repertoire), (2) speech sounds (phonetic repertoire), (3) the manner in which sounds contrast with each other (phonemic repertoire), and (4) the behaviors of different sounds in different contexts (phonological processes). Each of these subcomponents may influence the others; each may interfere with successful communication. Therefore, individual sound or structure errors are treated within the context of the child's whole phonological system rather than one by one; remediation begins at the level of communicative function—the word.

The assumption underlying this type of treatment is that the child's phonological system—his or her (subconscious) mental organization of the sounds of the language—is not developing in the appropriate manner for the child's language or at a rate appropriate for the child's age. The goal is for the child to adjust his or her phonological system in the needed direction. Initiating a change in one part of the phonological system is expected to have a more general impact on the whole system.

The sounds of a language are organizable into various categories according to their articulatory or acoustic features. The consonants /p, t, k, b, d, g/, for example, are all members of the category stops. They are noncontinuant because the airflow is discontinued in the oral cavity during their production. Some stops also have the feature voiceless because they are produced without glottal vibration. Each speech sound can be categorized in different ways according to such features. These features are called distinctive because they differentiate each sound from all others in the language. The (implicit) knowledge of distinctive features is presumed to be one organizational basis for phonological systems.

Distinctive features therapy (McReynolds and Engmann, 1975) focuses on features that a child's system lacks. A child who produces no fricatives lacks the continuant feature, so distinctive features therapy would focus on this feature. In theory, establishing a continuant in any place of articulation (e.g., [s]) would lead to the child's generalization of the feature to other, untrained fricatives. Once the feature is included in the child's feature inventory, it can be combined with other features (e.g., voice, palatal) to yield the remaining English fricatives.

The function of distinctive features is to provide communicative contrast. The more features we use, the more lexical distinctions we make and the more meanings we express. A child who produces several target phonemes identically (e.g., all fricatives as [f]) will have too many homonyms. Therapy will focus on using more distinctive features, for example, producing different target phonemes distinctively. Williams (2000a, 2000b) recommends simultaneously contrasting all target sounds with the overgeneralized phoneme. In some disordered phonologies, features are noncontrastive because they are used in limited positions. For a child who uses voiced stops only in initial position and voiceless stops only in final position, for example, “bad,” “bat,” “pad,” and “pat” are homonyms because they are all pronounced as [bæt]. Conversely, some children with disordered phonology may maintain a contrast, but without the expected feature. A child who does not voice final stops may lengthen the preceding vowel to indicate voicing; “bat” as [bæt] and “bad” as [bæ:t]. Such a child has phonological “knowledge” of the contrast and may independently develop voicing skills. Therefore, Elbert and Gierut (1986) suggest that features of which children have little knowledge are a higher priority for intervention. In other studies (e.g., Rvachew and Nowak, 2001), however, subjects have made more progress when most-knowledge features were addressed first.

Often, distinctive features are remediated with an emphasis on contrast, through minimal pair therapy. Treatment focuses on words differing by one distinctive feature. For example, the continuant feature could be taught by contrasting “tap” versus “sap” and “met” versus “mess.” Typically the sound that the child substitutes for the target sound (e.g., [t] for /s/) is compared to the target sound ([s]). Therapy may begin with discrimination activities; the child indicates the picture that corresponds to the word produced by the clinician (e.g., a hammer for “tap” versus an oozing tree for “sap”). This highlights the confusion that may result if the wrong sound is used. Communication-oriented production activities are designed to encourage the child to produce the feature that had been missing from his system (e.g., to say “mess” rather than “met”). (See phonological awareness intervention for children with expressive phonological impairments for a discussion of approaches in which the contrastive role of phonological features and structures is even more explicitly addressed.)

Gierut (1990) has tested the use of maximal pair therapy, in which the contrasting sounds differ on many features (e.g., [s] versus [m]). She has found that children may be able to focus better on the missing feature (e.g., continuant) when it is not contrasted with the substituting feature (e.g., noncontinuant) than when it is.

Markedness is another phonological concept with implications for remediation of sounds. Markedness reflects ease of production and perception: [θ] is marked, due to low perceptual salience; [b] is unmarked because it is easy to produce and to perceive. Children typically acquire the least marked sound classes (e.g., stops) and structures (e.g., open CV syllables) of their languages first (Dinnsen et al., 1990). It is unusual for a child to master a more marked sound or structure before a less marked one. Gierut's (1998) research suggests that targeting a more marked structure or sound in therapy facilitates the acquisition of the less marked one “for free.” However, other researchers (e.g., Rvachew and Nowak, 2001) report more success addressing less marked sounds first.

Therapy based on nonlinear phonology stresses the importance of syllable and word structures as well as segments (Bernhardt, 1994). Minimal pair therapy is often used to highlight the importance of structure (e.g., “go” versus “goat,” “Kate” versus “skate,” “monkey” versus “monk”). The goal of this therapy is to expand the child's phonotactic repertoire. For example, a child who previously omitted final consonants may or may not produce the correct final consonant in a given word, but will produce some final consonant. Structural and segmental deficits often interact in such a way that a child can produce a sound in certain positions but not others (Edwards, 1996).

The approaches described above focus primarily on what's missing from the child's phonological system. Another set of approaches focuses on what's happening instead of the target production. For example, a child whose phonological system lacks the /θ, ð, s, z/ fricative phonemes may substitute stops (stopping), substitute labial fricatives ([f, v]; fronting), substitute palatal fricatives ([]; backing or palatalization), or omit /θ, ð, s, z/ in various word positions (initial consonant deletion, final consonant deletion, consonant cluster reduction). These patterns are referred to as phonological processes in speech-language pathology.

Phonological process therapy addresses three types of error patterns:

As in distinctive feature therapy, in phonological process therapy classes of sounds or structures that pattern together are targeted together. Again, either the entire class can be directly addressed in therapy or some representative members of the class may be selected for treatment, in the expectation that treatment effects will generalize to the entire class. The goal is to reduce the child's use of that process, with resultant changes in her phonological system. For example, if the child's phonological system is expanded to include a few final consonants, it is expected that she or he will begin to produce a variety of final consonants, not just those that were targeted in therapy.

Some therapists use traditional production activities (beginning at the word level) to decrease a child's use of a phonological process; others use a minimal pair approach, comparing the targeted class (e.g., velars) with the substituting class (e.g., alveolars). The cycles approach (Hodson and Paden, 1991) to process therapy has some unique features. First, each session includes a period of auditory bombardment, during which the child listens (passively) to a list of words that contain the targeted sound class or structure. Second, each pattern is the focus for a predetermined length of time, regardless of progress. Then, treatment moves on to another pattern. Hodson and Paden argue that cycling is more similar to the phonological development of children without phonological disorders.

In summary, the goal of linguistically based approaches to phonological therapy is to make as broad an impact as possible on the child's phonological system, making strategic choices of treatment goals that will trigger changes in untreated as well as treated sounds or structures.