Abstract
Noun-related morphosyntax has not been emphasized in the literature on children with specific language impairment (SLI), yet, across languages, problems in this area are quite apparent. This review is designed to highlight noun-related difficulties that seem to be especially troublesome for these children. A review of the research literature on children with SLI was conducted, including work on a wide range of languages. Children with SLI were found to have weaknesses greater than those of younger typically developing compatriots in such noun-related areas as articles, possessive inflections, full subject pronouns, agglutinating noun suffixes, and pronominal clitics. Factors seemingly responsible for these difficulties include prosody, case assignment, agreement, and both grammatical and phonological opacity. Each language reviewed provides evidence of serious problems with select noun-related morphemes in the speech of children with SLI. The emphasis in language assessment will necessarily have to vary across languages, not because children vary in their ability to overcome the obstacles that these factors present, but rather because the typology of a language will dictate the extent to which these factors are operative in the language.
Introduction
Children with specific language impairment (SLI) show a significant deficit in language ability while exhibiting normal hearing, earning age-appropriate scores on nonverbal tests of intelligence, and showing no signs of frank neurological damage or disease (Bishop, 2014b; Leonard, 2014a). Many of these children show subtle weaknesses in select areas of motor and nonlinguistic cognitive processing, which has led some investigators to prefer alternative terms for these children (e.g., primary language impairment, developmental language disorder), though SLI remains by far the most frequent term applied to these children in the past 30 years (Bishop, 2014a; Leonard, 2014b).
Much investigative attention has been directed toward the morphosyntactic weaknesses of children with SLI. Although these weaknesses can cover a wide range, special attention has been devoted to these children’s difficulties with the verb system, especially morphemes that express tense and agreement. In English, children with SLI show less use of tense and agreement verb inflections as well as copula and auxiliary be forms than younger typically developing peers matched on mean length of utterance (e.g., Conti-Ramsden, Botting, & Faragher, 2001; Eadie, Fey, Douglas, & Parsons, 2002; Hoover, Storkel, & Rice, 2012; Joseph, Serratrice, & Conti-Ramsden, 2002; Leonard, Eyer, Bedore, & Grela, 1997; Oetting & Horohov, 1997; Rice & Blossom, 2013; Rice & Wexler, 1996) or number of verbs in their inventory (Leonard, Miller, & Gerber, 1999). In languages such as German, Swedish, and Dutch, such differences are also apparent, even though the absolute levels of use are somewhat higher than in English (e.g., Blom, Vasić, & de Jong, 2014; de Jong, 1999; Hansson, Nettelbladt, & Leonard, 2000; Ott & Höhle, 2013). In Romance languages, verb inflections seem less severely affected, but auxiliary forms involving tense and agreement are quite weak relative to younger typically developing peers (Leonard, Bortolini, Caselli, McGregor, & Sabbadini, 1992; Paradis & Crago, 2001). Children with SLI learning other types of languages have also been reported to have special difficulties with verb-related details, such as the use of grammatical aspect in Arabic (Abdalla & Crago, 2008) and Cantonese (Fletcher, Leonard, Stokes, & Wong, 2005).
In English, morphemes related to nouns have received less attention in the SLI literature. When taken as a group, these morphemes seem to present less serious problems than verb-related morphemes for children with SLI (Bedore & Leonard, 1998), However, several specific noun-related morphemes can be especially difficult for these children. In many other languages, noun-related problems can represent the dominant deficit in SLI. The purpose of this review is to describe many of the noun-related problems across different languages, and discuss the factors that seem to underlie them. Collectively, the evidence should serve to urge careful examination of the noun-related morphosyntactic abilities of children with SLI, as these may lie closer to the core of a child’s language problems than previously assumed.
Noun-related morphemes and factors that may impede their development
We define ‘noun-related’ quite broadly, as grammatical morphemes that can be associated with nouns. The morphemes considered include noun plural inflections, articles, possessive inflections, full pronouns, agglutinating noun suffixes, and pronominal clitics. For each morpheme type, we discuss the factors that might make its acquisition challenging for all children – even those showing typical language development. These may represent fault lines that could create especially serious problems in children with SLI. We then follow with a review of the actual evidence.
It will be seen that the factors that seem to contribute to noun-related weaknesses cut across languages and across morpheme types. We will argue that these factors probably serve universally as potential obstacles to children with SLI. The crosslinguistic differences seen in the SLI profile are not because these factors are more easily overcome by children in some languages than in others, but rather that these factors do not come into play in languages with particular typologies.
To make a case that a noun-related morpheme stands out as problematic for children with SLI, we must go beyond the simple and common finding that the children are less proficient with the morpheme than their same-age typically developing peers. Children with SLI have more difficulties than their age-mates in learning many details of language, so finding that some noun-related morpheme is also below age-level would not be surprising. To identify areas of potentially greater importance, then, we look for evidence of extraordinary difficulty with the noun-related morpheme. Most often, this is seen when children with SLI show a greater weakness with the morpheme than younger typically developing children who are similar to the children with SLI on some detail of language that represents an important control. For example, when some noun-related morphemes are produced, they increase children’s utterance length. Others depend on the children having enough words in their vocabulary to create the need to express the noun-related morphemes in the first place. In the studies reviewed here, we focus primarily on comparisons between children with SLI and younger typically developing children who are matched on such things as mean length of utterance (MLU) or vocabulary scores.
Noun plural inflections
Inflections that express the plural number of nouns are probably the most straightforward of the noun-related elements because they are an integral part of the lexical category, Noun. In the languages represented in the SLI literature, the plural is formed by attaching an inflection at the end of a bare noun stem (e.g., book-books in English), or appearing in word-final position instead of a singular inflection (e.g., libro-libri ‘book-books’ in Italian). Because plurals often have referents that can be verified in the real world (e.g., several books as opposed to one book), their function is quite transparent. If they pose a learning challenge, it is likely to be in their articulatory or phonological requirements. In some languages, the plural is formed by adding a word-final consonant, which, at a minimum would result in a closed syllable (as in keys) and, often, a word-final consonant cluster (as in books). Perception may represent less of an obstacle because, in these languages, nouns often appear in sentence- or clause-final position, where both sibilants and vowels are lengthened, resulting in greater duration than if they had appeared elsewhere in the sentence (Hsieh, Leonard, & Swanson, 1999).
The evidence
The early studies on English suggested that the consonantal nature of the noun plural might serve as a source of difficulty. English-speaking children with SLI appeared to show less use of plurals than younger children matched for MLU (Leonard et al., 1992; Leonard et al., 1997; Rice & Oetting, 1993). Even with these group differences, percentages of use by the SLI groups were rather high. Other studies of English reported either no differences or differences that, while statistically significant, were viewed as not clinically significant due to high levels of use by the children with SLI (Oetting & Rice, 1993; Polite, 2011; Rice & Wexler, 1996).
