Since the 1960s, the term articulation disorder has been replaced in many circles by the term phonological disorder. This shift has been driven by the recognition that children with articulation disorders show general patterns in their speech that are not easily identified by an articulatory defect approach. Further, these patterns appear similar to those used by younger, typically developing children. The notion of phonological versus articulatory impairment, however, has not been examined in depth through comparisons of typically developing children and children with phonological impairment across a range of languages. Such comparisons would provide the most revealing evidence in support of one view over the other. If articulatory factors are behind children's phonological impairment, children with such impairments should show patterns somewhat independent of their linguistic environment. They should look more like one another than like their linguistically matched peers. If there is a linguistic basis to phonological impairment, such children should look more like their typically developing linguistic peers than like children with phonological delay in other linguistic environments. The basic structure of this research line is presented below, using English and Italian children as examples (TD = typically developing, PI = phonologically impaired): Cross-Linguistic Predictions Articulatory Deficit English TD + Italian TD = different English PI + Italian PI = same English TD + English PI = different Italian TD + Italian PI = different Linguistic Deficit English TD + Italian TD = different English PI + Italian PI = different English TD + English PI = same Italian TD + Italian PI = same

The pursuit of this line of research requires two steps. First, it needs to be verified that there are indeed cross-linguistic differences in phonological acquisition between typically developing children. The truth of this claim is not self-evident. Locke (1983), for example, has argued that children will be similar cross-linguistically until some point after the acquisition of the first 50 words. Elsewhere, however, it has been argued that such cross-linguistic differences are evident at the very earliest stages of phonological development (Ingram, 1989). Resolution of this issue will require extensive research into early typical phonological development in a range of languages. To date, the data support early cross-linguistic differences.

The second, critical step is to determine whether children with phonological delay look like their typically developing peers or like children with phonological impairment in other linguistic communities. The data on this issue are even more sparse than for the first step, but some preliminary data exist, and those data support the linguistic account of phonological delay over the articulatory one.

These two steps can be summarized as follows:

Examination of a range of studies on early phonological development shows that children in different linguistic environments converge in their acquisition toward a basic or core phonetic inventory of speech sounds that is different for each language. This is demonstrated here by examining the production of word-initial consonants in English, French, K'iche', and Dutch.

Below is an inventory of the English consonants typically used in the early stage of phonological acquisition (Ingram, 1981). English-speaking children show early acquisition of three place features, a voicing contrast among stops, and a series of voiceless consonants.

French shows some similarities to English, but also two striking differences, based on my analysis of selected diary studies. French-speaking children tend to acquire velar consonants later, yet show an early use of /l/, a sound that appears later in English.

K'iche' (formerly spelled Quiché) is a Mayan language spoken in Guatemala. K'iche' children (Pye, Ingram, and List, 1987) show an early /l/, as in French, and an affricate, /t∫/, which is one of the most frequent early sounds acquired. The first fricative tends to be the velar /x/, despite the fact that the language has both /s/ and /∫/.

Lastly, below is an intermediate stage in Dutch, based on data in Beers (1994). The Dutch inventory shows early use of the velar fricative /x/, as in K'iche', but also the full range of Dutch fricatives, which appear at more or less the same time.

Data such as these from English, French, K'iche', and Dutch show the widely different ways that children may acquire their early consonantal inventories. It has been proposed that these differences result from the varying roles these consonants play in the phonologies of the languages discussed (Ingram, 1989). The more frequently a consonant is used in a wide range of words, the more likely it is that it will be in the early inventory.

The question now is, what do the inventories of children with phonological impairment look like when comparisons like those above are done? The limited evidence to date indicates that they look like their same language typically developing peers. Data from Italian, Turkish, and Swedish suggest that this is the case.

The first language examined here is Italian, using the data reported in Bortolini, Ingram, and Dykstra (1993). Typically developing Italian children use the affricate /∫/ and the fricative /v/, both later acquisitions for English-speaking children. Italian-speaking children with phonological impairment have an inventory that is a subset of the one used by typically developing children. They do not have the affricate but do show the early acquisition of /v/. The early use of /v/ can be traced back to the fact that the voiced labiodental fricative is a much more common sound in the vocabulary of Italian-speaking children than it is in the vocabulary of their English-speaking peers.

Topbas (1992) provides data on Turkish for both typically developing children and children with phonological impairment. The Turkish inventory of the typically developing children is noteworthy for its lack of early fricatives, despite a system of eight fricatives, both voiced and voiceless, in the adult language. For English-speaking children, the lack of fricatives is often a sign of phonological impairment, but this appears to be expected for typically developing Turkish-speaking children. The data on phonological impairment are from a single child at age 6 years, 0 months. This child shows the same lack of fricatives and the early affricate, just as in the typically developing data.

Lastly, Swedish data are available to pursue this issue further (Magnussen, 1983; Nettelbladt, 1983). The data on typically developing Swedish children are based on a case study of a child at age 2 years, 2 months. This child lacked the velar stop and the voiced fricative /v/. The voiced stops were also missing in a group of ten children with phonological impairment; however, these children did show early use of /v/, just as the Italian-speaking children did.

Thus, preliminary data from a range of languages support the phonological rather than the articulatory account of phonological impairment. This results in two preliminary conclusions: (1) Typically developing children show early phonological inventories unique to their linguistic environment. (2) Children with phonological impairment show systems more similar to those of their typically developing peers in their own linguistic enviroment than to those of children with phonological impairment in other language environments.