Introduction

The acquisition of language is one of the signature achievements of childhood. Adequate language functioning includes the ability to understand what others say (also known as receptive language or language comprehension) and to produce utterances comprehensible to others (also known as expressive language or language production). Language capabilities can be differentiated further into several subdomains, including the knowledge and use of (a) words (also known as lexical, semantic, or vocabulary development); (b) sequences of words (also known as grammatical or syntactic development); (c) linguistic forms in social interactions (also known as pragmatic development); and (d) the system of speech sounds (also known as phonology). Children with specific language impairment (SLI) are broadly defined as displaying significant difficulties in one or more of these language domains. These difficulties cannot readily be explained by deficits in other aspects of development that are linked to language acquisition. Such aspects include intelligence, hearing, oral-motor skills, emotional developmental, and language exposure. However, due to the multiple aspects of language that may be affected, the difficulties of separating language from other developmental skills, and the variability in the course of language development within and across individual children, the precise criteria for diagnosing SLI have been the subject of debate for more than 20 years.

Research questions

1) How is SLI diagnosed?
2) Is SLI actually “specific”?
3) How early can SLI be diagnosed?

1) How is SLI diagnosed?

SLI is diagnosed when a child has significantly poor language skills despite the presence of normal-range intelligence, hearing, oral-motor skills, emotional development and language learning environment. Normal intelligence is usually defined as either a full-scale or nonverbal IQ score no more than one standard deviation (SD) below the mean; although specific thresholds for normal-range hearing, oral-motor skills, emotional development, and language learning environment have not been established, the general consensus is that a child should not be diagnosed with SLI if deficits in any of these domains could reasonably be held responsible for the child’s language deficits.

With respect to the language deficits in SLI, significantly poor language skills have generally been defined as scores placing the child in the lowest ten percent of children his or her age (i.e., approximately 1.25 standard deviations below the mean) on one or more well-constructed norm-referenced tests reflecting several of the linguistic domains and processes mentioned above (e.g., Tomblin et al., 1997). Formal analyses of the child’s language during conversation are also conducted in order to determine whether the number and type of words, grammatical constructions, and sentences are consistent with those expected for his or her age and language background (Leonard, 1998). Children with SLI generally talk less than typically developing peers from the same language community, producing shorter, less complex sentences and employing a more restricted set of vocabulary words.

Based on evidence that both norm-referenced tests and language sample analyses are affected by experiential factors such as sociodemographic factors and language background (Dollaghan et al., 1999), several investigators (e.g., Paul, 2007) have suggested that in addition to poor language test scores, the diagnostic criteria for SLI should also include significant caregiver or teacher concern about the child’s language skills when viewed within the context of his or her community. In addition, a number of investigations have suggested that measures of language that are less affected by the child’s language background and experiences, such as repeating nonsense words, may be helpful in identifying children with language disorders (e.g., Dollaghan & Campbell, 1998; Ellis Weismer et al., 2000; Graf Estes, Evans & Else-Quest, 2007). The increase in empirical evidence concerning accuracy of various criteria for diagnosing SLI, including formal tests and other measures (e.g., Dollaghan, 2004; Oetting, Cleveland & Cope, 2008; Perona, Plante & Vance, 2005; Spaulding, Plante & Farinella, 2006), means that diagnostic standards for the condition will become even better specified over the next few years.

2) Is SLI actually “specific”?

These general recommendations notwithstanding, several issues continue to be debated concerning the diagnosis of SLI. One ongoing controversy concerns the question of whether the deficits in children with SLI are actually specific to language. Many studies have shown that even when individual children’s IQ scores fall within normal limits, group mean IQs for children with SLI generally fall at the lower end of the normal range, significantly lower than average IQ scores for groups of children developing typically. Because even ostensibly nonverbal IQ assessments depend on language or language-related capabilities (including memory coding and rehearsal, meta-cognitive strategies, and the availability of a fund of general knowledge typically acquired via linguistic means), it is reasonable to question whether intelligence could ever be assessed without some confounding by language skills. Mounting evidence suggests that many children with SLI have significant difficulties in cognitive operations such as phonological coding, working memory, and the like (e.g., Archibald & Gathercole, 2007; Ellis Weismer et al., 2000; Montgomery & Windsor, 2007) that are linked not only to intelligence but also to perceptual-motor skills and prior experiences. Thus, although children with SLI by definition have normal IQs, deficits in other language-related aspects of cognition are likely.

3) How early can SLI be diagnosed?

Moderate-to-severe language deficits can be diagnosed by approximately three years of age, and a number of studies suggest that treatment is effective in improving the expressive vocabulary and grammatical skills of such children (Law, Garrett, & Nye, 2004). However, diagnosing SLI requires testing IQ, which is difficult to do with confidence in children younger than approximately four years of age. Although the desirability of identifying children with SLI as early as possible is widely acknowledged, efforts to determine valid and reliable diagnostic indicators of the condition in children younger than approximately four years of age have been notable for their lack of success. Several large studies (e.g., Dale et al., 2003) have shown that that neither late onset nor slowed growth of expressive vocabulary in two- or three-year-old children reliably predicts those who will exhibit the lexical and grammatical deficits associated with SLI at later ages. The fact that delayed and/or slowed language acquisition is characteristic of many other developmental disabilities, including autism spectrum disorders, hearing impairment, Fragile X syndrome, seizure disorders, and severe speech articulation disorders, further complicates efforts at early differential diagnosis of the condition. Although at present it appears that SLI cannot be diagnosed with confidence in children younger than four years of age (Paul, 2007), similar treatment recommendations apply regardless of whether the child’s deficits are determined to be specific to language. Accordingly, treatment may be warranted for language disorders identified at younger ages, whether or not a diagnosis of SLI has been made.

Directions for future research

There is a clear need for more credible evidence concerning the accuracy and precision of proposed indicators of SLI. As noted in the epidemiologic and evidence-based practice literature (e.g., Klee, 2008), the accuracy of a proposed diagnostic marker can only be determined by comparison to a reference standard; both the proposed diagnostic indicator and the reference standard must be administered independently and with masking to large samples of children representing the full spectrum of severity and the presence of other contributing variables. Future research is also needed to address the question of whether clinical course, optimal intervention approaches, prognoses, or long-term outcomes differ for children diagnosed with specific and nonspecific language impairments.