In Spanish, noun plurals – /s/ or /es/ – also involve a word-final consonant, though, unlike English, plurals in consonant clusters (as in books in English) are not expected. In a study making use of probe items, Bedore and Leonard (2001) found that children with SLI were more likely to omit the plural inflection than a group of younger typically developing children matched for MLU. However, an examination of the same children’s spontaneous speech failed to reveal differences (Bedore & Leonard, 2005). Finding only small differences between children with SLI and MLU controls, Grinstead, Cantú-Sánchez, and Flores-Ávalos (2008) concluded that noun plurals were not central to the morphosyntactic deficit in Spanish SLI.
Noun plurals in Swedish would appear to pose even fewer articulatory challenges to children with SLI. Although consonants are included in most contexts, the largely syllabic nature of Swedish plurals (e.g., stol-stolar ‘chair-chairs’) renders them identifiable as attempts at plural in most instances. Leonard, Salameh, and Hansson (2001) found only small, nonsignificant differences between children with SLI and younger MLU controls.
In Hebrew, noun plurals are also syllabic, and are distinguished according to gender (-im masculine, -ot feminine). Neither Rom and Leonard (1990) nor Dromi, Leonard, and Shteiman (1993) found differences between children with SLI and typically developing children matched according to MLU.
In a study of Hungarian-speaking children, Lukács, Leonard, and Kas (2010) compared the use of the plural suffix by children with SLI and younger typically developing children matched on vocabulary test scores. In this language, the plural suffix involves the addition of -k, often with a vowel that links it to the stem. Examples include hajó-hajók ‘ship-ships’ and madár-madarak ‘bird-birds.’ Notice in the second example that there is sometimes a change in the vowel in the stem; without this change the plural would be madárak, which is incorrectly formed. Lukács et al. found that the children with SLI were as accurate as the younger controls when only the plural suffix had to be added to the noun. Later, we will see that this changes when the plural suffix is accompanied by another suffix.
Italian is a language whose noun plural inflections are exclusively vowels, though a distinction is made for gender. Rather than representing an addition to a stem, use of the plural is more properly characterized as changing the singular ending to the plural (as in -o to -i in uccello-uccelli ‘bird-birds’). Leonard et al. (1992) found no differences between children with SLI and younger MLU controls.
The findings for noun plural inflections indicate that this area of morphosyntax is not highly problematic for children with SLI. Group differences seem to be confined to languages whose plural forms involve word-final /s/ or /z/, and, even in these languages, the results are mixed.
Articles
Articles represent another type of noun-related element that might cause difficulties for children with SLI. According to certain linguistic frameworks (e.g., Radford, 2004), articles are elements associated with a functional category. In particular, articles form part of the functional category of Determiner, whose structure is shown in (1).
Here, the phrasal projection Determiner Phrase (DP) branches into a specifier (Spec) and an intermediate category, D’. This, in turn, branches into Determiner (D), considered the head, and another phrasal projection, Noun Phrase (NP). It is assumed that articles (e.g., a, the in English) are represented in D. Elements from functional categories tend to be acquired later than lexical category elements, and this is certainly true for children with SLI.
When articles make an indefinite–definite distinction, as in English a–the, a linguistic pragmatic function is also involved. Schafer and deVilliers (2000) proposed that in children, this function is most relevant to the definite article. Specifically, they suggested that an indefinite phrase such as a car might be equivalent to a number phrase as in one car, whereas use of the definite article (the car) requires children to recognize those nouns whose referents are identifiable by the listener.
The functional category and pragmatic status of articles are not the only obstacles to acquiring these elements. In most languages, articles are monosyllabic forms that are in a less prosodically privileged position. Consider the structure appearing in (2), adapted from the Nespor and Vogel (2007) framework of prosodic phonology.
CG refers to a clitic group (also called a composite group) that consists of a word from a lexical category (noun, verb, adjective, adverb), which may be inflected, and all adjacent phonologically interacting grammatical morphemes – an article in the case of (2). One or more clitic groups will form a Phonological Phrase (PhP). The PhP will house all CGs within a single syntactic phrase. In (3), the PhP contains two CGs.
In certain languages, including English, short PhPs can be merged, a process that occurs more readily when one is the complement of the other, as in (4), where the car serves as the complement of wash.
The relevance of this structure to articles and certain other noun-related elements is that the prosodic position of the element may dictate the children’s success in producing it. For example, for some time, researchers have noted that children acquiring Germanic languages have a strong preference for strong–weak syllable sequences (e.g., basket, jacket) over weak–strong sequences (e.g., balloon, giraffe) when producing words, owing to both the dominant prosodic pattern of their ambient language and, possibly, articulatory factors. The relative disadvantage of weak–strong sequences is seen not only at the single-word level, but also when such a sequence involves an article with a noun, as in the car. The article, as a weak syllable preceding a strong syllable, is vulnerable to omission. The word-plus-inflection combination going, in contrast, is less vulnerable on prosodic grounds because it involves a strong–weak syllable sequence.
However, when the clitic group the car can be merged with that of the preceding verb, as in (4), the article can participate in a strong–weak sequence (wash the). Such a prosodic reorganization results in a greater likelihood of successful production (e.g., Gerken, 1996).
In many languages, articles must agree with their associated nouns in number and gender. Furthermore, in some languages, the syntactic distinction between, say, the subject of the sentence and the direct object of the sentence – a distinction of case – is marked on the article rather than the noun.
In a language such as German, only the article marks case; the noun is unchanged. Thus, the article der is used with the masculine singular noun Hund ‘dog’ if it is used in subject position, but the article den is used with Hund if it is used as the direct object. The omission of an article in such languages, then, can cause a problem not seen in many other languages. In other languages, omission will remove specification of the definite–indefinite distinction. For languages whose articles carry gender and number (e.g., Spanish), this information will be lost only if gender and number are not already clear from the noun. However, when articles also mark case, as in German, the syntactic role of the noun may be unclear, especially if, for pragmatic reasons, the speaker chooses a non-canonical word order.
Articles are not the only free-standing morphemes that mark case. In Hebrew, there is an accusative case marker used when the direct object is definite. This form, et, is a weak monosyllable and thus may pose a prosodic as well as morphosyntactic challenge.
The evidence
Several studies of English have reported lower levels of article use by children with SLI than by MLU controls (e.g., McGregor & Leonard, 1994; Polite, Leonard, & Roberts, 2011; Rice & Wexler, 1996), though Leonard et al. (1992) found only a nonsignificant trend favoring younger MLU-matched children. Some studies have combined the indefinite a and definite the in their calculations (e.g., Leonard et al., 1992), though other investigations have treated the two types of articles separately (e.g., Polite et al., 2011).
