Case Marking in Turkish Heritage Children With and Without Developmental Language Disorder

Abstract

Recent studies on agglutinative languages, such as Japanese, Finnish, and Turkish, have reported case marking deficits in children with developmental language disorder.In this study, we investigate case marking in bilingual children speaking Turkish as a heritage language in Germany in comparison to those in France and the U.S. and late successive bilinguals in Germany. The research focuses on the potential use of case marking to identify developmental language disorder in Turkish as a heritage language. In this study, we compare data obtained from 73 children with and without developmental language disorder (age 5;1–11;6) that speak Turkish as a heritage language to those obtained from 10 late successive bilinguals (age range 7;12–12;2) in Germany, France, and the U.S., analyzing case marking and the possessive markers included in genitive–possessive constructions and using both standard and heritage Turkish as reference varieties. The results show that the groups differ significantly (p < 0.05) regarding the use of case and possessive markers. Current first language use is the leading predictor of performance in case marker production in the TEDİL when using heritage Turkish as the reference variety in scoring. The results demonstrate that children with developmental language disorder that speak heritage Turkish produce fewer case markers and show higher rates of omission and substitution errors, particularly in accusative/dative and genitive markers, thus confirming the results of previous research. The omission of possessive and genitive markers in simpler structures may serve as a clinical marker of developmental language disorder, allowing for children with typical language development that speak heritage Turkish to be distinguished from those with a developmental language disorder.

1. Introduction

The CATALISE consortium (Bishop et al., 2016, 2017) recommends using the term developmental language disorder (DLD) to refer to children’s language deficits that create communication or learning barriers and are unlikely to resolve by the age of five. These language difficulties are not associated with a known biomedical condition, such as brain injury, neurodegenerative conditions, genetic conditions, or chromosome disorders (Bishop et al., 2016, 2017). DLD affects the comprehension and production of language in children, with morphosyntax being the most notoriously affected area (Leonard, 2014). According to estimates of epidemiological studies, 5% to 7% of school-age children are affected by DLD, an impairment of unknown origin specific to language. DLD is considered a public health concern (Law et al., 2013), as it is associated with an increased risk of school failure and long-term academic underachievement (Stothard et al., 1998; Tomblin, 2008; Durkin et al., 2015), poor employment outcomes (Johnson et al., 2010), and socio-emotional and behavioral problems (Yew & O’Kearney, 2013).

1.1. DLD in Monolinguals

Grammatical morphology is widely documented as an area of extraordinary difficulty for children with DLD. The typology of the language being acquired is assumed to contribute to the severity of the problems encountered with morphology (Ruigendijk, 2015). Verbal inflectional morphology has been identified as a vulnerable domain in children with DLD when acquiring different languages, e.g., English (Bishop, 1994; Rice & Wexler, 1996, 2001; Bedore & Leonard, 1998; Chondrogianni & Marinis, 2012), German (Clahsen et al., 1997; Roberts & Leonard, 1997; Rothweiler et al., 2012), and Swedish (Hansson et al., 2000). Studies on inflectionally rich languages, such as Italian, Spanish, Hebrew, and Hungarian, have shown that verb morphology appears to be less severely affected than nominal morphology in such languages (Leonard & Dromi, 1994; Bortolini et al., 1997; Bedore & Leonard, 2001; Leonard et al., 2002; Lukács et al., 2009). In languages with clitic pronouns, e.g., French, selective deficits with (object) clitics are considered a clinical marker of DLD (Cronel-Ohayon, 2004; Hamann et al., 2003; Jakubowicz, 2003; Jakubowicz & Tuller, 2008; Tuller et al., 2011).

In addition to verb morphology, several recent studies have identified nominal morphology as a locus of difficulty for children with DLD in different languages. For example, it has been shown that genitive inflections, indefinite articles, and “article + adjective + noun constructions” are difficult for Swedish-speaking children with DLD (Leonard et al., 2001). Spanish-speaking children with DLD have been reported to have deficits at the noun phrase (NP) level, especially with articles and clitics (Bedore & Leonard, 2005; Castilla-Earls et al., 2020). Further studies have shown that Italian-speaking children with DLD experience difficulties with both inflectional verb morphology, e.g., third-person plural inflections, and noun morphology, such as articles and clitics (Bortolini et al., 1997, 2006). Studies on case marking in monolingual children with DLD have obtained results for different languages. Monolingual English-speaking children with DLD have been shown to overapply accusative pronouns for nominative forms in subject positions, as in *“me drink at all” (Loeb & Leonard, 1991; Moore, 1995; Wexler et al., 1998; Radford & Ramos, 2001; Lukács et al., 2013). In their study, Rom and Leonard (1990) compared the spontaneous speech of a group of Hebrew-speaking children with DLD (age 4;4 to 5;3) to that of a group of younger typically developing (TD) children (age 2;4 to 3;3) matched according to the number of morphemes per utterance (MPU). They found that the definite accusative marker “et” in Hebrew was produced less by children with DLD than by younger TD children. Dromi et al. (1993) studied spontaneous speech samples of 15 monolingual Hebrew-speaking children with DLD aged 4;1 to 5;11, 15 TD children matched for age, and 15 TD children matched for the mean length of utterance in words (MLUw). The percentage of definite accusative markers used in obligatory contexts was lower for monolingual Hebrew-speaking children than for their peers matched for age and MLUw. Mastropavlou and Marinis (2002) examined speech samples of three monolingual Greek-speaking children with DLD, ranging in age from 6;7 to 8;2, and naturalistic data of four TD children aged from 1;7 to 2;9. The study showed that case marking on definite articles—especially in the genitive case—seemed more problematic in the speech of the children with DLD than gender and number marking, which results from a syntactic deficit (Mastropavlou & Marinis, 2002). Clahsen (1991) examined spontaneous speech samples of monolingual German-speaking children with DLD, ranging in age from 3;2 to 9;6, and in MLU in words from 1.46 to 2.84. It was observed that they incorrectly marked the case of articles by opting for nominative case marking in accusative or dative contexts. The German-speaking children with DLD were prone not only to subject–verb agreement errors but also to case marking errors in accusative and dative contexts; thus, Clahsen (1991) proposed a broader agreement to account for this. In contrast to these results, Eisenbeiss et al. (2005) investigated spontaneous speech data of monolingual children with DLD (age range from 5;8 to 7;11) and with TD (age range from 2;6 and 3;6) and showed that a distinction was made between structural and semantic cases, upon which the original agreement deficit account of DLD was modified. Eisenbeiss et al. (2005) concluded that German-speaking children with DLD encountered problems with semantic case marking rather than structural case marking. In their experimental study on motion verbs, Papadopoulou et al. (2009) asked monolingual Greek- and German-speaking children with DLD (age range from 5;1 to 6;4) and TD (age range from 5;0 to 6;0), as well as adults (range in age from 19 to 33), to describe what they saw in short videos. The monolingual German-speaking children with DLD “appear to experience problems with marking the dative case in the atelic condition, which is the last acquired case marking in unimpaired children” (Papadopoulou et al., 2009). Scherger (2019) investigated German dative case marking by comparing longitudinal studies of three simultaneous (bilingual language acquisition or 2L1) bilingual children with DLD aged 4 and 7 years to aged-matched cross-sectional data of monolingual TD children, 2L1 TD children, and sequential (L2) TD children, as well as monolingual children with DLD and L2 children with DLD. The results of this study showed “no differences between 2L1 and DLD by age 4. However, by age 7, there are differences between 2L1 and SLI on the one hand, but no differences between L2 and SLI on the other” (Scherger, 2019).

1.2. DLD in Agglutinative Languages

Previous studies on agglutinative languages, such as Japanese and Finnish, have reported case marking deficits in children with DLD. Fukuda and Fukuda (1999) assessed monolingual Japanese-speaking children with DLD (age range between 8;9 and 12;1) and eight age-matched children with TD using a battery of linguistic tests. The nominative, accusative, and dative cases on a grammaticality judgment task were reported to be problematic for Japanese children with DLD. Similar results were presented by Fukuda et al. (2007) concerning grammatical cases, namely, nominative, accusative, and dative. In their study, they investigated the grammatical knowledge of three Japanese children aged 9;7–13;3 with DLD by using elicited tasks. It was observed that the performance of children with DLD was poorer than that of children with TD. Murao et al. (2012) investigated structural and inherent cases in the spontaneous speech samples of two Japanese children with DLD who were 9 and 10 years old. The results showed that the number of errors was significantly higher in structural cases than in inherent cases in the samples of both children (Murao et al., 2012). To determine whether Finnish-speaking children with DLD (N = 15, mean age 5;2) were weaker in grammatical case use (i.e., accusative, partitive, and genitive case suffixes) than younger (N = 15, mean age 3;8) or same-age TD children (N = 15, mean age 5;2), their use of case suffixes was assessed through specifically designed probe tasks (Leonard et al., 2014). The Finnish-speaking children with DLD were less accurate in their case marking, especially in contexts requiring the accusative case, and they were more reliant on using the nominative form, a more frequent and less complex form in Finnish, than their TD peers (Leonard et al., 2014). In their studies, Lukács et al. (2010) examined the production of Hungarian irregular and regular noun morphology in the elicited production of nouns with plural, accusative case, and plural plus accusative case suffixes in two groups of children (8–10 years and 4–7 years) with language impairment and two control groups matched for vocabulary size. It was stated that the younger Hungarian-speaking children with language impairment were less accurate than the younger verbal control children in marking plural and accusative cases. The same study also showed a significant overgeneralization of stem forms with the correct selection of suffixes in all groups—the older and younger children with language impairment and the older and younger verbal control children.

1.3. DLD and Heritage Language

In the growing body of research focusing on case marking as a potential clinical marker of DLD in bilingual children, it was observed that the normal course of case acquisition is delayed in bilingual children with DLD, and these children performed below norms, even if the norms of the assessment tools were adjusted for bilinguals (see Rothweiler et al., 2010; Chilla & Şan, 2017; Şan, 2018; Gökgöz et al., 2020; Chilla, 2022, for Turkish–German; De Jong et al., 2010, for Turkish–Dutch; Blom et al., 2022; Scherger, 2019, for Italian–German; Werthmann, 2020 for German–Russian; and Schwartz et al., 2015, for Russian as a first language (L1) in bilinguals with different second languages (L2s)).

Most of the research on the intersection of bilingualism and DLD has focused on the acquisition of the majority language (L2) in simultaneous and successive child bilinguals (Blom et al., 2019). In contrast, only a few studies have examined DLD markers in the heritage language (HL). Heritage language refers to a language that is acquired at home and different from the majority language (Rothman, 2009; Lohndal et al., 2019). The language production and comprehension of heritage speakers may differ from those of monolingual control groups (Polinsky, 2006, 2008; Montrul, 2008). Whether this difference follows from incomplete acquisition or attrition in the linguistic knowledge of heritage speakers is still being discussed (Köpke, 2007; Montrul, 2008; Laskowski, 2009; Rothman, 2009). Language attrition refers to the phenomena in which certain structures are acquired and later undergo changes due to a lack of use or non-use and/or interference from the dominant language (Schmid & Yılmaz, 2018), whereas incomplete acquisition refers to the case where heritage speakers fail to learn certain structures due to their limited input and use in the HL (Montrul, 2008; Polinsky, 2011). The term “incomplete acquisition” is found to be problematic by heritage speakers and policymakers due to its adverse connotations and scientific backdrop (Putnam & Sánchez, 2013; Rothman & Kupisch, 2018; Bayram et al., 2017). According to Bayram et al. (2017), heritage grammars cannot be incomplete because they are “coherent grammars on their own” (Polinsky, 2008) and have structures that have been reanalyzed in terms of input quality and quantity. Łyskawa and Nagy (2020) state that neither incomplete acquisition nor attrition is the reason for the variation in heritage languages. Therefore, HLs are discussed as varieties on their own (Pires, 2011; Flores, 2015; Nagy, 2016).

