The phoneme /r/ is one of the most frequently misarticulated sounds that a speech-language pathologist (SLP) hears. Despite advances in research, its misarticulation is still considered one of the most difficult sounds to correct and one of the most persistent speech errors in school-aged children (Clark, Schwarz, & Blakeley, 1993; Elbert & McReynolds, 1975). According to a recent survey of approximately 200 SLPs (plus anecdotal evidence), many SLPs are not making adequate progress with students who have /r/ disorders, and they feel inadequately prepared in diagnostic and evaluation procedures (Ristuccia, 2004). These factors, coupled with a critical shortage of speech-language pathologists, mandate the investigation of new, research-based approaches relevant to school settings. Thus, the discussion presented here seeks to demonstrate that using an approach based on phonetically consistent word lists, which use 21 variations of vocalic /r/, will improve students' speech production.
The Traditional Approach to Articulation Therapy & How Many /r/'s Are There?
As previously stated, misarticulation of /r/ presents a significant challenge to the speech student and clinician. Several reasons have been presented in the literature to account for the persistence of /r/ speech errors. A primary reason is the variability in the manner of /r/ production (Zawadzki & Kuehn, 1980). The phoneme /r/ has allophones that may be considered consonantal forms, and others that are considered vocalic in nature. Given the variable nature of the /r/ phoneme, it is possible that its complexity is not adequately addressed in traditional approaches of articulation remediation, also known as the "stimulus approach" (Van Riper, 1978). In the traditional approach, the focus is on the single misarticulated sound, in this case, consonantal /r/. Traditional articulation remediation is characterized as a continuum ranging from ear/perceptual training to hierarchically arranged steps of production training. Depending upon the level of stimulability of the sound in error, production training could begin by establishing correct production of the sound in isolation, syllables, words, phrases, sentences, and ultimately ending with correct production in conversation. Although the approach is generally considered to be effective at speech remediation, most would agree that it is often not effective with remediation of /r/ errors.
In the traditional approach, the phoneme /r/ is simply classified in terms of word position: prevocalic (before a vowel, as in red); intervocalic (between vowels, as in arrow); and postvocalic (after a vowel, as in father). However, actual pronunciation of /r/ is far more complex. It comprises six vocalic combinations as well as pre-vocalic /r/, initial /r/ blends, and medial and final /r/. The prevocalic classification of the traditional approach works for words such as red, because /r/ in the initial position functions as a consonant. However, the other initial vocalic /r/s (as in Ernie, earphone, airplane, Archie, ornament, and Ireland) cannot be classified as prevocalic/consonantal because a vowel precedes the /r/. It would be difficult to remediate [er] initial words (e.g., Ernie) using prevocalic/consonantal /r/ words such as red and run. Establishment/training on /r/ in the traditional approach often does not distinguish among the different allophones of /r/, especially the vocalic allophones. Therefore, traditional practice on "medial /r/" might involve the mixing of disparate words in practice lists such as carpet, butterfly, parachute, hamburger, or bird. Taking the complexity and variability of articulation of /r/ into account, it is apparent why the traditional approach might not be very effective in correcting the production of misarticulated /r/ (Clark et al., 1993). One possible reason given for the lack of success is the variability of the /r/ phoneme that is not taken into account, particularly with /r/ allophones that function as vowels, such as [er].
The Rationale for Targeting Different /r/ Allophones
The difficult and intricate nature of /r/ is well-documented in the literature. The phoneme /r/ is generally considered to be a consonant, especially when used in the pre-vocalic position, such as ring, race, and run. However, as noted above, the phoneme /r/ has vowel-like properties and therefore is labeled as a glide, a semi-vowel, or a liquid. The position of /r/ changes the quality of the vowel that immediately precedes it in such words as bear, pear, pour, and arrow (Carmona, 1987), clearly affecting the position of the tongue and mandible. Clinically, this suggests that /r/ articulation is variable depending on the preceding vowel. Researchers suggest that postvocalic /r/ could be considered the second element of a diphthong, and that it varies as a function of the preceding vowel, in contrast to the steady state and lower formant frequency of the prevocalic /r/ (Zawadzki & Kuehn, 1980).
Research supports the relevance of the /r/ division in assessment and treatment of /r/ disorders. Pollock (1991) suggests that clinicians consider six vocalic /r/s: (1) the unstressed / ɚ/ as in father, (2) the stressed /ɝ/ as in bird, (3) /ɪɚ/ as in fear, (4) / Ɛɚ/ as in fair, (5) /ɑɚ/ as in car, and (6) / Ɔ ɚ/as in oar. Shine, Downes, & Denning (1982) postulated 10 phones of /r/: (1) consonantal (or prevocalic) /r/ as in red; (2) voiceless fricative di- and tri- clusters as in price and spring; (3) voiced fricative di-clusters as in broke; (4) /ɪɚ / as in fierce; (5) /Ɛɚ/ as in shared; (6) / Ɔ ɚ/ as in fork; (8) /ʊr/ as in toured; (9) stressed / ɝ / as in hurt; and (10) unstressed / ɚ / as in sister. Curtis and Hardy (1959) identified four classes of phonetic events in which their subjects showed differential behavior: (1) the consonantal /r/; (2) the stressed vocalic / ɝ /; (3) the unstressed vocalic / ɚ / as in butterfly; and (4) the intervocalic / Ɛɚ / as in share. Interestingly, Curtis and Hardy documented their subjects' ability to produce 43 different /r/ combinations divided into seven distinct classifications depending upon context. They concluded:
Subjects behave differently with respect to the different types of /r/