Polite et al. (2011) adapted tasks from Schafer and deVilliers (2000) to determine if English-speaking children with SLI had difficulty with both indefinite and definite articles, or problems primarily with only one of these. Examples of indefinite article items and definite article items can be seen in (5) and (6), respectively.
Stephen wants to call home. What can he use to call home with? (5) A little boy was playing with his toys. He had a car and an airplane. (6) One of them had wings. Guess which?
The children with SLI were indistinguishable from MLU controls in their accuracy on indefinite article items, but were less accurate than controls on definite article items. Unlike other studies that report frequent omissions of articles, the children with SLI in this study produced more substitutions (replacing the with a) than omissions of definite articles. It was also true, however, that omissions, though fewer in number than substitutions, were more numerous than omissions by the MLU control children. The difference in accuracy between definite and indefinite articles suggests that the linguistic pragmatic function of definite articles may have played a role in the children’s difficulty.
Earlier it was noted that the weak-syllable nature of articles in many languages is probably a factor in children’s difficulty with this morpheme type. McGregor and Leonard (1994) pursued this question in a study of English. They found that articles that were in sentence-initial position – where they could not be adjoined to a preceding strong syllable – were especially likely to be omitted. Articles in sentence-medial position that could align with a strong syllable were more likely to be produced. This finding also held for young MLU controls, though the SLI group showed less article use overall.
Article use appears to be a major obstacle for children with SLI learning Italian. Several studies report less article use by children with SLI than by MLU controls (e.g., Bortolini, Caselli, & Leonard, 1997; Bottari, Cipriani, Chilosi, & Pfanner, 1998, 2001; Leonard & Bortolini, 1998; Leonard et al., 1992). Bortolini, Caselli, Deevy, and Leonard (2002) asked a different type of question, whether a measure of article use produced good diagnostic accuracy. In one of two studies, they found both good sensitivity (the correct identification of children with language impairments) and specificity (the correct identification of truly typically developing children). However, in the second study, the statistically significant group differences were not large enough to meet the more stringent criterion of diagnostic accuracy.
Omissions of the article are by far the most frequent type of error seen in Italian-speaking children with SLI, and prosody has been cited as a major factor in such errors (Leonard & Bortolini, 1998). When substitutions occur, they tend to be based on over-use of more frequent article forms, at the expense of less frequent article forms of the same number and gender. For example, although il is the most common definite article for masculine singular nouns, lo is used for nouns with select word-initial consonants (/z/) or consonant clusters (e.g., /st/). Yet, children with SLI, like younger controls, will sometime produce, for example, il zaino rather than lo zaino ‘the backpack.’
In Spanish, article use appears to be more problematic for children with SLI, whether these children are compared directly to MLU-matched peers (Bedore & Leonard, 2005), or MLU is used as a covariate (Restrepo & Gutiérrez-Clellan, 2001). Omissions are the most frequent type of error (e.g., Anderson & Souto, 2005). However, Restrepo and Gutiérrez-Clellan (2001) reported a number of substitution errors. The larger proportion of substitution errors in Spanish compared to Italian might be partially explained by prosodic factors, as the durations of weak syllables in articles in Italian are shorter than in Spanish (Delattre, 1966), notwithstanding the typological similarities between these two languages.
The evidence is somewhat mixed for articles in French SLI. Le Normand, Leonard, and McGregor (1993) found no differences between children with SLI and younger MLU control children, though differences favoring younger controls were observed by both Parisse and Maillart (2007) and Pizzioli and Schelstraete (2008). The latter authors also found that changes from intransitive sentence structures to transitive structures created a larger decline in article use by the children with SLI than by the control children.
Swedish differs from many other languages because, whereas indefinite forms are articles (en or ett depending on the gender of the noun), definite forms are suffixes (-en, -et) attached to the ends of nouns. Thus, there are indefinite–definite distinctions such as en stol-stolen ‘a chair’-‘the chair’ and ett tåg-tåget ‘a train’-‘the train’). Studies comparing children with SLI and MLU controls have found lower degrees of indefinite article use by the SLI group, with no group differences for definite suffixes (Håkansson & Nettelbladt, 1996; Hansson, 1997; Hansson & Nettelbladt, 1995; Leonard et al., 2001). Omissions of the indefinite article are most common, though occasionally the more frequent article en substitutes for the less frequent article ett.
In one sense, these findings from Swedish are surprising, because whereas Polite et al. (2011) found definite articles to be more vulnerable than indefinite articles in English, in Swedish, the definite forms are more consistently produced than the indefinite forms. A likely factor would seem to be prosody; indefinite articles are weak syllables that appear in contexts making them vulnerable to omission whereas definite suffixes are weak syllables that are often attached to a strong syllable (as in stolen and tåget). To test this idea, Hansson, Nettelbladt, and Leonard (2003) also studied the Swedish children’s use of indefinite and definite forms when an adjective appeared before the noun. In such a context, both the indefinite (e.g., en) and definite (e.g., den) forms are monosyllabic weak-syllable articles and should be equally susceptible to omission. These investigators found that both types of articles were omitted more frequently by children with SLI than by younger typically developing children matched for MLU.
This is not the complete story of article use in Swedish SLI; article + adjective + noun combinations also involve a rather complicated system of interactions between gender, number, and definiteness. We shall discuss this more fully in a later section.
In Hebrew, there is a definite prefix, ha- that is used not only with the noun but also with any modifying adjectives, as in ha-mita ha-gdola ‘the-bed the-big.’ There is no marker for the indefinite. Rom and Leonard (1990) found more omissions by children with SLI than by MLU controls. However, Dromi et al. (1993) did not find group differences for this morpheme type. Thus far, the focus has been on nouns without accompanying adjectives. Thus, it is not known whether children with SLI have special difficulties marking the adjective as well as the noun with the definite prefix.
Studies of German-speaking children with SLI were some of the first to appear in the literature on SLI. Among them was Clahsen’s (1991) work showing that children with SLI learning this language often omit articles and, when they do produce them, they often have incorrect case. This issue was pursued in greater detail by Eisenbeiss, Bartke, and Clahsen (2005). In particular, they asked whether problems with case reflected difficulties with structural case (e.g., not grasping that nouns in direct object position require accusative case), or alternatively, represented problems with lexical case. The latter type of case is seen when a noun might appear in a position that suggests one type of case, but the particular verb actually requires a different case. For example, the verb danken ‘to thank’ requires dative case on the noun rather than accusative case. Eisenbeiss et al. found that a group of children with SLI were similar to younger MLU controls in using structural case, but had difficulties in selecting the proper case when the verb required a less typical (lexical) case for that structural position.