Bilingual language acquisition (2L1) and early second language acquisition (eL2) are defined in accordance with the cut-off between L1 and L2 acquisition, which is at around 4 years of age (Meisel, 2007, 2009; Rothweiler, 2006). Considering that the disadvantages of late phenomena result from reduced input as in 2L1s, it may be assumed that the acquisition of the late phenomena of HLs may benefit more from a higher AoO of the L2 than early phenomena (Makrodimitris & Schulz, 2021). In terms of heritage language acquisition, the AoO of the L2 is identified as an important factor for the L1 skills of heritage children (Montrul, 2008). Because successive bilingual children acquire some phenomena in L1 and master them before their L2 exposure, their abilities in the HL are more advanced than those of simultaneous bilingual children with an early AoO of the L2 (Rothman & Kupisch, 2018; Makrodimitris & Schulz, 2021).

1.4. Diagnostic Challenges in Assessing Bilingual Children

Due to variations in their linguistic environments and divided language input, which mean that bilingual children are exposed to separate and unequal amounts of input in both of their languages, bilingual children differ from their monolingual age-matched peers in several aspects. Recent studies have focused on disentangling typical language development (TD) from DLD in bilingual children (Marinis & Armon-Lotem, 2015; Marinis et al., 2017; Tuller et al., 2018). Linguistic research in this vein has pointed out the overlap between the morphosyntactic error patterns of bilinguals with TD and those of monolingual children with DLD (MoDLD) (Armon-Lotem et al., 2015; Paradis, 2010; Paradis et al., 2016). This leads to methodological and clinical confounds, which may result in the misdiagnosis of bilingual children with TD (BiTD) as bilingual children with DLD (BiDLD). In order to avoid this diagnostic dilemma, it is recommended to assess both languages of the child or at least the dominant language (ASHA, 2004; IALP, 2006; Fredman, 2006; RCSLT, 2007). International research has discussed clinical markers, such as non-word repetition, working memory deficits, and sentence repetition (Bortolini et al., 2006; Ziethe et al., 2013; Chiat, 2015; Blom & Boerma, 2017; Abed Ibrahim & Hamann, 2016; Hamann & Abed Ibrahim, 2017).

Studies on case marking in monolingual and bilingual Turkish-speaking children with DLD found that noun suffix use is more vulnerable than verb morphology and that difficulties arise in case marking (Çavuş, 2009; De Jong et al., 2010; Rothweiler et al., 2010; Acarlar & Johnston, 2011; Topbaş et al., 2016; Chilla & Şan, 2017; Güven & Leonard, 2020; Gökgöz et al., 2020; Chilla, 2022). Moreover, the acquisition of Turkish in a heritage setting has been the subject of several studies in an attempt to characterize grammatical deviations (e.g., Yağmur, 1997; Türker, 2000; Akıncı & Decool-Mercier, 2010; Backus, 2013; Schroeder, 2016; Chilla & Şan, 2017; Şan, 2018; Felser & Arslan, 2019; Krause & Roberts, 2020; Schmid & Karayayla, 2020; Tekin, 2020; Chilla, 2022). Although Turkish exists as an HL in countries such as Germany, France, and the U.S., its sociolinguistic profile differs among these countries, which may result in differences in the grammatical structures of its varieties (Özsoy et al., 2022). The difference in the sociolinguistic profile results from the way of migration, the population of the Turkish-speaking community, the domains in which Turkish is spoken, the geographical distance of the countries in which people of Turkish origin live, the literacy in Turkish, and the access to Turkish media. In light of diagnostic challenges due to the observed overlap between DLD markers in the monolingual acquisition of Turkish and grammatical features of heritage Turkish, this study investigates whether case marking can be used to identify DLD in the bilingual German–Turkish context as opposed to in the French and U.S. contexts, as well as how Turkish-speaking heritage children with DLD compare to their typically developing peers in their use of case marking.

2. Case in Standard Turkish

2.1. Case Marking in Turkish

As an agglutinative language, Turkish has a rich morphological system. Morphosyntactic relations in Turkish are expressed by the suffixation of several different morphemes on verbal and nominal stems. Unlike fusional languages, each suffix encodes a single (semantic) dimension (e.g., plural or dative case, but not both) in Turkish and can vary in form according to the rules of vowel harmony (Göksel & Kerslake, 2005). Suffixes conveying different information are consecutively attached to the noun stem in a fixed order (e.g., the case suffix always follows the suffix used for number). The default word order in Turkish is subject–object–verb (SOV), and it can alternate for pragmatic purposes, such as signaling topic focus (Erguvanlı, 1984). Given the flexible word order in Turkish, basic grammatical relations are conveyed with inflectional markings on verbs and nouns. In Turkish, the structural arguments of the verb are expressed through case marking on nouns rather than the word order (Çavuș, 2009). Structural cases in Turkish include the nominative (zero marking), accusative, and dative. For example, the accusative case suffix is usually attached to the noun as a direct object (DO) of a transitive construction. The remaining case suffixes (genitive, locative, ablative, and semantic dative referring to directional goals and instrumental) mark the semantic case. Turkish has seven cases, all of which are marked on nouns, question words, pronouns, and the nominalized forms of verbs or adjectives by an overt case suffix, except for the nominative case, which is not overtly marked on the noun and has no phonological realization (see

Table 1. Case paradigm of standard Turkish (e.g., ev “house”).

Case	Singular	Plural
Nominative	ev	ev-ler
Accusative	ev-i	ev-ler-i
Dative	ev-e	ev-ler-e
Locative	ev-de	ev-ler-de
Ablative	ev-den	ev-ler-den
Genitive	ev-in	ev-ler-in
Instrumental	ev ile (ev-le)	ev-ler ile
		(ev-ler-le)

).

A fixed order exists in the combination of case suffixes with number and possessive markers, such as stem + number + possessive + case. Nouns without a case suffix are typically subjects and, thus, nominative. The unmarked nominative can be signaled using agreement markers on the verbs, excluding third-person singulars, which do not have an overt agreement marker on the verbs. It is challenging to differentiate between the nominative case and the caseless bare form, which is used as a non-specific direct object (gazete oku- “newspaper-reading”) or a subject (su bas- “flood-, literally, water-cover”), especially in one-word utterances (F. N. Ketrez & Aksu-Koç, 2009). In addition, the emergence of the nominative case cannot be traced due to it not being a phonologically realized morpheme. The accusative case marks the direct object of a transitive verb. The accusative case is obligatory, where the direct object is definite or indefinite. As previously stated, some direct objects are left in a non-case-marked form, which results in difficulties in differentiating between an omission error or referring to an object in a non-specific or generic sense in the utterances of children (F. N. Ketrez & Aksu-Koç, 2009). The accusative case marker is obligatory in pronouns and nouns with a possessive morphology. The following examples show the accusative suffix with a noun (1a) and the accusative suffix with a pronoun (1b):

(1)	a.	(Ben)	kalem-i	al-dı-m.
		(I)	pencil_ACC	take-PAST-1SG
		“(I) took the pencil.”
	b.	(Ben)	o-nu	al-dı-m.
		(I)	It-ACC	take-PAST-1SG
		“(I) took it.”

The dative case expresses both structural and semantic cases. It marks the indirect objects of intransitive verbs, such as ver- “give” in (2a), and the direction or the goal of an action as in (2b). Additionally, the performer of an action in a causative construction, in which a transitive verb (e.g., let/make somebody do something) is causativized, can be expressed by a noun phrase with a dative case marker. Dative-marked complements are obligatorily taken by some verbs, such as -y)A inan- “believe” and -(y)A güven “trust”. The dative case suffix is also required for some adjectives, such as -(y)A așık “in love with”, and post-positions, such as -(ya)A kadar “until”. Compared with other case markers, the use of dative case suffixes is less transparent and predictable; however, their use in different obligatory contexts increases their frequency.

(2)	a.	Kız	oğlan-a	çiçek	ver-di
		girl	boy-DAT	flower	give-PAST-3SG
		“The girl gave a flower to the boy.”
	b.	Kız	okul-a	git-ti.
		girl	school-DAT	go-PAST-3SG
		“The girl went to school.”

The locative case expresses static location, and the locative case suffix is typically used in adverbial constructions specifying where the action takes place, as in (3):

(3)	Ahmet	köy-de	yaş-ar.
	Ahmet	village-LOC	Iive.PRES
	“Ahmet lives in the village.”

The ablative case marks notions, such as departure, separation, source, or cause, as in (4). Despite being more transparent than the other case markers, the ablative case suffix can be used in unpredictable relations with a few verbs, such as –DAn hoșlan “like”, -DAn vazgeç “give up”, and –DAn faydalan “benefit”.

(4)	Adam	ev-i-nden	çık-ıyor-du.
	the man	home-POSS.3SG-ABL	come out-PROG-PAST-3SG
	“The man was coming out of his home.”

The genitive case marker marks the “possessor” and often co-occurs with the possessive marker in a genitive–possessive construction, as in (5). The latter is a composite noun phrase (NP) constructed from two nouns. The possessive marker is attached to the object “possessed” (noun) and has to agree with the possessor in number and person; however, the possessive marker denoting the “possessed” does not count as a case suffix.

(5)	Kadın-ın	boy-u	kısa.
	women-GEN.3SG	height-POSS.3SG	short
	“The woman is short.”

The instrumental case marker ile “with” is discussed as a marginal case because this marker can have comitative, instrumental, and conjunctive meanings (F. N. Ketrez & Aksu-Koç, 2009; Göksel & Kerslake, 2005). Commutative and instrumental markers form postpositional phrases. They can also be attached to noun phrases, which enables them to function as oblique objects, as seen in the dative, locative, and ablative cases. The suffixal form of the commutative/instrumental case marker ile is –(y)lA, as shown in 6. As –(y)lA can be attached to bare pronouns, it can be attached to genitive-attracting pronouns, as shown in (7). Furthermore, the suffix form of the instrumental case marker is a clitic, and it does not allow stress.

(6)	Çocuk	oyuncak-la	oynuyor.
	Child	toy-INST	play-PROG-3SG
	“The child is playing with the toy.”

(7)	Simple personal pronouns: seninle “with you”
	Demonstrative pronouns: bununla “with this”
	kim “who”: kiminle “with whom”

2.2. The Acquisition of Case Marking in Turkish

The theoretical framework of Dressler (1997) summarizes the acquisition of morphemes in three stages: pre-morphology, proto-morphology, and morphology. In the pre-morphology stage, the child is between the ages of 1;3, and 1;5, and the grammatical module has not yet been developed. In the proto-morphology stage, the child is between the ages of 1;6 and 1;9, and the system of morphological grammar is still developing. By the end of this stage, at the age of 1;9, the inflectional, derivational, and compounding morphology subsystems start to develop (see Aksu-Koç, 2010). According to Güven and Leonard (2020), the “learner-friendly properties of Turkish noun morphology” provide early learning opportunities for typically developing children. Such properties include rich inflectional suffixes that mark a single dimension, and their syllabic nature is assumed to facilitate children’s acquisition of Turkish noun morphology. The early acquisition of Turkish noun suffixes has been investigated by various researchers (Ekmekçi, 1987; Topbaş et al., 1997; N. Ketrez, 2004; F. N. Ketrez & Aksu-Koç, 2009; Topbaş et al., 2012). In their investigation of the spontaneous speech production of monolingual children, Topbaş (1997) pointed out that dative and accusative case suffixes are the first to emerge at the age of 1;3, followed by locative, ablative, and instrumental case suffixes, which emerge between the ages 1;4 and 1;6 (Table 1). Genitive and possessive case suffixes are produced from the age of 1;3 onwards, and the use of the possessive suffix starts from early on in combination with case markings (Rothweiler et al., 2010).

