Elicitation Task for Subject–Verb Agreement

Jan de Jong

The marking of verbs for grammatical features of the subject of a sentence is called subject–verb agreement. Features that are marked this way are number, person and sometimes gender. Marking of agreement may be fused with tense. For instance, the third person singular –s in English verbs is used only for the present tense. Importantly, agreement is an inflection that is determined by the syntactic context in which the verb is used (contextual inflection; Booij, 1994).

The agreement paradigm is different for each language where agreement occurs. Some languages (notably English) have very sparse paradigms, while other languages (e.g. Italian) have paradigmatic cells not only for singular and plural but also for first, second and third person. Slovene, for instance, has dual as well as plural marking. On the other hand, some languages (e.g. Swedish) do not have verb agreement morphology (although tense is marked in Swedish). There are also within-language differences between the marking for agreement of present and past.

Subject–Verb Agreement as a Vulnerable Area in Specific Language Impairment

Verb morphology is known to be a vulnerable domain in specific language impairment (SLI) in many languages, though certainly not in all: in some languages, noun morphology is more vulnerable than verb morphology (cf. Chapter 2 on case morphology). There are also cross-linguistic differences in the severity of verb agreement difficulties in SLI. Research by Leonard (1998) has revealed that inflection is better preserved if it is highly salient in the target language. The same is true for languages that have rich (or uniform) morphology. Together, these characteristics make agreement a strong cue in such languages. According to Leonard, this explains, for instance, why accuracy rates for inflection on verbs are much higher in the output of children with SLI in Italian than in English.

In the languages where verb inflection is affected, difficulties with verbal morphology are often seen as a reliable marker of SLI. Some authors claim that this marker should be identified with tense rather than agreement (Rice et al., 1995). However, the language for which the claim of tense as a clinical marker has most consistently been made, English, does not distinguish between overt marking of (present) tense and agreement, as in third person singular –s (the only overt affix in the present tense inflectional paradigm).

Other researchers, notably Clahsen (2008), claim that agreement problems outnumber tense problems and that the reverse pattern, i.e. impaired tense marking and intact subject–verb agreement marking, does not seem to exist in SLI (Clahsen, 2008: 176). In earlier work, Clahsen (1992) proposed that the contextual nature of verb agreement, captured in the Control Agreement Principle, is where the weakness of children with SLI should be located. This would explain why other features, such as plural marking on nouns, are better preserved in SLI: number marking on nouns is not contextual.

Tense and agreement errors may show as omissions or substitutions of inflectional morphemes. In English, a language with little overt marking in the first place, omission is the dominant error type. In languages with a more elaborate verb paradigm, substitutions are also found. Recently, Clahsen (2008) has suggested that problems in agreement may not show across the board, but rather in specific cells of the agreement paradigm. These affected cells may be language specific. This, again, may be explained alternatively by differential patterns of saliency (Bedore & Leonard, 1996). Another factor concerns the number and complexity of features to be marked. In Hebrew, past tense forms are marked for person, number and gender; in the present tense, person is not marked. Dromi et al. (1999) showed that this led to a discrepancy in performance for the present and past tense. They found that errors in second person past tense (masculine and feminine, singular and plural), as well as errors in present tense plural feminine, revealed differences between children with SLI and typically developing (TD) children. They characterised the errors as feature simplification/reduction since they involved omission of the person marking in the past and the number or gender marking in the present.

When choosing a test domain for verb morphology that is valid across languages, however, agreement is a proper target, languages that lack agreement notwithstanding. The variation of inflectional paradigms makes it a fruitful area for research into the sources of linguistic disability in SLI. Studies on agreement morphology in monolingual children with SLI have addressed typologically different languages like German (Clahsen, 2008; Rothweiler et al., 2012), Hebrew (Dromi et al., 1999), Spanish (Bedore & Leonard, 2001), Italian (Leonard et al., 1992), Finnish (Kunnari et al., 2011), French (Franck et al., 2004) and Greek (Stavrakaki et al., 2008). These studies concern different linguistic contexts, test different hypotheses and sometimes propose different explanations for agreement difficulties in SLI. Together, however, they demonstrate that agreement morphology is a domain that warrants investigation when assessing the linguistic profile of SLI.

Agreement morphology is also sometimes found to be weak in bilingual children without language impairment (e.g. Paradis & Crago, 2000, 2003). This observation leads to a diagnostic confound: are agreement difficulties in bilingual children with language delay due to their bilingualism or are they indicative...