In Greek, articles are marked for case as well as gender and number. As in many other languages, they are weak monosyllables. Problems in this area are seen in Greek-speaking children with SLI (Tsimpli & Stavrakaki, 1999). For example, Tsimpli (2001) found that preschoolers with SLI omitted definite articles to a greater extent than younger typically developing children. Indefinite articles appear to be less problematic than definite articles, a finding that parallels the findings of Polite et al. (2011) for English. Chondrogianni, Marinis, Edwards, and Blom (2015) noted that Greek-speaking children with SLI show much higher levels of accuracy with definite articles as they approach seven years of age.
Articles clearly pose serious difficulties for children with SLI. There is good evidence indicating that prosody plays an important role. However, linguistic pragmatic factors and case are also implicated. The range of languages showing weaknesses in article use is noteworthy.
Possessive (genitive) inflections
Languages such as English and Swedish have a possessive (genitive) inflection that is attached to the noun serving as the possessor (e.g., mommy’s). As we saw in (1), when viewed from a linguistic perspective, this inflection appears in D, with the possessor occupying the specifier location. This inflection, then, contrasts with the noun plural inflection, as the latter forms part of the lexical category, whereas the possessive inflection is the result of a specifier–head relationship in a functional category. On morphosyntactic grounds, then, children might be expected to have more difficulty acquiring possessive inflections than noun plural inflections. One might argue that possessive ’s differs from noun plural -s in another respect as well. In most contexts, a phrase like Mommy car implies possession even without the inflection. This is true for fewer contexts when a noun plural is omitted (compare the car and the cars), though plural nouns are sometimes accompanied by quantifiers or number-sensitive determiners.
The evidence
Studies that have included possessive inflections are relatively sparse in the SLI literature. In a study of spontaneous speech in English, Leonard (1995) observed more frequent omission of the possessive ’s inflection by children with SLI than by a group of MLU controls. Swedish employs a possessive inflection not unlike that of English. For example, possessive mammas ‘Mommy’s’ is formed by attaching the -s inflection to the stem (mamma + s). Using a probe task, Leonard et al. (2001) found more omissions on the part of children with SLI than by younger children matched for MLU.
Possessive inflections are problematic for children with SLI in the two languages that have been studied. Unfortunately, these languages – English and Swedish – both employ word-final /z/ and/or /s/ as the possessive inflection. These phonemes cannot adequately explain the findings given that English-speaking children with SLI have less difficulty with noun plural inflections that employ the same phonemes. However, until a wider range of languages can be studied, the precise basis for difficulties with this morpheme type will not be clear.
Full pronouns
Along with their clear referential function, full pronouns in many languages are marked for case. Although pronouns are clearly noun-related, as they serve as a context-specific stand-in for nouns, they are also related to verbs. In many languages, the structural position of a pronoun relative to the verb (e.g., subject, direct object) will determine its case. For example, the direct object gets accusative case (or partitive case in some languages) from the verb. To understand how the subject gets nominative case, we need to consider another functional category, that of Subject Agreement (or AGRs), along with the functional category, Tense (T). (In some linguistic frameworks, AGRs and T form part of a broader functional category called Inflection or INFL.) The structure is illustrated in (7).
The subject can be thought of as originating in the Spec position of VP and then either ‘moving’ to the Spec position of TP to obtain tense and then on to the Spec position of AGRsP where the agreement features license nominative case. Alternatively, the process can be characterized as a feature being checked at each of these positions. In either instance, if a child’s utterance has the structure assumed in (7), a nominative case pronoun is expected in the subject position.
However, if some of the child’s verb productions are missing an obligatory agreement morpheme, as when a child says like in a context that requires likes, it is possible that the utterance has a structure that is lacking the phrasal projection AGRsP, as in (8).
If so, an utterance such as Her like quinoa and Her pushed Paul could occur. It can be seen, then, that from this linguistic framework, correct use of nominative case depends on more than the child knowing the meaning of she and that it is used as the subject of a verb. The use of nominative case might also depend on the presence or absence of agreement marked on the verb.
There are other factors that might influence a child’s use of nominative case. Rispoli (1994) proposed that the common phonetic core seen in some pronouns such h- for he, him, and his, may lead children to over-apply the core in instances that are inappropriate. For example, the presence of m- in accusative (me) and genitive (my) pronouns might lead children to use me or my in place of I. In contrast, pronoun forms that seem to have no common core, such as us, are quite unlikely to be used in place of another pronoun.
A third factor that could play a role comes from the ‘input ambiguity’ hypothesis put forth by Pelham (2011). She pointed out that certain pronouns, notably you and it, are case-ambiguous in the sense that they have the same phonetic form regardless of their grammatical role (e.g., You saw the boys and the boys saw you). Pelham hypothesized that if children hear such forms frequently, they may overgeneralize and treat other pronouns as equally applicable regardless of grammatical role, leading to productions such as Me saw the boys and the boys saw me. Examining parent–child language samples, Pelham observed that there was a correlation between the input frequency of case-ambiguous pronouns and children’s pronoun case errors.
A fourth factor is the pronoun form that serves as the default in the language. English differs from many languages in that accusative case is the default form. An answer to a question such as ‘Who wants ice cream?’ can be ‘Me’ in English. In many other languages, the nominative form would be the form used – the equivalent of ‘I.’ Thus, if an English-speaking child produces an accusative form in subject position, it might be because the child was simply selecting the default form (e.g., her) that matched the referent in other key respects (e.g., third person, feminine, singular).
The evidence
A longstanding observation in the SLI literature on English is the frequent substitution of accusative case for nominative case in pronouns, as in Her running, Him see that, Me wanna do that, and Them really pretty. Although such errors are also seen in young typically developing children, Loeb and Leonard (1991) found that such errors were more frequent in the speech of children with SLI than in the speech of younger MLU control children. Similar findings were reported by Wexler, Schütze, and Rice (1998).
Earlier, it was noted that, according to some linguistic accounts, nominative case is related to the functional category AGRs, and when pronoun errors such as her for she occur, the child’s sentence is probably lacking a morpheme (such as is or -s) that marks agreement with the subject. Indeed, in the Loeb and Leonard (1991) study, the children were also less likely than the MLU controls to use agreement morphemes, and in both groups, there was a correlation between the tendency to produce nominative case and the tendency to use subject agreement morphemes.
Wexler et al. (1998) examined this relationship in a very systematic way, and found that although nominative case pronouns could appear with (e.g., She is running) or without (e.g., She running) agreement morphemes, accusative case pronouns were mostly associated with the absence of agreement morphemes (e.g., Her running). (Recall that the absence of either T or AGRs could block the realization of is, as this morpheme requires both tense and agreement. In the case of She running, AGRs is presumably present, but not T.) However, there were occasional instances of unexpected utterances such as Her is running, though they were attributed to performance errors on the part of the children. Charest and Leonard (2004) essentially replicated the Wexler, Schütze, and Rice findings, and they, too, found occasional exceptions to the pattern, such as Her runs.