Table 2. Acquisition of case markers in standard Turkish (from Rothweiler et al., 2010, p. 546; Topbaș, 1997, p. 131).

Age	Case Marking
1;3	DAT	ACC	GEN
1;4	DAT	ACC	GEN	LOC
1;5	DAT	ACC	GEN	LOC	ABL
1;6	DAT	ACC	GEN	LOC	ABL	INST

(Rothweiler et al., 2010, p. 546), adapted from Topbaș (1997, p. 131; N = 96), details the acquisition of case morphology in monolingual Turkish.

2.3. Case Marking in Heritage Turkish

Studies examining noun suffixes as a potential DLD marker in heritage Turkish-speaking children should consider the linguistic differences between the Turkish variety acquired as an immigrant language and standard Turkish. To date, the typical and atypical acquisition of Turkish as a first language and as a heritage language has been studied in Germany (Pfaff, 1991; Cindark & Aslan, 2004; Babur et al., 2007; Uzuntaş, 2008; Şan, 2008; Rothweiler et al., 2010; Şimşek & Schroeder, 2011; Schellhardt & Schroeder, 2013; Schroeder & Dollnick, 2013; Felser & Arslan, 2019; Krause & Roberts, 2020; Bayram, 2020, Chilla 2022), the Netherlands (Schaufeli, 1993; Çavuș, 2009; De Jong et al., 2010; Doğruöz & Backus, 2009), France (Akıncı & Koçbaş, 2002; Akıncı, 2008), Sweden (Bohnacker et al., 2016; Bohnacker & Karakoç, 2020), Belgium (Altinkamis & Simon, 2020), and the U.S. (Coşkun Kunduz & Montrul, 2022a, 2022b).

Language contact, the orate structure of heritage Turkish, and the context of acquisition have led to structural changes in the language system (Küppers et al., 2015). Language attrition, the use of L1 with the loss of many original features (Aalberse et al., 2019), is observed in the L1 skills of Turkish heritage speakers living in a second-language environment (Yağmur et al., 1999). Homogeneity among Turkish heritage speakers in terms of language production has not been found (Johanson, 1991) because their output in L1 ranges from a monolingual mode in both of their languages to a bilingual mode with varying degrees of language mixing. Backus (2003) describes spoken Turkish in migrant settings as “mixed lects” characterized by a high degree of variability. Rehbein et al. (2009) define the majority language “German” in the migrant context in Germany as a “catalyst” language, which may cause Turkish heritage speakers (HSs) to either develop new forms or use existing ones in different ways. The onset of standard Turkish, which may be called the acquisition of a formal register of Turkish, starts in Turkish classes upon formal schooling. In contrast, the orate structures of Turkish as a heritage language are acquired informally in the private environment, including through family, relatives, and acquaintances, and in public environments, such as in shops, on the street, at the market, on the bus, and through the media (Hamurcu-Süverdem & Akıncı, 2017; Schroeder, 2020; see Maas, 2008, 2010, for socio-situational register).

Several studies have claimed that case is a vulnerable area of HSs’ grammar (De Groot, 2005; Leisiö, 2006; Montrul & Bowles, 2009; Polinsky, 2018; Flores, 2020). While some experimental studies (Polinsky, 1995) stated that case marking in the HL diverges from the homeland baseline, other naturalist studies (Preston, 1986) reported that case is not broadly affected. Studies on case markers in heritage Turkish have mostly provided results of systematic investigations of heritage Turkish in Germany, one of the target countries in this study. The limited studies on the linguistic features of heritage Turkish in France and the U.S. have investigated relative clauses (Akıncı, 2018; Coşkun Kunduz & Montrul, 2022a); clause combining in heritage Turkish (Iefremenko & Schroeder, 2019); lexical access (Akıncı, 2021); language use, attitude, and maintenance (Akıncı & Yagmur, 2000; Coşkun Kunduz, 2022); and narratives (Akıncı, 2002). The use of case markers in heritage Turkish in France was investigated by Chilla and Şan (2017) by comparing BiTD and BiDLD Turkish–French children with BiTD and BİDLD Turkish–German children aged 5;0–8;9 using TEDIL (Topbaş & Güven, 2011), which is a language assessment tool used to identify DLD in monolingual Turkish-speaking children. The results of the study showed that, although BiTD heritage Turkish children in France used the extended ablative case marker, which is a characteristic of heritage Turkish in Germany, their results did not particularly differ from standard Turkish, as standard Turkish is the dominant language in the heritage context in France due to the migration status and socioeconomic situation (Chilla & Şan, 2017). Previous research on ablative case marking in heritage Turkish in Germany has provided evidence for the overextension of the ablative case to other contexts (Boeschoten, 1994; Cindark & Aslan, 2004; Sirim, 2009; Şimşek & Schroeder, 2011). This linguistically interesting use seems to be modeled on the functions of the German equivalent von “from” (Backus & Boeschoten, 1998). Investigating the speech samples of 24 heritage child bilinguals aged 8–12 in Germany, Boeschoten (1994) reported that the use of the ablative, as in the example below (8), is certainly extended compared to any other Turkish variety.

(8)	Target
	(Oda-m-ın		duvar-lar-ı-nda)		diploma	ve	resimli	yapboz-lar	var.
	(Room-POSS.3SG-GEN		wall-PL-POSS:3G-LOC)		certificate	and	illustrated	puzzle-PL	exist
	“(On the wall of my room) there’s a certificate and illustrated puzzles.”
	Response
	Urkunde	var,	ondan sonra	Puzzle-ler-den		resimler
	Certificate	exist	then	puzzle-PL-ABL		picture-PL
	*“There is a certificate, and then pictures (made) from puzzles.” (Boeschoten, 1994, p. 261)

Further research on deviations in heritage Turkish (Şimşek & Schroeder, 2011; Backus et al., 2013) indicates that the Turkish-speaking population in migrant settings in Germany substitutes the accusative case marker with a dative case marker in contexts in which an accusative case is needed. Example (9) was taken from LAS-Project Literacy Acquisition in Schools in the Context of Migration and Multilingualism, in which written and/or narrative data were obtained from heritage bilingual school children attending classes one and seven (Şimşek & Schroeder, 2011).

(9)	Target
	kadın	da	bir	kapı-yı	göster-di
	woman	and	DEF	door-ACC	show-PAST
	“And the woman pointed out a door.”
	Response
	kadın	da	bir	kapı-ya	göster-di (FEH, 1. Class)
	woman	and	DEF	door-DAT	show-PAST
	“And the woman pointed out a door.”

The loosening of standard Turkish norms is also found in the substitution of the dative case marker with the accusative case marker in heritage Turkish in Germany (Cindark & Aslan, 2004; Sirim, 2009; Backus et al., 2013; Chilla, 2022). In their study, Sirim (2009) used the corpus of “Powergirls” aged 13–24 (see Keim, 2007), including in-group conversations, and they showed that bilingual Turkish speakers use object NPs marked with the accusative case suffix, which is used with the dative case in standard Turkish, as in (10):

(10)	Target
	İşyeri-ne	telefon ed-iyor-um.
	workplace- DAT	phone-PRES.PROG-1SG
	“I call to work.”
	Response
	telefon aç-ıyo-m	işyeri-ni
	phone-PRES.PROG-1SG	workplace- ACC
	“I call to work.” (Sirim, 2009, p. 55).

In terms of the use of the accusative case on direct objects, Turkish, allowing overt and zero marking, is considered a Differential Object Marking (DOM) language, whereas German is considered a non-DOM language due to obligatory accusative marking. Krause and Roberts (2020) focused on Differential Object Marking (DOM) phenomena in their studies to determine whether German, as a non-DOM language, affects DOM in the language of Turkish heritage speakers with German as their dominant language. The results of the study, which included the corpus of 32 heritage adult bilinguals aged 18–34, suggested that there was no cross-linguistic (CLI) effect of German as the dominant language on heritage Turkish as the weaker language, but heritage bilingual speakers with a low L1 proficiency showed “a sensitivity to the animacy level of the indefinite direct objects” (Krause & Roberts, 2020, p. 313). In Turkish, the accusative case marker and prenominal indefinite determiner bir “one” marks the definiteness status of the object NP (Felser & Arslan, 2019). Focusing on heritage Turkish in their studies, Felser and Arslan compared the data of heritage Turkish speakers (N = 20) with those of lL2 (N = 21) and Turkish monolinguals (N = 25) aged 19–29 to determine whether heritage speakers had difficulties in appropriately using (in-)definite NPs, and they found that heritage speakers showed overextension in “the use of definite NPs to pragmatic contexts that normally require an indefinite one” (Felser & Arslan, 2019, p. 39).

The changing nature of genitive–possessive constructions has been a research theme of several studies of heritage Turkish in Germany (Boeschoten, 2000; Johanson, 2002; Cindark & Aslan, 2004; Sirim, 2009; Şimşek & Schroeder, 2011; Schroeder, 2014; Schellhardt & Schroeder, 2015; Chilla, 2022). In the language production of bilingual Turkish-speaking young people, it is observed that they omit either the genitive case marker (11) or the possessive marker (12). As stated by Csató and Johanson (1998, p. 219), it is important to note that, in informal spoken Turkish and in some dialects, omission of the possessive marker in a genitive–possessive construction is acceptable. Johanson (2002, p. 55) states that HSs use prepositive possessive constructions common to German and Turkish, such as bizim köy, unser Dorf’, our village. As per the “alternation hypothesis” (cf. Johanson, 2002), if there are two alternative constructions in the target language, the structure for which a parallel exists in the L1 is mostly overgeneralized by HSs. Additionally, the redundant use of the possessive marker in (13) is typically found in the early stages of monolingual Turkish acquisition, but it stabilizes in heritage Turkish (Küppers et al., 2015, p. 36).

(11)	Target
	Park-ta-ki	kadın-lar-ın	dedikodu-lar-ı	ne	üzerine?
	park-LOC	woman-PL-GEN	gossip-PL-POSS-3SG	what	about-DAT
	“About what are the woman in park gossiping?”
	Response
	Park-ta-ki	kadın-lar	dedikodu-lar-ı	ne	üzerine?
	park-LOC	woman-PL-Ø	gossip-PL-POSS.3SG	what	about-DAT
	*“The woman in the park what about is their gossip?” (Cindark & Aslan, 2004, p. 4)

(12)	Target
	O kişi-nin	değişik	bir	ülke-den	gel-diği-nden		veya
	this-person-GEN	different	DEF	country-DAT	come-FNOM-POSS-ABL		or
	kültür-ü-nden	dolayı	dışla-n-ma-sı
	culture-POSS.3SG-ABL	due to	exclude-PASS-FNOM-POSS-
	çok	gör-ül-üyor.
	often	see-PASS-PROG-3G
	*“It is very common for that person to be excluded due to being from a different country or having (a different) culture.”
	Response
	O kişi-nin	değişik	bir	ülkeden,	veya	kültürü	dolayı	dışla-n-ma
	this-person-GEN	different	DEF	country-DAT	or	culture-ACC	due to	exclude-PASS-FNOM-Ø
	çok	görülüyor
	often	see-PASS-PROG-3G
	“It is very common for that person to be excluded due to from a different country or culture.”(Schroeder, 2014, p. 37)

(13)	Target
	Kopya	çekmek-le	başka	insan-ın	hakkını	yersin.
	copy	take-INSR	another	person-GEN.3SG	right-POSS3SG-ACC	take-AO-2SG
	“By copying (in the exam) you cheat others of their rights.”
	Response
	Kopya	çekmek	başka	insan-ın	hakkı-sı-nı	yiyorsun
	copy	take	another	person-GEN.3SG	right-POSS.3SG-POSS.3SG-ACC	take-IPFV-2SG
	sayılır.
	count-PASS-AOR
	*“To copy it is considered you are cheating others of their rights.” (Küppers et al., 2015, p. 36).