When AGRs is not reflected in the utterance, the use of an accusative case pronoun as a substitute makes sense for English because, in this language, accusative case is the default. Substitutions of this type are not expected in many other languages, as nominative case is the default. One curious exception is seen in the study by Bol and Kasparian (2009), who found that accusative case pronouns (e.g., hem ‘him’) were used in place of nominative pronouns (hij ‘he’) more frequently by children with SLI than by younger MLU control children in Dutch – even though nominative case is the default in this language. The basis of this substitution is not clear.
Problems with full pronouns have been most clearly documented for subject pronouns in English. In this language, subject pronouns must assume nominative case, which may depend in part on subject–verb agreement factors. The selection of nominative case also runs counter to the default form in English, case ambiguity seen in some pronouns (you, it), and the fact that some of the competing forms show a common core (e.g., me, my; her, her). If the accusative-for-nominative case substitutions seen for Dutch SLI are confirmed in future work, the default form explanation in particular would seem less plausible.
Noun suffixes in agglutinating languages
In some of the languages for which there are SLI data, case is expressed on a noun suffix. In some of these languages, the noun suffixes appear as part of an agglutinating morphology (e.g., Finnish). That is, a suffix marking case might be sequenced after a suffix marking plural. The sequence of suffixes is fixed in the language, but the number of combinations of suffixes can be fairly large. Given that the sequence is consistent (e.g., noun stem + plural suffix + accusative case suffix), the identification of each suffix might not be challenging in principle. However, morphophonological rules often apply, such that the phonemes of a suffix change according to the phonemes that immediately precede them, either in the stem, or in the suffix that lies closer to the stem. In other instances, the phonemes of the stem itself might change. These characteristics can complicate the identification and learning of each suffix and the grammatical function it serves.
In addition, the agglutinating languages studied to date in the SLI literature have a dominant pattern of stress on the first syllable. If the sequence of suffixes results in several weak syllables having to be produced in a row, there could be weak syllable deletion, with one of the suffixes being the casualty. In agglutinating languages in which multisyllabic words are the norm, young typically developing children only rarely omit syllables from three- and four-syllable words (see Saaristo-Helin, Kunnari, & Savinainen-Makkonen, 2011 for a review of Finnish). However, to the extent that weak syllable deletion is a factor in SLI, the additional length due to the attachment of suffixes may well be an obstacle.
The evidence
Leonard et al. (2014) studied the use of case suffixes by Finnish-speaking children with SLI and a group of younger controls with similar MLUs. These suffixes marked genitive case (attached to nouns in response to a question such as ‘Whose baseball cap is this?’), partitive case (used with nouns in response to a question such as ‘What is the man building?’), and accusative case (applied to nouns in response to a question such as ‘What does the woman catch?’). The children with SLI were less accurate than the younger control children across the three case suffixes but, like the typically developing children, they found accusative case suffixes the most difficult. The most frequent substitution was the use of a noun in nominative case. The nominative form can be regarded as a bare stem, but its phonological characteristics are not identical to the form that would result if a child were to simply omit the genitive, partitive, or accusative suffix. For example, the nominative form for ‘coat’ is takki, whereas omission of the accusative suffix from takin would yield taki, not takki. Yet, takki was the form produced.
The children with SLI were less accurate than the younger control children in one other respect. For some nouns and suffixes, the addition of the suffix requires a change in the phonological characteristics of the stem. Note in the above example, takki ‘coat’ becomes takin when the accusative suffix is added. Yet, many children with SLI added the suffix without making the stem change, resulting in takkin.
Hungarian, like Finnish, employs case suffixes. Earlier, it was noted that Lukács et al. (2010) studied children’s use of noun plural suffixes. In the same study, they also studied the children’s use of the accusative suffix in the singular, and when combined with the plural suffix. For example, the nominative case, kutya ‘dog’ becomes kutyát in accusative case. When both plural and accusative are obligated, these suffixes appear in sequence, as in kutyákat. Lukács and her colleagues found that the children with SLI were as accurate as the younger vocabulary-matched children when only accusative case was required, but were less accurate than the younger children when both the plural and the accusative were required. The most common error was omission of one of the two suffixes, though not both. The children with SLI also had difficulty handling the morphophonological details, as when the stem was not modified when the suffixes were added, leading to productions such as kutyakat instead of kutyákat.
Lukács, Kas, and Leonard (2013) also examined lexical case in Hungarian-speaking children with SLI. As noted earlier, unlike structural case that is based on the syntactic position of the noun, lexical case is dictated by the subcategorization rules of the particular verb as well as the semantics of the entire predicate. For each lexical case, these investigators selected two types of probe items. One type involved spatial meanings. For example, when the suffix -ből ‘from’ is used with a noun that follows a verb such as ‘escaped,’ a spatial meaning is clearly conveyed. The other type of item was non-spatial in nature, as when -ből is used with a noun that follows a verb such as ‘learned.’ In this instance, there is only a metaphorical rather than literal connection to a spatial meaning. Lukács and her colleagues found that the children with SLI were less proficient than younger control children in marking the proper lexical case, though both groups were more accurate when a spatial meaning was expressed than when a non-spatial meaning was involved.
Hebrew makes use of a definite accusative case marker, et, that appears between the verb and the direct object when the latter is definite, whether a common or a proper noun. An example is Ima mexabeket et Daniel ‘Mommy hugs Daniel.’ Both Rom and Leonard (1990) and Dromi et al. (1993) found that children with SLI were more likely to omit the accusative case marker than were typically developing children matched for MLU.
Considering the variety of noun suffixes that are used in agglutinating languages, relatively little is known about their status in SLI. However, from the two languages studied to date – Finnish and Hungarian – problems have been identified in this general area. First, children with SLI can be inaccurate in their choice of the appropriate case for the suffix. Second, when more than one suffix is required in a sequence, one of these suffixes is vulnerable to omission. Finally, the morphophonological changes required in suffixation are often difficult for these children.
Pronominal clitics
Many languages have both full pronouns and pronominal clitics that carry the same person, number, and case information. For example, in Italian, the direct object ‘her’ will be a full pronoun form (lei) in the context of No, vedo lei, non lui ‘No, I see her, not him’ but in clitic form (la) in the context of Dov’è Carol? Ah, adesso la vedo ‘Where’s Carol? Oh, now I see her.’ Despite the fact that pronominal clitics are much more frequent than full pronoun forms in many languages, clitics are often very challenging for young typically developing children (Wexler, 2003). There are probably both morphosyntactic and prosodic reasons for this difficulty. Consider the structure in (9), based on Wexler (2003). (In this framework, for null-subject languages such as Italian, checking for agreement at AGRsP is not required and is thus not shown below.) The sentence La bacia as in ‘(He) kisses her’ would have the literal translation as ‘(He) her kisses.’
Whether in terms of movement or feature checking, we can think of the direct object pronoun originating in the DP position of V’. The pronoun then moves to (or checks features at) the functional category Object Agreement (AGRo) where accusative case is assigned. It then moves to its preverb position, which is shown here as the Spec position of CliticP.
Clearly, if clitics involve the type of morphosyntactic operations just outlined, they would be challenging for young children. However, even without making detailed assumptions about movement or feature checking, one can see how clitics are difficult because they occupy a sentence position that differs from the usual direct object position. For example, as seen in the Italian sentence Bacia Gemma ‘(He) kisses Gemma,’ when a noun (or a full pronoun) serves as the direct object (Gemma, in this instance), it follows the verb. In many other languages, including English, nouns and their pronoun equivalents occupy the same sentence position (‘He kisses Gemma/her’).
Clitics also pose a prosodic challenge. In many languages, they are unstressed monosyllables. Just as the in the car is difficult because it represents a phrase-initial weak syllable, so too is the Italian clitic la in La bacia ‘(He) kisses her’). Furthermore, whereas there are many instances in which a weak syllable can be reorganized with a preceding strong syllable in a language such as English, languages such as Italian contain many multisyllabic words that end in a weak syllable. Thus, even if a subject is overtly produced, as in Franco la bacia, the clitic la immediately follows a weak syllable, not a strong syllable. Reorganization with the preceding word, then, would not lead to a strong syllable–weak syllable sequence, rendering the clitic vulnerable to omission even in this instance.
It can be seen, then, that pronominal clitics represent an interesting mix of elements. On the one hand, they serve as arguments of the verb. For example, without the direct object clitic la in the Italian sentence Franco la bacia ‘Franco kisses her,’ the sentence would seem incomplete. On the other hand, pronominal clitics are closely associated with functional categories, as shown in (9). As weak syllables, often in prosodically challenging sentence positions, they can also present production difficulties for children.
The evidence
In the SLI literature, pronominal clitics have been studied most extensively in Romance languages. Italian-speaking children with SLI are more likely to omit pronominal clitics than MLU-matched compatriots (e.g., Bortolini et al., 2006; Dispaldro, Leonard, & Deevy, 2013; Leonard & Dispaldro, 2013). In two different studies emphasizing diagnostic accuracy, Bortolini et al. (2002; Bortolini et al., 2006) found that clitics demonstrated good sensitivity and specificity in distinguishing children with SLI from their typically developing peers. Most often, clitics are tested in their common pre-finite-verb position, as in the Franco la bacia example. Leonard and Bortolini (1998) found that children with SLI are more likely than younger control children to omit clitics from this position than when the clitic attaches to the end of an infinitive (e.g., lo in Vuoi farlo? ‘Want to do it?’). In the latter context, children with SLI tend to include the clitic and not differ from MLU controls. This context often results in a strong syllable–weak syllable sequence (as in farlo), which would render successful production more achievable. On the other hand, many of the infinitive + clitic examples studied by Leonard and Bortolini were in relatively common expressions. For this reason, they may have been highly rehearsed through regular use.
In Spanish, pronominal clitics are also troublesome. In studies employing both probe tasks and spontaneous speech, Bedore and Leonard (2001, 2005) found that children with SLI were less successful than MLU controls in the use of clitics. Unlike the findings for Italian, Bedore and Leonard observed that clitic substitutions were as frequent as omissions. Most noticeable was the finding that plural clitics were often replaced by singular forms. Earlier, in our discussion of articles, we noted certain prosodic differences between Spanish and Italian; the same factor may have contributed to the crosslinguistic difference in the findings for clitics.
French-speaking children with SLI may also have problems with object clitics, though the evidence is somewhat mixed. Hamann et al. (2003) observed that preschool-age children with SLI were less proficient with object clitics in their spontaneous speech than typically developing children who were two years younger. Also using spontaneous speech samples, Thordardottir and Namazi (2007) reported very different findings. These investigators found that preschoolers with SLI used object clitics to the same degree as both same-age peers and younger children matched for MLU. All groups used object clitics infrequently in the speech samples, but instances of omissions were rare. The largest group differences are seen in a study by Jakubowicz, Nash, Rigaut, and Gérard (1998) that employed experimental tasks to elicit object clitics. In this study, younger typically developing children showed much higher percentages of use (79%) than the children with SLI (25%). Because spontaneous speech allows children to be selective in the types of structures they use, such samples may provide overestimations of the children’s true ability. In addition, Pérez-Leroux, Pirvulescu, and Roberge (2011) note that in certain contexts in French, speakers tend to prefer a clitic but may also use a lexical NP. Only by carefully controlling the specific sentence contexts might it be possible to gain a precise impression of children’s abilities with object clitics.
Clitics were among the grammatical morphemes examined in a study of Czech-speaking children with SLI by Smolík and Vávrů (2014). As in other languages, pronominal clitics serve as arguments of the verb; in Czech they occupy second position in the sentence and form phonological units with adjacent words. Using a sentence imitation task, Smolík and Vávrů found that the children with SLI were more likely to omit pronominal clitics than were a group of younger typically developing children matched on vocabulary test scores.
Pronominal clitics serving as direct objects seem to be an especially weak area for Greek-speaking children with SLI. In Greek, these clitics precede the finite verb. Stavrakaki and van der Lely (2010) found that school-age children with SLI were less accurate in both production and comprehension of object clitics than were three different groups of younger typically developing children. In production, the children with SLI had a tendency to either omit the clitic, or to provide a full noun phrase in a context where a clitic would ordinarily be used. For example, in response to ‘What does the elephant do to the dog?’ these children would sometimes respond with ‘(The elephant) kicks the dog’ rather than ‘(The elephant) him-kicks.’ Tsimpli (2001) reported significant omission of object clitics relative to a small group of younger typically developing children who had participated in a longitudinal study. Chondrogianni et al. (2015) found that children with SLI had difficulty with object clitics on an online comprehension task even when their production accuracy was high.
With the exception of one study of spontaneous speech in French, the evidence indicates that object clitics represent an area of serious concern in SLI. Prosody probably contributes to this weakness, as does the pre-finite-verb position of object clitics. It is not yet clear if the difficulty with the latter can be attributed to a morphosyntactic operation of movement/checking, or to the simple fact that object clitics occupy a different syntactic position than nouns that play the same object role. When object clitics are attempted, substitutions are sometimes seen.
Opaque and complex noun phrases
When discussing agglutinating languages, we noted how a noun suffix sequence can cause complications in part because of the morphophonological rules involved and in part because of the vulnerability of weak syllables in sequence with other weak syllables. Another kind of problem involving sequences is seen when a language requires agreement in a sequence such as article + adjective + noun. If a language marks agreement in a very transparent way, this is less of a challenge for children (Leonard et al., 1992). For example, the Italian sequence la bella ragazza ‘the beautiful girl’ clearly marks feminine singular on the article, adjective, and noun. However, in some languages, the agreement is much more opaque and inconsistent across the three elements.