As a dialect-leveling feature of heritage Turkish in Germany and other western countries, the use of the instrumental marker ile “with” in -lEn is found in typical utterances of the older generation, who brought -lEn from the eastern Anatolian dialect of Turkey (Boeschoten, 2000; Şimşek & Schroeder, 2011). The use of -lEn is observed in young bilingual Turkish students’ language production in Germany, which may result from the influence of L2 German, reinforcing closed syllables, as in the example below (14) (Şimşek & Schroeder, 2011).

(14)	Target
	Çanta-sı-yla	oynu-yor-du.
	bag-POSS3SG-INST	play-PROG-PAST-3SG
	“S/he played with his bag.”
	Response
	Çantasın-lan	oynu-yor-du.
	bag-POSS3SG-INST	play-PROG-PAST-3SG
	“S/he played with his bag.” (Şimşek & Schroeder, 2011, p. 213)

2.4. Case Marking in DLD in Turkish

Recent research indicates that Turkish-speaking children with DLD have difficulties with counterfactuals, tense, and negation in verb morphology and case marking in nominal morphology (Babur et al., 2007; Chilla & Babur, 2010; Acarlar & Johnston, 2011; Yarbay-Duman et al., 2015; Topbaş et al., 2016; Yarbay-Duman & Topbaş, 2016; Chilla & Şan, 2017; Chilla, 2022). Moreover, compared to their age-matched typically developing peers, many Turkish-speaking children with DLD produce shorter utterances with fewer morphemes, particular morphological errors, and less complex sentences (Topbaş et al., 2016). Derivational morphology has also been shown to pose problems for children with DLD in sentence repetition tasks (Topbaş, 2010). A study conducted by Bernreuter (2004) focusing on the inflectional errors in speech samples of two heritage language children with TD and two children with DLD argued that the nominal domain is vulnerable and that learners mostly make omission errors rather than substitution errors. Çavuș (2009) studied structural and semantic case marking in heritage child bilinguals with DLD (N = 19) by comparing their narrative data with those of heritage child bilinguals with TD (N = 19) and monolinguals with TD (MOTD) (N = 20) aged 6–8;5. The study showed that there were no difficulties with structural case and that heritage child bilinguals with DLD produced utterances with fewer semantic case markers than BITD children; however, they did not make more errors than the others when they used case forms (Çavuș, 2009). De Jong et al. (2010) investigated DLD in Turkish and Dutch in heritage child bilinguals. They compared four groups, each consisting of 20 children: MODLDs (L1 Dutch), younger children (L1 Dutch), and two groups of heritage child bilinguals with and without DLD aged 6–8 years. Regarding case morphology, the results suggested that heritage child bilinguals with DLD had more difficulties in noun morphology (case marking: accusative: 56% correct) than verb morphology (89% correct). Rothweiler et al. (2010) investigated the grammatical features of heritage Turkish by focusing on case morphology in the spontaneous speech data of successive bilingual children (N = 2 with BiTD, N = 2 with BiSLI) aged 3–7 years. The study showed that BiDLD children had more deficits in case marking than BiTDs and that they produced errors, such as omission (15) and substitution (of the dative with the accusative) (16). Topbaş et al. (2016) compared the data of three children with DLD (mean age 5;2) with those of three age-matched TD children from the T-SALT Database. Their study aimed to present several morphological difficulties and errors based on spontaneous speech samples of children with DLD, concerning the assessment of DLD (Topbaş et al., 2016). TEDİL (Topbaş & Güven, 2011) was used to identify DLD in monolingual Turkish children, and picture descriptions and the narrative tasks of the same test were applied to collect speech samples. According to the results, children with DLD used fewer case suffixes than their typically developing monolingual peers; they demonstrated a high rate (87%) of case suffix usage in obligatory contexts (Topbaş et al., 2016). Errors occurred as substitutions, such as the accusative case for the dative case, as in (17), and the omission of the accusative (17) in obligatory contexts. Furthermore, Topbaş and Maviş (2016) investigated the use of case markings (dative, accusative, locative, and ablative) of 18 monolingual DLD children aged between 3 and 5 years compared to the data of children aged 3;8–4;10 (±6 months from the DLD group, mean age 4;2) retrieved from T-SALT (Turkish Systematic Analysis of Language Transcripts, Acarlar et al., 2006). The same study reported that the mean use of the dative and accusative was the highest in the utterances of TD children and that the use of case marking was low in DLD children. Lastly, a very recent study focused on the noun morphology of monolingual Turkish-speaking children by comparing speech samples of DLD children (N = 20) aged 4–7;10 with those of 20 age-matched TD children (4;0–7;3) and 20 younger MLU-matched children (2;0–4;3) (Güven & Leonard, 2020). The results suggested that children with DLD used fewer noun suffixes than both the younger and age-matched TD children, and the use of bare nouns in place of an overt suffix was a possible error by all groups.

Further studies on DLD in Turkish-speaking heritage children (Babur et al., 2007; Rothweiler et al., 2010; Topbaş, 2010; Chilla & Şan, 2017; Şan, 2018; Gökgöz et al., 2020; Chilla, 2022) identified the production of case morphology as a vulnerable area in DLD child heritage bilinguals, obtaining results similar to the abovementioned studies, such as a less frequent use of case suffixes in obligatory contexts and the production of errors, such as omission and substitution. Regarding case marking, Rothweiler et al. (2010) investigated the data of two TD and two DLD Turkish heritage children in Germany aged 3–6;5, and they discovered high accuracy, although the Bi-DLD children produced more errors (15%) than the BI-TD children (5.6%). The study found that the substitution of the accusative case for the dative case particularly disentangled DLD from TD (Rothweiler et al., 2010). A recent study (Blom et al., 2022) investigated the grammatical language development of Turkish and Dutch in 10 Bi-DLD Turkish–Dutch and 10 Bi-TD Turkish–Dutch children in comparison to that of control groups of 10 MO-DLD and 10 MO-TD Dutch children aged 5–6 years. In this longitudinal study, the authors analyzed semi-spontaneous language samples and found that, in terms of case marking, the lack of genitive suffixes in simple constructions and the omission of possessive marking in simple genitive–possessive constructions could be clinical markers of DLD (Blom et al., 2022). The most distinguishing result of Blom et al. (2022)’s study is that the accusative case is less promising for use as a clinical marker to differentiate DLD in Turkish heritage children in the Netherlands, as its omission and substitution can be a feature of the input in heritage Turkish regardless of the relative frequency of its omission and substitution in the data.

It is evident from these studies that case morphology presents a significant challenge for children with DLD, in both monolingual and heritage bilingual settings, and a common pattern emerges: children with DLD demonstrate lower accuracy in case marking, frequently omitting obligatory case suffixes or substituting them incorrectly (e.g., the accusative for the dative). This vulnerability appears in various linguistic contexts, from spontaneous speech samples to controlled narrative tasks. In particular, nominative and structural case marking pose fewer challenges, whereas semantic case markers (dative, accusative, ablative, and locative) are often problematic, with errors manifesting as omission and substitution. A longitudinal study by Blom et al. (2022) further highlights that, while accusative case marking is often affected, it may not serve as a strong clinical marker of DLD in heritage Turkish due to similar patterns appearing in the general heritage input. These findings underscore the importance of considering both linguistic and environmental factors when assessing DLD in bilingual populations.

(15)	Target
	Şimdi	ben	bu-nu	tak-a-yım	mı?
	Now	I	this-acc	attach-OPT-1SG	INT
	“Shall I attach it now?”
	Response
	Dimmi	ben	bu-Ø	gada (DLD-Fer, 6;5)
	now	I	that- Ø (for ACC)	attach
	*“I attach this now.” (Rothweiler et al., 2010, p. 550)

(16)	Target
	O	kedi	o-na	bak-ıyor.
	This	cat	he-DAT	look-PRES.PROG
	“This cat looks at him.”
	Response
	O	Katze	bakıyo	o-nu. (DLD-Dev, 5;5)
	This	cat	look-PRES.PROG	he-ACC
	“This cat looks at him.” (Rothweiler et al., 2010, p. 550)

(17)	Target
	Atta-ya	gid-iyor.
	Out-DAT	go-PRES.PROG
	“S/he is going out.”
	Response
	Atta-yı	gid-iyo.
	Out-ACC	go-PRES.PROG
	“S/he is going out.” (Topbaş et al., 2016)

(18)	Target
	Ayı	kepçe-yi	it-ti.
	bear	truck-ACC	hit-PAST
	“The bear hit the truck.”
	Response
	Ayı	it-ti	çepçe-Ø
	bear	hit-PAST	truck
	*”The bear hit the truck.” (Topbaş et al., 2016)

2.5. Current Study

As case in Turkish appears to be affected by language contact in the heritage setting, this study investigates whether it can still be used as a clinical marker of DLD in child heritage bilinguals. Accordingly, this study provides a detailed analysis of the case and possessive markers in the utterances of heritage BiTDs and BiDLDs in Germany, France, and the U.S., comparing them to typically developing lL2 BiTD children living in Germany. The reason for including lL2 BiTD children is that, as first-generation immigrant children, they have better abilities than simultaneous and successive bilingual children in the HL (Montrul, 2016; Paradis et al., 2021). This is associated with their acquisition of the standard Turkish variety in a qualitatively rich language environment, their exclusive exposure to it for an extended period, and their access to formal language registers during the school-age period.

In light of the features of heritage Turkish and DLD markers regarding case and possessive markers, we pursue the following research questions:

• How do heritage BiDLD children compare to their heritage BiTD peers and lL2 BiTDs in their production of case and possessive markers of standard Turkish?
• Which age and input variables predict variance in performance using case and possessive markers in standard and heritage Turkish?
• Given the potential overlap between features of heritage Turkish and clinical markers of DLD in standard Turkish, can case and possessive markers be used to identify DLD in the bilingual heritage context? Does performance on case and possessive markers improve if certain features of heritage Turkish are considered?

3. Materials and Methods

3.1. Participants and Procedures

This study includes data of 83 Turkish-speaking heritage children from Germany, France, and the U.S., comprising 73 bilinguals with TD and lL2 BiTD, and 10 with DLD (age range 5;0–12;4), who participated in the BiLaD1 and BiliSAT2 projects (see

Table 3. Participants and overview of background variables related to bilingualism (mean, SD, and range).

Background Variables (PaBiQ)	Germany			France		U.S.
	BİTD (N = 38)	BİDLD (N = 12)	IL2_BITD (N = 10)	BİTD (N = 14)	BİDLD (N = 4)	BİTD (N = 5)
Age (months)	95.11	79.00	126.9	85.79	85.50	78.80
	(27.23)	(13.12)	(24.96)	(8.59)	(12.97)	(21.39)
	49–161	64–111	86–149	70–99	67–95	61–108
L2 AoO	18.87	25.92	97.20	7.07	6.0	19.20
	(16.31)	(13.75)	(25.42)	(13.40)	(12.0)	(19.63)
	0–42	0–41	54–126	0–39	0–24	0–42
L2 LoE	76.24	53.08	29.70	78.71	79.50	59.60
	(34.30)	(17.21)	(13.37)	(18.10)	(13.28)	(39.37)
	13–161	29–81	9–54	31–99	67–95	20–108
% Early L1 Exposure	67.89	75.42	94.90	72.36	78.50	87.60
	(20.11)	(12.01)	(8.33)	(12.17)	(5.20)	(11.67)
	8–100	50–100	80–100	42–83	71–83	75–100
Current L1 richness (/14)	5.68	6.50	9.50	7.57	6.50	7.20
	(2.46)	(2.11)	(2.42)	(1.45)	(4.04)	(3.83)
	1–11	3–11	7–15	5–10	1–10	3–10
Current L1 Skills (/15)	10.13	9.25	14.30	12.14	7.75	10.60
	(3.56)	(3.44)	(1.25)	(2.74)	(6.13)	(3.85)
	0–15	2–15	11–15	6–15	0–14	6–14
Current L1 use (/16)	9.08	9.83	12.10	11.36	11.75	9.60
	(3.03)	(2.12)	(2.47)	(1.50)	(1.50)	(2.19)
	3–15	6–14	7–15	9–15	10–13	6–12

). Participants living in different federal states of Germany and France were recruited in kindergartens, mainstream schools, speech and language therapy centers, and community associations and through private contacts. The authors recruited participants in the U.S. only through private contacts. Written informed consent was obtained from all children’s parents/legal guardians for data collection through a parental questionnaire and for their children’s participation in this research. The protocol was approved by the “Kommission für Forschungsfolgenabschätzung und Ethik” of the Carl-von-Ossietzky University of Oldenburg. All the child heritage bilinguals were born in Germany, France, or the U.S. and were primarily exposed to heritage Turkish in their daily life, alongside the majority language German, French, or English, to which they were exposed either from birth (i.e., simultaneous bilingualism) or upon enrolment in kindergarten (early successive bilingualism). In contrast, the late participants, referred to as the late successive bilingual typically developing group (lL2 BiTD), were born in Turkey and moved to Germany at school age. Compared with the heritage bilingual groups, the lL2 BiTDs were relatively old when they started acquiring the majority language as their L2; however, they are expected to retain abilities in standard Turkish because they experienced monolingual exposure for an extended time in a qualitatively rich language environment. The clinical status of each child in Germany and France as having typical language development (BiTD) or developmental language disorder (BiDLD) was verified using an extensive procedure based on a standardized language assessment of each of their languages, with the pathology cut-off points adjusted according to their language dominance (cf. Thordardottir, 2015). For the assessment of L1 and L2, norm-referenced L1 and L2 tests that are frequently used by speech-language pathologists and that cover the age range were used. For an assessment of L2 German, tests assessing different language domains, including expressive and receptive vocabulary (WWT 6–10; Glück, 2011), morphosyntax comprehension (LiSe-DaZ; Schulz & Tracy, 2011), and production (LiSe-DaZ; Schulz & Tracy, 2011), as well as phonology (PLAKSS; Fox-Boyer, 2014), were selected. For French, N-EEL (Chevrie-Muller & Plaza, 2001) was used to assess language skills in domains such as expressive and receptive vocabulary and morphosyntax comprehension and production, and BILO (Khomsi et al., 2007) was used to evaluate phonology. For a view of standardized tests used to assess cognitive abilities and language skills in German, French and Turkish, see

Table A1. Standardized tests used to assess cognitive abilities and language skills in German and Turkish.

Language	Test	Language Skills Tested					Scoring Method	Age Range
		Phonology	Vocabulary Expression	Vocabulary Reception	Morphosyntax Comprehension	Morphosyntax Production
German	WWT 6–10a		Picture naming	Picture selection	-		Individual subtest scores	5;6–10;11
	Lise-DaZ b	-	-	-	Picture–sentence matching, TVJT	Story, sentence completion, lead-in questions	Individual subtest scores	Monolinguals: 3;0–6;11 Bilinguals: 3;0–7;11
	PLAKSS-II c	Picture naming		--	-	-	Individual subtest scores	2;6–7;11
French	NEE-L d	-	Picture naming	Picture selection	Picture–sentence matching	Sentence completion	Individual subtest scores	5;7–8;6
	BILO e	Word repetition	-	-	-	-	Individual subtest scores	5;0–15;0
Turkish	TEDİL f	--	Picture naming	Picture selection	Picture selection	Sentence completion/ construction	2 composite scores, 1 production and 1 comprehension	2;0–7;11

Notes: ^a Glück (2011); ^b Schulz and Tracy (2011); ^c Fox-Boyer (2014); ^d Chevrie-Muller and Plaza (2001); ^e Khomsi et al. (2007); ^f Topbaş and Güven (2011).

in Appendix A. As the heritage bilingual children from the U.S. were not included in the BiLaD project, they were not assessed in terms of their L2 English. The clinical status of each child in Germany and France was verified as typically developing or language impaired based on whether they scored below the pathology cut-off on monolingual norm-referenced tests in two domains of each of their languages after adjusting the pathology cut-off criterion in line with the dominance status of the language being assessed, as estimated using the Parents of Bilingual Children Questionnaire (PABIQ, Tuller, 2015). In contrary to the L2 tests, the TEDİL (Topbaş & Güven, 2011) test battery, which is used to assess language skills in Turkish, offers only two global composite scores for each of the comprehension and production subtests, collapsing morphosyntax and lexical semantics together; hence, heritage Turkish bilingual children were assigned to the BiDLD group if they scored in the pathology range in either the production or comprehension subtests. An overview of the participants (based on the verified clinical status) is given in Table 3.

Comparisons of the background variables between the child groups (heritage BiTD in Germany, France, and the U.S.; heritage BiDLD in Germany and France; and lL2 BiTD in Germany) revealed significant overall effects as a function of group on all variables (Kruskal–Wallis test): chronological age, [Χ²(5, N = 83) = 18.217, p = 0.003]; L2 AoO, [Χ²(5, N = 83) = 35.436, p = 0.000]; L2 LoE, [Χ²(5, N = 83) = 22.993, p = 0.000]; early L1 exposure, [Χ²(5, N = 83) = 22.884, p = 0.000]; current L1 use at home, [Χ²(5, N = 83) = 18.111, p = 0.003]; L1 richness [Χ²(5, N = 83) = 17.276, p = 0.004]; and current L1 skills, [Χ²(5, N = 83) = 20.286, p = 0.001].

Subsequent Mann–Whitney U tests with Holm–Bonferroni adjustment showed no significant differences between the heritage BiTDs and heritage BiDLDs in Germany and France or between the heritage BiDLDs in Germany and heritage BiTDs in the U.S. for any of the input variables, implying comparable exposure patterns between the groups. A significant difference (p < 0.001) emerged in all variables between the heritage BiTDs in Germany and lL2 BiTDs. Similarly, significant differences (p < 0.001) with large effect sizes emerged between the heritage BiDLDs in Germany and lL2 BiTDs in all variables. Except for in current L1 use, the heritage BiTDs in France differed from the lL2 BiTDs in all variables (p < 0.001).

3.2. Materials

3.2.1. The Parental Questionnaire PaBiQ

In this study, the PaBiQ (Tuller, 2015), developed during COST Action IS0804, was used to collect background information, including the children’s age, past and current quantity and quality of L1/L2 input and use, and risk factors for DLD (Hamann & Abed Ibrahim, 2017). The questionnaire allows for the calculation of the experiential language dominance index (see Unsworth, 2016), serving as a baseline for adjusting the pathology cut-off criterion. The age and input variables in the current study were as follows: chronological age, age of onset of exposure to the majority/societal language (AoO_L2), length of exposure to the L2 (LoE_L2), early L1 exposure (measured by the frequency of early language use and the exposure and diversity of exposure contexts before the age of 4), current L1 use (measured by the proportion of relative L1 use within the family), and L1 linguistic richness (the richness of the L1/HL language environment within and outside the child’s home). In addition, the current L1 skills based on parental estimation were explored as a further potential predictor of performance. The Turkish translation of the PaBiQ was used with the children’s parents in a phone or face-to-face interview.

3.2.2. Elicited Production of Turkish Case and Possessive Markers

This study made use of a corpus, including all the utterances of bilingual Turkish-speaking children in Germany, France, and the U.S. that were produced in the Turkish Test of Early Language Development (Turkish Early Development Test (TELD-3: T; TEDIL, Topbaş & Güven, 2011)). The TEDİL is an adaptation of the Test of Early Language Development (TELD-3), originally developed by Hresko et al. (1999). The test is frequently used by SLPs in Turkey as a screening tool for language delay and impairment. It is a norm-referenced language assessment tool that measures receptive and expressive language performance skills in children (2;0–7;11) and aims to identify a child’s strengths and weaknesses regarding verbal language competencies. In this study, we used the form for the expressive language knowledge of set A, which includes 22 questions for semantics knowledge and 17 questions measuring syntax/morphology. Scoring covers expressive, receptive, and global language performances.

The TEDİL testing process does not necessarily begin with the first item. The entry point depends on the child’s chronological age, calculated in years and months. This is “based on the assumption that the child can successfully perform all of the items that fall below the starting point” (Benner & Grim, 2012, p. 137). Participants must provide three correct consecutive responses to items for their age level to determine the entry point. For each correct response, the child is awarded 1 point. If the child fails three items in a row, testing proceeds backward until this criterion is met. Once the entry point is specified, testing continues forward, and the test is terminated when the child incorrectly responds to three successive items. In the current study, all the utterances produced by children on test items providing obligatory contexts, even if they were higher in number than listed in

Table 4. Overview of item and case types in TEDİL (Topbaş & Güven, 2011) and number of obligatory contexts per structure type.

Item Types	Number of Item Types	Case Types and Possessive	Number of Obligatory Contexts
Sentence Repetition	19	Accusative	10	35
		Dative	6
		Locative	0
		Ablative	1
		Instrumental	1
		Genitive	5
		Possessive	12
Sentence Construction	3	Accusative	1	4
		Dative	1
		Locative	0
		Ablative	0
		Instrumental	0
		Genitive	1
		Possessive	1
Sentence Completion	1	Accusative	0	1
		Dative	0
		Locative	1
		Ablative	0
		Instrumental	0
		Genitive	0
		Possessive	0
Question–Answer	2	Accusative	1	4
		Dative	0
		Locative	0
		Ablative	0
		Instrumental	0
		Genitive	1
		Possessive	2
Interpretation	4	Accusative	0	1

, were used as a corpus to extract case and possessive markers. For an overview of the case and possessive markers investigated in this study, please refer to Table 4. Note that the test itself does not provide any solution for the obligatory contexts regarding narratives; they are not listed or included in Table 4. However, the case and possessive markers produced in these tasks were also included in this study. Some sample items from the TEDİL-SRT depicting every case type and possessive marker are also listed in

Table 5. Sample items from the TEDİL-SRT depicting every case type and possessive marker.

Sentence Type	Sample Sentence			Number of Obligatory Contexts
Simple Sentence	Baba-m	iş-e	git-ti.	Possessive (1) Dative (1)
	father-1SG.POSS	work-DAT	go-PAST
	“My father went to work”.
Adverbial Clause	Çocuk-lar	cam-ı	kır-ıp	Accusative (1)
	Child-P	window-ACC	break-VN(-Ip)
	kaç-tı-lar.
	run away-PAST-3PL
	“Having broken the window, the children run away”.
Noun Clause	Anahtar-ı	anne-m-e	ver-diğ-im-i	Accusative (2) Dative, (1) Possessive (1)
	key-ACC	mother-POSS-DAT	give-FNOM-
			1SG.POSS-ACC
	unut-muş-um.
	forget-EV/PF-1SG
	“I forgot that I gave the key to my mother”.
Relative Clause	Ağaç-a	tırman-an	kedi	Dative (1) Ablative (1)
	Tree-DAT	climb-SUBJP	cat
	köpek-ten	kaç-ıyor-du.
	dog-ABLT	Run away-PROG-PAST
	“The cat that climed the tree run away from the dog”.
Adverbial Clause	Anne-m-le		baba-m	Instrumental (1) Possessive (2)
	Mother-1SG.POSS.INST		father-1POSS
	konuş-ur-ken	biz	ders çalış-tı-k.
	talk-AOR-CV	we	study-PAST-3PL
	“As my mother and my father talked, we studied”.
Simple Clause	Ayşe-nin	arkadaş-ı-nın		Genitive (3) Possessive (3)
	Ayşe-GEN	friend-3SG.POSS-GEN
	bebek-i-nin	elbise-si
	Doll- 3SG.POSS-GEN	cloth-3SG.POSS
	mavi-y-di.
	blue-P.COP
	“The cloth of Ayse’s friend’s tool was blue”.