In an earlier presentation of Swedish SLI data, we noted that articles in article + adjective + noun combinations are often omitted by children with SLI, which supported a prosodic explanation for some of the children’s difficulties. However, articles in particular and prosody more generally are not likely to be the only obstacles that such combinations present to children with SLI. These combinations are rather complicated not only because they involve distinctions of number, gender, and definiteness but also because some of these distinctions are only partial and quite opaque. For example, the gender distinction is made in both the article and the adjective for singular indefinite combinations, as in en stor stol ‘a big chair’ versus ett stort tåg ‘a big train.’ However, in the definite, the article and noun make the gender distinction, but the adjective does not, as in den stora stolen ‘the big chair’ versus det stora tåget ‘the big train.’ Recall as well, that when an adjective is not used in a definite phrase, no article is used either, as in tåget ‘the train’ versus det stora tåget ‘the big train.’ Swedish article + adjective + noun sequences are of this type, as we shall see. Discovering the rules of agreement for such sequences might be quite problematic for children with SLI.
The evidence
Using a probe task, Leonard et al. (2001) found that children with SLI were significantly less accurate on article + adjective + noun sequences than typically developing children matched for MLU. Omissions of the article were the most frequent error, but errors in gender and definiteness were also seen, including instances in which indefinite and definite forms were used within the same combination (e.g., ett stort tåget instead of det stora tåget). Errors in gender or definiteness of the adjective also co-occurred with article omissions. Although the omission of articles might be partially attributable to prosodic factors, Leonard et al. found that the children with SLI in particular were much more likely to omit the article from an article + adjective + noun combination (e.g., en stor stol) than from a simple article + noun combination (e.g., en stol), even when the particular article needed in the two contexts was the same and in both instances the article was a weak syllable preceding a strong syllable. This suggests that the complexity of the entire combination probably played a role in the children’s difficulty.
Although article + adjective + noun sequences necessarily introduce features such as gender, number, and definiteness into the picture, and probably render some element vulnerable to omission for prosodic reasons, it seems likely that the problem is exacerbated when, as in Swedish, the morphemic forms to be selected for the sequence are not uniform or transparent.
Evidence based on composite measures of noun-related morphemes
In several studies, investigators have used a composite measure made up of several different noun-related morphemes. Bliss (1989) found that English-speaking children with SLI showed lower scores than younger controls matched on MLU on a composite consisting of noun plural -s, articles, and pronouns. In their study on English, Bedore and Leonard (1998) used a composite made up of noun plural -s, possessive ’s, and articles, and assessed the diagnostic accuracy of this measure. In two experiments, specificity was quite good, but in one of the experiments, sensitivity was not sufficiently high (21% of the children with SLI were misclassified as typically developing). In both experiments, a composite measure composed of verb-related morphemes involving tense and agreement showed better diagnostic accuracy.
The opposite pattern was found for Icelandic, although the research design was somewhat different. Elin Thordardottir (2008) compared SLI and age-matched groups and found that group differences were larger for a noun-related composite than for a verb-related composite. This finding is much less surprising considering that, in Icelandic, nouns are marked for case as well as number, and adjectives are marked for case, number, and gender.
Other investigators have employed composites containing key noun-related elements but also other types of morphemes. In a study of English, Leonard et al. (1997) created a composite consisting of noun plural -s, possessive ’s, articles, and infinitival to. Children with SLI earned lower composite scores than younger MLU controls. Eadie et al. (2002) also reported lower scores by English-speaking children with SLI relative to MLU controls using a composite made up of plural -s, possessive ’s, articles, and progressive -ing. Note that in both the Leonard et al. and Eadie et al. studies, the verb-related morphemes contained in the composite (infinitival to, progressive -ing) did not involve tense or agreement.
In a study of Dutch, Zwitserlood, van Weerdenburg, Verhoeven, and Wijnen (2015) used a ‘non-verb-related’ composite that included determiners and pronouns, though morphemes unrelated to nouns were also involved. The children with SLI were less accurate than a group of younger typically developing children matched according to a language test score. Gutiérrez-Clellan, Restrepo, and Simón-Cereijido (2006) used articles and clitics as well as elements unrelated to nouns in a composite to study diagnostic accuracy in Spanish. The measure showed fair to good diagnostic accuracy in distinguishing children with SLI from age-matched peers.
It is probably not surprising that composite measures of noun-related morphemes have revealed difficulties on the part of children with SLI. Most or all of these morphemes have been tested singly in some study in the same language. These composites have not always demonstrated high levels of specificity, but are unanimous in showing statistically significant group differences favoring young typically developing children.
The sources of difficulty with noun-related morphemes
In this review, we have identified several types of noun-related morphemes that are quite difficult for children with SLI. Although verb-related morphemes have received more research attention, it is clear that problems with noun-related morphemes are far from trivial for this clinical population. They represent more than below-age-level abilities; recall that these weaknesses emerged in comparisons with younger typically developing children – children who were usually two or three years younger than the children with SLI. An inspection of the morphemes that are troublesome offers some hints as to possible sources of difficulty. Here, we consider some of the possibilities.
Prosody is a factor leading to omissions, and these omissions, in turn, may mask additional noun-related weaknesses
The frequent omissions of weak-syllable morphemes such as articles and clitics certainly implicate prosody. However, languages that put these morpheme types in a more favorable prosodic position, with fewer omissions, reveal substitutions as well. This raises the possibility that if, say, Italian-speaking children with SLI had less difficulty with prosody, they might show approximately the same distribution of substitutions and omissions seen in children with SLI who speak Spanish – a language whose weak syllables have longer durations than Italian.
Difficulties with weak syllables may also mask other types of problems. For example, when two suffixes appear in sequence in an agglutinating language, one of the suffixes is often omitted. Because both morphemes are weak syllables, a prosodic explanation seems plausible. However, it is possible that planning the expression of two different grammatical functions (one for each suffix) could also be a problem, such that one of the suffixes might be left out even if prosody were not a factor.
We also saw how articles were often omitted in Swedish article + adjective + noun combinations at a rate that was even higher than article omissions in article + noun combinations, even though the latter context also placed the article in the more vulnerable weak syllable–strong syllable sequence. This suggests that there was something in the complexity of the longer combination that might have prompted omission over and beyond the prosody itself.