.

The data for this study were obtained from the children’s responses in the individual sessions. Special dictaphones were used to record the children’s responses during the testing. Two linguistically trained raters phonetically transcribed, verified, and coded the data.

3.3. Scoring

For each participant, all contexts requiring case markings were extracted, and the proportion of correct case markings (case and possessive markers) was calculated once using standard Turkish as the reference variety and once allowing for case and possessive markers characteristic of the heritage Turkish variety. As the nominative case cannot be detected due to it not being a phonologically realized morpheme, it was not included in this study.

Table 6. Examples of responses not conforming to standard Turkish; case and possessive markers are given in bold.

Case and Possessive Conforming to Heritage Turkish but Not to Standard Turkish
Overextension of ablative	Expected response in standard Turkish:
	Kitap		oku-ma-yı	ders-ten		daha çok	sev-er-im
	Book		read-VN-ACC	lesson-ABL		more	like-AOR-1SG
	“I like reading book more than lesson.”
	Response of child BiTD HS (acceptable in heritage Turkish):
	Kitap-tan		oku-ma-yı	ders-ten		daha çok	sev-er-im
	Book-DAT		read-VN-ACC	lesson-ABL		more	like-AOR-1SG
	“I like reading from book more than lesson.”
Substitution of accusative with dative	Expected response in standard Turkish (sentence repetition):
	Dede-yi			sev-me-miş
	grandfather-ACC			like-NEG-EV/PF
	“S/he did not like the grandfather.”
	Response of child BiTD (acceptable in heritage Turkish):
	Dede-ye			sev-me-miş.
	grandfather-DAT			like-NEG-EV/PF
	“S/he did not like to the grandfather.”
Substitution of dative with accusative	Expected response in standard Turkish (sentence repetition):
	Zeynep	ve	arkadaş-lar-ı-na			birşey	alı-yor-du.
	Zeynep	and	friend-PL-3SG.POSS-DAT			something	buy-PRES.PROG-PAST
	“S/he was buying something for Zeynep and her/his friends.”
	Response of child BiTD and BiDLD HS (acceptable in heritage Turkish):
	Zeynep	ve		arkadaş-lar-ı-nı		birşey	al-ıyor-du
	Zeynep	and		friend-PL-ACC		something	buy-PRES.PROG-PAST
	“S/he was buying something for Zeynep and her/his friends.”
Substitution of dative with locative	Expected response in standard Turkish (sentence repetition):
	Sonra		dışarı	çık-ınca	araba-ya	eşya-lar-ı-nı	koy-uyor
	And then		out	go-VN	car-DAT	belonging-PL-ACC	put-PRES.PROG
	“And then when s/he goes out, s/he puts her/his belongins into the car.”
	Response of child BiTD and BiDLD HS (acceptable in heritage Turkish):
	Sonra	dışarı	çık-ınca	araba-da		eşya-lar-ı-nı	koy-uyor
	And then	out	go-VN	car-LOC		belonging-PL-ACC	put-PRES.PROG
	“And then when s/he goes out, s/he puts her/his belongins in the car.”
The use of instrumental marker ile "with" in -lEn (dialect leveling feature)	Expected response:
	Mehmet	baba-sı		ile		balık tut-ma-ya	git-ti
	Mehmet	father-3SG.POSS		with		angle	go-PAST
	“Mehmet went to angle with his father.”
	Response by adult and child HS BiTD (acceptable in heritage Turkish):
	Mehmet baba-sın-lan			balık tut-ma-ya		git-ti.
	Mehmet father-3SG.POSS-INST			angle-DAT		go-PAST
	“Mehmet went to angle with his father.”
Omission of genitive in genitive–possessive construction	Expected response:
	Ayşe-nin	arkadaş-ı-nın		bebeğ-i-nin		elbise-si	mavi-ydi
	Ayşe-GEN	friend-ACC-GEN		doll-ACC-GEN		cloth-POSS	blue-P.COP
	“The cloth of doll of Ayşe’s friend was blue.”
	Response by adult and child HS BiTD/BiDLD (acceptable in heritage Turkish):
	Ayşe-nin	arkadaş-ı-Ø		bebeğ-i-Ø		elbise-si	mavi-ydi
	Ayşe-GEN	friend-ACC		doll-ACC		cloth-POSS	blue-P.COP
	“The cloth of doll Ayşe friend was blue.”
Omission of possessive marker in genitive–possessive construction	Expected response:
	Ben-im	arkadaş-ım		üç		gündür	hasta.
	I-GEN	friend-POSS		tree		for tree days	sick
	“My friend has been sich for tree days.”
	Response HS BiTD (acceptable in heritage Turkish):
	Ben-im	arkadaş-Ø		üç		gün	hasta
	I-GEN	friend-Ø		tree		day	sick
	“My friend has been sick tree days.”
Redundant use of possessive	Expected response:
	Ev-i-ne		git-ti
	home-3SG.DAT		go-PAST
	“He went to his home.”
	Response HS BiTD (acceptable in heritage Turkish):
	Ev-i-si-ne git-ti.
	home-3SG.POSS-3SG.POSS-DAT go-PAST
	“He went to his his home.”
DLD Markers in Monolingual Turkish
Omission of accusative	Expected response
	Anahtar-ı anne-m-e			verdi-ğim-i			unut-muş-um.
	Key-ACC mother-POSS-DAT			give-FNOM-ACC			forget-EV/PF-1SG
	“I forgot that I gave the key to my mother.”
	Response by child HS BIDLD
	Anahtar-Ø		anne-m-e	ver-diğ-im-Ø			unutmuşum
	key-Ø		mother-POSS-DAT	give-FNOM-ACC-Ø			forget-EV/PF-1SG
	“I forgot that I gave the key to my mother.”
Omission of dative	Expected response:
	Dün		araba	adam-a			çarp-tı
	Yesterday		car	man-DAT			hit-PAST
	“The car hit the man yesterday.-“
	Response by child HS BiDLD:
	Dün		araba	adam-Ø			çarp-tı
	Yesterday		car	man-Ø			hit-PAST
	“The car hit the man.”
Omission of locative	Expected Response:
	Yarın		sınıf-ta	bayram			yok.
	Tomorrow		class-LOC	festival			exist-NEG
	“Tomorrow there is no festival in the class.”
	Response by child HS BIDLD
	Yarın		sınıf-Ø	bayram			yok.
	Tomorrow		class-Ø	festival			exist-NEG
	“Theres is no festival class.”

provides examples of responses that do not follow standard Turkish, responses that are classified as target-like in heritage Turkish, and errors, namely, DLD markers.

3.4. Data Analysis

In this study, IBM SPSS 26 software (IBM Corp. Released, 2020) was used to conduct statistical analyses. Non-parametric tests, such as the Kruskal–Wallis test, were used for group comparisons due to the unequal and small sample sizes and the violation of normality assumptions. A post hoc Mann–Whitney U test with Bonferroni–Holm correction was used to further explore the between-group differences. Note that the Bonferroni–Holm correction is less conservative than the Bonferroni correction, reducing the risk of type I and II statistical errors. The effect sizes (r) were also calculated, with an effect size > 0.5 corresponding to a strong effect.

Before the regression analyses, non-parametric Spearman correction for significant correlations with the rate of successful case and possessive marker use was performed to examine the background variables obtained via the PABIQ, which were treated as dependent variables in the subsequent hierarchical regression analyses. Only those yielding significant moderate-to-strong bivariate Spearman correlations with the dependent performance variables were selected as potential predictors. These were later entered as independent variables in regression modeling to determine which background factors influenced the children’s performance on case and possessive markers in the TEDIL, using standard and heritage Turkish as reference varieties. Therefore, the standardized regression coefficients, R² values, and p values are reported for each model.

4. Results

4.1. TEDIL: Case and Possessive Marker Comparisons

As mentioned in Section 3.3, we calculated the rate of the successful case and possessive marker production of the bilingual children in the TEDİL, once with reference to the standard Turkish variety alone and once allowing for morphological errors typical of heritage Turkish, which do not affect the production of case or possessive markers. Firstly, to determine the general results, we focused on the results of the BiTDs, BiDLDs, and lL2ers without dividing the groups into countries. Regarding the standard Turkish and standard or heritage Turkish baselines, the means of all the heritage BiTDs, heritage BiDLDs, and lL2ers can be seen in Figure 1 and

Table 7. Group means (SD) of correct responses and error types. BiTD = bilingual children with typical language development; BiDLD = bilingual children with developmental language disorder; lL2_BiTD = late successive bilingual children with typical language development.

Group	%Target-like Case Clauses
	Reference Turkish Variety
	Standard Turkish	Standard or Heritage Turkish
Heritage_BİTD	82.15	88.01
	(16.19)	(21.13)
Heritage_BİDLD	69.04	85.33
	(23.01)	(13.98)
IL2_BİTD	97.57	99.27
	(2.99)	(1.54)

. The BiTDS, BiDLDs, and lL2 BiTDs significantly benefited from the consideration of heritage Turkish features while scoring (sequentially: Friedman: [X²(2) = 95.657, p = 0.000], Wilcoxon test: Z = −4.529, p = 0.000, η² = 0.360; Friedman: [X²(2) = 29.032, p = 0.000], Wilcoxon test: Z = −3.297, p = 0.001, η² = 0.679; Friedman: [X²(2) = 21.391, p = 0.000], Wilcoxon test: Z = −2.023, p = 0.043, η² = 0.409).

To examine further significant differences among the BiTDs, BiDLDs, and lL2ers from all countries in this study, we conducted the Kruskal–Wallis H test. The results suggested that there was a significant difference among these groups when both varieties were taken into consideration (Kruskal–Wallis test: standard Turkish [X²(2, N = 83) = 18.217, p = 0.003], standard or heritage Turkish [X²(2, N = 83) = 14.183, p = 0.001]). According to the results of pairwise comparisons (see

Table 8. The effect of (clinical) group membership on the rate of correct case and possessive attempt in the TEDIL (Kruskal–Wallis test) and post hoc pairwise comparisons (Mann–Whitney U tests); significant comparisons are given in bold. BiTD = bilingual children with typical language development; BiDLD = bilingual children with developmental language disorder; lL2_BiTD = late successive bilingual children with typical language development.

Group	%Target-like Case Clauses
	Reference Turkish Variety
	Standard Turkish	Standard or Heritage Turkish
	χ2(2, N = 83) = 21.391 p < 0.001	χ2(2, N = 83) = 14.183 p < 0.001
Heritage_BİTD vs. Heritage_BİDLD	U = 278.0	U = 333.0
	p < 0.05	p = 0.098
	r = 0.077	r = 0.038
Heritage_BİTD vs. IL2_BİTD	U = 67.0	U = 98.5
	p < 0.05	p < 0.05
	r = 0.221	r = 0.168
Heritage_BİDLD vs. IL2_BİTD	U = 7.5	U = 24.0
	p < 0.05	p < 0.05
	r = 0.569	r = 0.371

), although the heritage BiTDs scored significantly better than the heritage BiDLDs in terms of the standard Turkish variety, the lLers performed significantly better than the heritage BiTDs and heritage BiDLDs from all countries in terms of both varieties (see Table 8).

To determine how the scoring measure regarding the standard and heritage Turkish varieties affected the results of the heritage BiTD, BiDLD, and lL2 BiTD children, the results were analyzed again concerning the countries in which they lived at the time of testing. As can be seen in the within-group analyses in Figure 2 and

Table 9. Group means (SD) of correct responses and error types according to the countries where the groups of bilingual children lived. BiTD = bilingual children with typical language development; BiDLD = bilingual children with developmental language disorder; lL2_BiTD = late successive bilingual children with typical language development.

Country	Group	%Target-like Case Clauses
		Reference Turkish Variety
		Standard Turkish	Standard or Heritage Turkish
Germany	Heritage_ BİTD	80.04	87.95
		(17.94)	(20.76)
	Heritage_ BİDLD	67.79	84.28
		(23.68)	(14.89)
	IL2_BİTD	97.57	99.27
		(2.99)	(1.54)
France	Heritage_ BİTD	87.39	88.68
		(10.37)	(25.38)
	Heritage_ BİDLD	72.80	88.48
		(23.80)	(12.11)
U.S.	Heritage_ BİTD	83.55	86.63
		(14.30)	(13.03)

, the heritage BiTD and BiDLD children living in Germany scored significantly lower in standard Turkish than in heritage Turkish (sequentially: Friedman: [X²(2) = 65.803, p = 0.000], Wilcoxon test: Z = −4.166, p = 0.000, η² = 0.457; Friedman: [X²(2) = 21.574, p = 0.000, Wilcoxon test: Z = −2.936, p = 0.003, η² = 0.718). Interestingly, the rate of successful case and possessive marker use in the lL2ers who lived in Germany also significantly increased when the alternative scoring measure was adopted (Friedman: [X²(2) = 18.571, p = 0.000], Wilcoxon test: Z = −2.023, p = 0.409, η² = 0.409). The heritage BiTD and heritage BiDLD children living in France at the time of testing showed improvement when considering heritage Turkish features (sequentially: Friedman: [X²(2) = 21.294, p = 0.000], Wilcoxon test: Z = −1.599, p = 0.110, η² = 0.183; Friedman: [X²(2) = 7.6, p = 0.000], Wilcoxon test: Z = −1.604, p = 0.109, η² = 0.307). The case was similar for the heritage BiTD children living in the U.S. (Friedman: [X²(2) =9.294, p = 0.010], Wilcoxon test: Z = −1.342, p = 0.180, η² = 0.360). Generally, the heritage BiDLD groups in Germany and France significantly benefitted from the consideration of heritage Turkish.

The Kruskal–Wallis H test results showed that there was a significant difference among all groups regarding standard Turkish (Kruskal–Wallis test: standard Turkish [X²(5, N = 83) = 23.194, p = 0.000]; heritage Turkish [X²(5, N = 83) =15.114, p = 0.010]). According to the pairwise comparison results (see

Table 10. The effect of (clinical) group membership on the rate of correct case attempt in the TEDIL (Kruskal–Wallis test) and post hoc pairwise comparisons (Mann–Whitney U tests); significant comparisons are given in bold (regarding countries in which the groups of bilingual children lived). BiTD = bilingual children with typical language development; BiDLD = bilingual children with developmental language disorder; lL2_BiTD = late successive bilingual children with typical language development; G = Germany; F = France; US = U.S.

Group	%Target-like Case Clauses
	Reference Turkish Variety
	Standard Turkish	Standard or Heritage Turkish
	χ2(5, N = 83) = 23.194 p < 0.001	χ2(5, N = 83) = 15.114 p < 0.001
Heritage_BİTD_G vs. Heritage_BİDLD_G	U = 147.0	U = 164.0
	p = 0.07	p = 0.140
	r = 0.041	r = 0.026
Heritage_BİTD_G vs. IL2_BİTD_G	U = 41.0	U = 76.5
	p < 0.05	p < 0.05
	r = 0.379	r = 0.236
Heritage_BİTD_G vs. Heritage_BİTD_F	U = 194.5	U = 238.5
	p = 0.14	p = 0.573
	r = 0.042	r = 0.006
Heritage_BİTD_G vs. Heritage_BİDLD_F	U = 58.0	U = 63.0
	p = 0.468	p = 0.572
	r = 0.014	r = 0.008
Heritage_BİTD_G vs. Heritage_BİTD_U.S.	U = 89.0	U = 75.0
	p = 0.840	p = 0.472
	r = 0.001	r = 0.014
Heritage_BİDLD_G vs. IL2_BİTD_G	U = 0.00	U = 18.0
	p < 0.05	p < 0.05
	r = 0.720	r = 0.398
Heritage_BİDLD_G vs. Heritage_BİTD_F	U = 31.0	U = 50.0
	p < 0.05	p = 0.085
	r = 0.286	r = 0.118
Heritage_BİDLD_G vs. Heritage_BİDLD_F	U = 23.5	U = 19.5
	p = 0.953	p = 0.599
	r = 0.000	r = 0.019
Heritage_BİDLD_G vs. Heritage_BİTD_U.S.	U = 18.0	U = 28.5
	p = 0.234	p = 0.879
	r = 0.094	r = 0.001
IL2_BİTD_G vs. Heritage_BİDLD_F	U = 16.5	U = 16.0
	p < 0.05	p < 0.05
	r = 0.415	r = 0.449
IL2_BİTD_G vs. Heritage_BİDLD_F	U = 7.5	U = 6.0
	p = 0.076	p = 0.054
	r = 0.242	r = 0.380
IL2_BİTD_G vs. Heritage_BİTD_U.S.	U = 9.5	U = 6.0
	p = 0.055	p < 0.05
	r = 0.256	r = 0.459
Heritage_BİTD_F vs. Heritage_BİDLD_F	U = 17.5	U = 17.0
	p = 0.277	p = 0.242
	r = 0.069	r = 0.076
Heritage_BİTD_F vs. Heritage_BİTD_U.S.	U = 30.5	U = 26.0
	p = 0.687	p = 0.444
	r = 0.009	r = 0.037
Heritage_BİDLD_F vs. Heritage_BİTD_U.S.	U = 6.5	U = 9.5
	p = 0.413	p = 0.905
	r = 0.082	r = 0.002

), generally, the lL2 BiTDs performed significantly better than their heritage BiTD and BiDLD peers in Germany, France, and the U.S., regardless of which Turkish variety was used as the scoring measure (see Table 10).

With respect to error types, a common error pattern in the child heritage BiTDs and BiDLD, especially those in Germany, was the substitution of the accusative with the dative. Concerning the performance of the lLer BiTD children, the analyses showed no significant difference between them. Overextension of the ablative case marker is a phenomenon frequently found in heritage Turkish spoken in Germany (see Boeschoten, 1994; Chilla & Şan, 2017; Chilla, 2022). The heritage BiTD and BiSLI children in Germany used overextended ablative case markers, whereas the heritage children with and without DLD in France seldom used them. No significant difference was observed between the heritage BiTD and heritage BiDLD children in Germany regarding dialect leveling using instrumental markers (-lEn instead of -lE “with”). Although not as frequently as the heritage groups in Germany, the heritage children with and without DLD also used the dialectal version of the instrumental case. As in the study by Blom et al. (2022), the genitive–possessive constructions of the heritage children included other error types. Although the heritage children with and without DLD in Germany omitted the possessive suffix, the BiTD children omitted this suffix in more complex sentences. In contrast, the other group with DLD omitted it in mono-clauses or sentential fragments. This was also observed in the results of the BiTD and BiDLD children in France and the U.S., but not as frequently as in the results of those in Germany. In terms of the redundant use of possessive suffixes, originally typical in the early stages of Turkish language acquisition and exhibited by monolingual Turkish-speaking children up until primary school age in Turkey (Küppers et al., 2015), it was observed that the heritage BiTD children in Germany used them more than the other heritage BiTD and BiDLD and lL2er children in all countries. In comparison to the lL2 BiTD children, the heritage children with and without DLD in Germany omitted the genitive case marker more than the other groups in France and the U.S. In particular, in the task where the children were expected to add more than one genitive marker to different nouns in the genitive–possessive construction, they omitted the second and third genitive cases.

4.2. Factors Predicting Performance on Case and Possessive Markers

As mentioned in the Methodology section, we carried out hierarchical linear regression analyses and simple/bivariate regression analyses by using the background variables strongly correlated with performance on the use of case and possessive markers as potential predictors. Note that the regression analyses were limited to the BiTD (heritage BiTDs and lL2 BiTDs) and BiDLD groups in Germany and France because the size of the BiTD group in the U.S. was very small. The Spearman partial correlation analysis in Table 7 shows that there was a significant relationship (p < 0.05) between the percentages of case and possessive markers using both measures (case and possessive markers in standard Turkish or either in standard or heritage Turkish) and the quantitative exposure variables AoO_L2, LoE_L2, early exposure L1, current L1 use at home, current L1 skills, and current L1 use as in

Table 11. Partial correlations controlling for age between case and possessive markers in the TEDİL and PaBiQ variables related to bilingualism (regarding the countries in which the groups of bilingual children lived).

Background Variables (PaBiQ)	%Target-like Case Markers and Possessive Marker
	Reference Turkish Variety
	Standard Turkish	Standard or Heritage Turkish
L2 AoO	0.430 *	0.429 *
L2 LoE	−0.457 *	−0.382 *
% Early L1 Exposure	0.256 *	0.198
Current L1 richness (/14)	0.268 *	0.247 *
Current L1 Skills (/15)	0.352 *	0.331 *
Current L1 use (/16)	0.271 *	0.119

Notes: * p < 0.05.

. However, it was observed that the correlation between them was low (p < 0.50).

To gain a comprehensive understanding of which background variables can be used to predict the performance of each group in each country on case and possessive markers, we conducted Spearman partial correlational analyses again (see

Table 12. Partial correlations controlling for age between case and possessive markers in the TEDİL and PaBiQ variables related to bilingualism. BiTD = bilingual children with typical language development; BiDLD = bilingual children with developmental language disorder; lL2_BiTD = late successive bilingual children with typical language development.

Country	Group	Background Variables (PaBiQ)	%Target-like Case Markers and Possessive Marker
			Reference Turkish Variety
			Standard Turkish	Standard or Heritage Turkish
Germany	Heritage_ BITD	L2 AoO	0.282	0.229
		L2 LoE	−0.362	−0.262
		% Early L1 Exposure	0.075	0.140
		Current L1 richness (/14)	−0.209	−0.106
		Current L1 Skills (/15)	0.144	0.175
		Current L1 use (/16)	0.215	−0.087
	Heritage_ BIDLD	L2 AoO	0.229	0.231
		L2 LoE	0.295	0.410
		% Early L1 Exposure	0.007	−0.142
		Current L1 richness (/14)	−0.182	0.014
		Current L1 Skills (/15)	−0.530	−0.053
		Current L1 use (/16)	−0.185	0.182
	IL2_BITD	L2 AoO	0.643 *	0.646 *
		L2 LoE	−0.597	−0.182
		% Early L1 Exposure	−0.119	0.160
		Current L1 richness (/14)	0.085	0.158
		Current L1 Skills (/15)	−0.007	0.005
		Current L1 use (/16)	−0.866 *	−0.557
France	Heritage_ BITD	L2 AoO	0.537 *	0.698 *
		L2 LoE	−0.461	−0.741 *
		% Early L1 Exposure	0.294	−0.027
		Current L1 richness (/14)	0.303	0.470
		Current L1 Skills (/15)	−0.093	−0.072
		Current L1 use (/16)	0.151	0.298
	Heritage_ BIDLD	L2 AoO	0.258	0.258
		L2 LoE	-	-
		% Early L1 Exposure	0.632	0.632
		Current L1 richness (/14)	-	-
		Current L1 Skills (/15)	-	-
		Current L1 use (/16)	0.105	0.105
U.S.	Heritage_ BITD	L2 AoO	0.975 *	0.975 *
		L2 LoE	−0.900 *	-
		% Early L1 Exposure	−0.564	−0.564
		Current L1 richness (/14)	0.866	0.866
		Current L1 Skills (/15)	0.667	0.564
		Current L1 use (/16)	0.447	0.224

Notes: * p < 0.05.