Noun-related morphemes closely tied to verb-related functional categories seem to be at considerable risk
It is noteworthy that both subject pronouns in English and pronominal object clitics in several different languages are obligatory arguments in the sentence, yet they are difficult for children with SLI. From a strictly semantic or argument structure point of view, they should have priority in the child’s selection of words to include in the utterance. Yet both of these morpheme types are problematic. Subject pronouns seems to be closely associated with a functional category related to subject–verb agreement – an area of notorious difficulty for children with SLI. Pronominal clitics, too, seem related to tense and agreement in that, unlike full noun phrases that would play the same argument role, clitics must move to a preverb position when the verb expresses tense and agreement – reflecting a functional category in some linguistic frameworks.
Case is a grammatical function that presents significant difficulty
Whether case is determined by the features of a particular verb or by structural position or grammatical function, morphemes marking case are often used less accurately by children with SLI than by their younger typically developing peers. This is true for case expressed on articles, noun suffixes, or free-standing forms such as et in Hebrew. Because so much research on SLI has been conducted on English-speaking children and this language is so limited in the morphemes with overt case, it is perhaps not surprising that case has received very little attention among SLI researchers. Difficulties with subject pronouns have certainly been well documented, but little else would suggest difficulties with case given the default status of accusative forms in English and the use of the same pronoun forms for direct objects, indirect objects, and objects of the preposition, as in Let’s watch him, Give him the money, and Put it on him, respectively. In other languages, problems with case become clearer.
Linguistic pragmatic limitations may exacerbate problems with noun-related morphemes
When prosodic factors are controlled, definite articles appear to be more troublesome than indefinite articles, especially in tasks that require the child to select the appropriate article form without the benefit of a model (as in Polite et al., 2011). This was seen in Greek as well as in English. (The morpheme et in Hebrew is also used only when the referent is definite, though it also marks case.) These findings suggest that children with SLI may have difficulties recognizing those nouns whose referents are identifiable by the listener.
Although difficulties with pronominal clitics can be easily associated with problems with weak syllables and/or verb-related functional categories, the choice of a clitic (e.g., la ‘her’ in Italian) over a full noun phrase (la ragazza ‘the girl’) or a full pronoun (lei ‘her’) also requires a grasp of the kind of information that is available to the listener. Most tasks have not been designed to tap this aspect of clitic use, but an over-use of using a full noun phrase in contexts more appropriate for a pronominal clitic has been documented in Greek SLI.
Noun plurals do not represent an area of special difficulty
It is perhaps ironic that of all the noun-related morphemes studied to date, the one that seems to pose the least difficulty is the one that is most closely tied to the lexical category, Noun, namely, noun plural inflections. This morpheme type is not part of any functional category. This is not to say that children with SLI are at age level in their use of plural inflections; they are not. For example, some studies showing similar use of plurals by children with SLI and MLU controls nevertheless find greater use by age controls (e.g., Dromi et al., 1993; Polite, 2011). Abdalla, Aljenaie, and Mahfoudi (2013) did not employ a group of younger controls, but found very large differences between Arabic-speaking children with SLI and their same-age peers. However, in most studies in most languages, differences between children with SLI and younger typically developing children are not often found, even in languages where the inflection is consonantal.
Multiple morphemes involving grammatical or phonological opacity are challenging
Thus far, we have focused on individual morpheme types. However, it appears likely that problems also exist when children with SLI must learn the pattern involved in combinations of morphemes. In some instances, the combinations are difficult because the grammatical features reflected in the morphemes are not transparent. As seen in Swedish article + adjective + noun combinations, for example, the definiteness, number, and gender features are not uniform across the morphemes in some of the combinations. Children with SLI seem to have more difficulty in these instances than younger typically developing children.
In other instances, morphophonological rules that alter the form of a noun stem or a suffix can serve as a source of confusion. That is, even when children with SLI can learn and apply a suffix to a stem, they may not have learned the phonological changes that must accompany this application. Adding a suffix while recognizing other changes that must be made seems to be a burden for these children.
Implications for assessment
Perhaps an illuminating way to discuss appropriate targets for assessment is to contrast the noun-related morphemes that might be the focus in assessment protocols of English, Spanish, and Swedish. We select these languages because their noun-related morphemes serve many of the same grammatical functions, but they express these morphemes in quite different ways – ways that can lead to crosslinguistic differences in the degree to which they cause difficulty for children with SLI.
We begin with the qualifier that children with SLI will be below age-level on many details of language; here we single out those for which there is evidence of extraordinary difficulty in at least one of these three languages. Only English-speaking children are likely to have special problems with subject pronouns, given the default status of accusative rather than nominative case in that language. However when direct object pronouns are required in typical speaking contexts, it will be the Spanish-speaking children who will be at risk, given that pronominal clitics are the usual forms expected in these situations. When nouns are used as possessors (requiring a possessive inflection), Swedish-speaking children as well as English-speaking children may have difficulty.
Articles will be problematic for children with SLI in all three languages, though in different ways. In English, both indefinite and definite articles will often be omitted, but definite articles may be especially difficult given their linguistic pragmatic function. For Spanish, articles will also be difficult, but a higher proportion of substitutions will be seen. Omissions might be reduced given the prosodic characteristics of Spanish. In Swedish, indefinite articles will be subject to omission, but definite forms will benefit from the fact that they are syllabic suffixes that are attached to the ends of nouns rather than preceding them.
Swedish will stand out as especially challenging for children with SLI when article + adjective + noun combinations are required. In English, of course, only definiteness is distinguished (e.g., a/the pretty girl). In Spanish, number and gender along with definiteness is involved, but very often the agreement is transparent (e.g., la muchacha bonita). In Swedish, the same three dimensions are involved but the agreement is far less transparent. The gender distinction is lost on the adjective when changing from the indefinite to the definite, and an article is needed in the definite only when there is an adjective. Finally, plural inflections are not likely to be a major concern in any of the languages. If problems are seen, they are more likely to be seen in English- and Spanish-speaking children with SLI.
Next steps
Clearly additional noun-related morphemes and additional languages should be the focus of future research. Once sufficient data are obtained, meta-analyses of noun-related morphemes should be possible, probably using language typology as a moderator variable. In terms of moving toward explanatory adequacy, more work is needed to determine the degree to which the difficulty with certain noun-related morphemes can be attributed to their functional category status and/or to their relationship to verb-related phenomena. In the meantime, it should be clear that pockets of serious morphosyntactic difficulty in SLI exist beyond morphemes that directly mark tense and agreement. When viewed across languages, these noun-related weaknesses cover a wide range of morpheme types that can adversely affect direct objects as well as subjects, and disrupt semantic, pragmatic, phonological, as well as grammatical expression. Until we have a better picture of the details of these noun-related difficulties, our understanding of SLI will surely be incomplete.
Footnotes
Declaration of conflicting interests
The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
Funding
The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: Portions of this work were supported by Grant R01 DC00458 from the National Institute on Deafness and Other Communication Disorders, National Institutes of Health.