). There was no statistically significant difference (p > 0.05) between the background variables and performance on case and possessive markers regarding standard or heritage Turkish variables for the heritage BiTD, BiDLD, and lL2 BTD children in Germany. However, concerning the correct responses of the lLer BiTDs in case and possessive markers in standard Turkish, current L1 use emerged as the leading predictor, explaining 54% (β = −0.735, t = −3.067, p = 0.015) (β = 0.720, t = 2.391, p = 0.019), with the age of onset of the L2 being another predictor, explaining 52% (β = 0.720, t = 2.391, p = 0.019).

Concerning the rate of successful case and possessive marker use of the heritage Bi-TD children in France, significant strong correlations (p < 0.05) emerged between the length of exposure to L2 and standard Turkish. A further significant strong correlation (p < 0.05) was found between the age of onset of the L2 and both scoring measures (standard Turkish and either standard or heritage Turkish). Regarding the BiDLD children in France, the correlational results suggested no statistically significant difference (p > 0.05) between the background variables and performance on case and possessive markers regarding the standard Turkish or heritage Turkish variables. As the results were not significant, a regression model was not built.

5. Discussion and Conclusions

Extensive research (Rothweiler et al., 2010; De Jong et al., 2010; Scherger, 2019; Gökgöz et al., 2020; Werthmann, 2020; Chilla, 2022; Blom et al., 2022, for Italian–German; for German–Russian; Schwartz et al., 2015, among others) has shown that case markers are vulnerable in bilingual children with DLD but robust in typical bilingual language development. Hence, they are assumed to be potential clinical markers for the identification of DLD in bilingual contexts; however, this may not apply in heritage contexts where case markers are typically acquired under reduced HL input conditions, especially in contexts where some of the heritage L1 varieties change as a result of contact with the majority/societal language (L2), as in the case of heritage Turkish in Germany.

Even though several studies (Çavuş, 2009; De Jong et al., 2010; Rothweiler et al., 2010; Schroeder & Dollnick, 2013; Chilla & Şan, 2017; Gökgöz et al., 2020; Blom et al., 2022; Chilla, 2022; Özsoy et al., 2023, among others) have investigated case markers in Turkish in the heritage context, few have compared heritage children with TD and heritage children with DLD (see Blom et al., 2022). As case marking is considered a domain for identifying language impairment in bilingual children (Scherger, 2019), the purpose of this study was to explore whether the use of case and possessive markers as part of genitive–possessive constructions can allow us to distinguish between BiTD and BiSLI in Germany, France, and the U.S, with lL2ers serving as a control group. This study sought to determine whether there would be an improvement in the performance of heritage BiTD children in the use of case and possessive markers when an alternative scoring method was used considering features of the heritage Turkish variety in Germany, France, and the U.S. Accordingly, all the case and possessive markers in a corpus of elicited production using the TEDİL were examined. In order to score the responses of the participants, we used two scoring measures: correct responses with reference to standard Turkish and correct responses with reference to either standard Turkish or heritage Turkish. The underlying reason behind the latter scoring method was not to penalize bilingual children for using the characteristics of the heritage variety of Turkish. Non-target responses in either variety were investigated to identify the potential overlaps between DLD markers in monolingual Turkish DLD and the features of heritage Turkish. Null responses were counted as errors unless they arose due to technical issues or experimental errors. Additionally, we investigated which age and input variables influenced the performance on case and possessive markers in the heritage child BiTD groups. Therefore, we examined background variables related to bilingualism that were strongly correlated with performance on case and possessive markers as potential predictors in the regression models.

The first research question of this study asked how heritage children with BiDLD compare to their heritage BiTD and lL2 BiTD peers in their production of standard Turkish case and possessive markers in the TEDİL. This study examined the rate of the successful case and possessive marker production of bilingual children in the TEDİL. Concerning case and possessive markers produced in the TEDİL, the lL2 BiTDs significantly outperformed all their BiTD and BiDLD peers regardless of which country they lived in when the scorings methods considered standard Turkish or heritage Turkish as the baseline, even if the BiTDs generally scored significantly better than the BiDLDs (see Figure 1 and Table 7 and Table 8). This result is consistent with the results of previous studies (Rothweiler et al., 2010; Blom et al., 2022), in which it was found that bilingual DLD children produced either fewer case markers or made errors, such as substitution or omission. All groups of lL2 BiTDs, heritage child BiTDs, and BiDLDs were divided in accordance with the countries in which they lived, and the results of pairwise comparisons suggested that the lL2ers who had been living in Germany for several years at the time of testing significantly outperformed all BiTD and BiSLI groups in the three different countries (p = 0.05) (see Figure 2 and Table 9 and Table 10). Although the performance of all heritage child BiTDs and BiDLDs did not significantly differ across countries, their performance on the case and possessive markers showed improvement when using the alternative scoring method that took heritage Turkish into account. The heritage bilingual children with DLD in Germany and France benefitted more from this method than the others (see Figure 2 and Table 9). The results of the regression analyses suggest that the leading predictor of the performance of the lLers BiTDs in case and possessive markers was current L1 use, explaining 54% of the variance when the standard Turkish variety was considered. Another predictor of their performance was the age of onset of German as the L2, explaining 52% of the variance in the rate of successful case and possessive marker use in standard Turkish. Although the late successive bilingual children have contact with the heritage Turkish variety and German as their L2 and their access to literacy-boosting programs and schooling is very limited in the heritage context (Rothman, 2009; Schroeder, 2020), they acquired Turkish in a monolingual environment before they emigrated to Germany, and their LoE was lower than that of their heritage peers with and without DLD. Thus, their exposure to standard Turkish for an extended period of time positively affected their performance on case and possessive markers when the standard Turkish variety was used as the baseline.

Qualitative error analyses of the BiTD and BiDLD children, especially those in Germany, showed that a common pattern between them is the substitution of the accusative case marker with the dative case marker. As stated in Blom et al.’s (2022) study, this may not be a clinical marker for identifying DLD in heritage children, as it may be one of the characteristics of heritage Turkish. Overextended use of ablatives, which was pointed out as another phenomenon of heritage Turkish in previous studies (Chilla & Şan, 2017; Chilla, 2022), was used by the BiTD and BiDLD groups in Germany more frequently than by the other groups. The use of -lEn, which is the dialectal variety of the instrumental case marker ile “with” (Boeschoten, 2000; Şimşek & Schroeder, 2011), was observed in the heritage child BiTDs and BiDLDs in Germany and France, although the frequency of its use was higher in Germany than in France. Even though language use at home in L1 was not found to be a predictor in this study, it may be assumed that family style and generation affect language output (Chen & Shirai, 2010). Additionally, it may influence the use of the instrumental case ile “with” in a heritage context. As the parents of lL2er children in Germany and heritage children in the U.S. and some parents of heritage children in France and Germany are considered to be the first generation, these children are expected to receive linguistic input in their L1 in standard Turkish. Although the heritage families in Germany are more heterogeneous in terms of their preference for their family language (Küppers et al., 2015) than those in other countries, it is true for all families that holidays in Turkey, exposure to Turkish media, life partners from Turkey, and informal everyday interactions in private and general domains (see Bezcioglu-Göktolga & Yagmur, 2018) support the maintenance of standard Turkish. However, it must be noted that heritage speakers access the structures of formal registers of Turkish very rarely (Schroeder, 2016), regardless of the generation that they belong to; they use stabilized structures of heritage Turkish. In parallel to the study by Blom et al. (2022), some errors were observed in the production of genitive–possessive constructions, including genitive and possessive markers in the utterances of the BiTD and BiDLD children in all countries. The redundant use of possessive markers, originally typical in the early stages of Turkish language acquisition and exhibited by monolingual Turkish-speaking children up until primary school age in Turkey (Küppers et al., 2015), was observed even in the utterances of the heritage BiTD school children in Germany in comparison with the other groups. Although the early acquisition of case markers results in high accuracy in their production, even in the heritage context, the stabilization of possessive markers may persist, as seen in their redundant use. Apart from this, the reason why the omission of possessive markers may be a potential DLD marker is that, whereas BiTD children omit it in more complex sentences, BiDLD children omit it in mono-clauses and sentential fragments. In terms of genitive case use, the BiTD children in Germany in particular omitted this marker in the second or third noun (e.g., bebek-in elbise-si-Ø kol-u “arm of baby’s cloth” instead of bebek-in elbise-si-nin kol-u) if there was more than one genitive marker in the genitive–possessive construction. The heritage DLD children produced sentential fragments by producing only some of the nouns in this construction (e.g., bebek-Ø kol-Ø “baby arm” instead of bebek-in elbise-si-nin kol-u). The omission of a genitive marker may also be considered a DLD marker in a heritage context.

It should be noted that this study has its limitations. Firstly, it has small sample sizes, with the risk of excluding important population characteristics. The lack of standardized language assessment tools for heritage Turkish is another limitation because the low scores of bilinguals did not only result from having DLD but also from L1 acquisition in the heritage context. Despite the limitations, this study has implications for assessing case markers in L1 in a heritage context and the influence of background variables on the acquisition of heritage language. If heritage Turkish is used as a baseline in the systematic analysis of the corpus of case and possessive markers in the utterances produced in the TEDİL, heritage child BiTDs and BiDLDs may be differentiated. To avoid misdiagnosis in clinical settings (Herkenrath & Rehbein, 2012; Chilla, 2022), the varieties of Turkish must be analyzed systematically because DLD markers and heritage Turkish features may overlap. For the same purpose, heritage children should be tested with combinations of linguistically controlled L1 tasks, such as LITMUS-SRT and LITMUS-NWRT, which sensibly allow for a fast estimate of typical language development (Abed Ibrahim, 2025; Hamann et al., 2020; Hertel et al., 2022). The fact that late successive bilinguals have contact with an L2 and heritage Turkish may influence their performance in terms of their linguistic skills. This may be a further point and must be addressed in future studies. Furthermore, heritage children may have different levels of literacy due to living in different countries, which may result in varying L1 performances.

Overall, the results demonstrate that case and possessive morphology are vulnerable in heritage languages. In the case of BiDLDs, fewer case markers and higher rates of omissions and substitution errors, particularly in the accusative, the dative, and markers, were observed in this study. Supporting the results of Blom et al.’s (2022) study, the omission of possessive and genitive markers in simpler structures may serve as a clinical marker of DLD. The overextension of the ablative case marker and the dialectal version of the instrumental marker (-lEn) appear to be features of heritage Turkish, as shown in previous studies.

To summarize, these results are essential for clinical practitioners and educators with limited language assessment resources in a heritage context. Moreover, parental questionnaires, such as the PABIQ (Tuller, 2015), can be used to collect background information about heritage children’s language history, which may improve bilingual children’s performance on language tests (Hertel et al., 2022). Additionally, the identification of DLD in the heritage context will improve if a sensitive error analysis regarding the qualitative and quantitative differences between heritage children with and without DLD is carried out (Chilla, 2022; Şan, 2023). Lastly, considering the high numbers of Turkish children who receive speech and language therapy services in the countries in which heritage Turkish is spoken, it is necessary and crucial to develop reliable language assessment tools normed for bilingual Turkish populations in different heritage contexts.