Language experience in LSF development: Behavioral evidence from a sentence repetition task.

In psycholinguistics and clinical linguistics, the Sentence Repetition Task (SRT) is known to be a valuable tool to screen general language abilities in both spoken and signed languages. This task enables users to reliably and quickly assess linguistic abilities at different levels of linguistic analysis such as phonology, morphology, lexicon, and syntax. To evaluate sign language proficiency in deaf children using French Sign Language (LSF), we designed a new SRT comprising 20 LSF sentences. The task was administered to a cohort of 62 children- 34 native signers (6;09-12 years) and 28 non-native signers (6;08-12;08 years)-in order to study their general linguistic development as a function of age of sign language acquisition (AOA) and chronological age (CA). Previously, a group of 10 adult native signers was also evaluated with this task. As expected, our results showed a significant effect of AOA, indicating that the native signers repeated more signs and were more accurate than non-native signers. A similar pattern of results was found for CA. Furthermore, native signers made fewer phonological errors (i.e., handshape, movement, and location) than non-native signers. Finally, as shown in previous sign language studies, handshape and movement proved to be the most difficult parameters to master regardless of AOA and CA. Taken together, our findings support the assumption that AOA is a crucial factor in the development of phonological skills regardless of language modality (spoken vs. signed). This study thus constitutes a first step toward a theoretical description of the developmental trajectory in LSF, a hitherto understudied language.

Introduction

Assessing language abilities in deaf people is a real challenge as their language mastery varies widely depending on factors such as their environment, health and socioeducational level. A direct consequence of this situation is that professionals such as teachers, clinicians, and researchers have to assess language skills almost case by case basis. The challenge is even greater since an assessment must satisfy the needs of each professional who works with Sign Languages (SL), not only measuring the individual’s general proficiency in sign language but also providing a specific focus corresponding to the professional’s needs (e.g., phonological development, proficiency level of second language learners of sign language, modeling sign language acquisition).

Since the 1970s, the Sentence Repetition Task (SRT) has frequently been used in spoken language assessment [1,2], and has proven efficient for measuring children’s linguistic abilities in various populations (e.g., native speakers, second-language learners, bilinguals, children, adults with language disorders, socioeconomically disadvantaged individuals). The SRT is also known to be a reliable clinical marker of language impairment [3-10]. Concretely, the task involves repeating a sentence produced by a speaker (experimenter, teacher, etc.). The participant’s repetition (or recall) should be as similar as possible to the sentence produced by the speaker. Once the sentence has been recalled, error types can be analyzed, thereby providing information about the level of language analysis that may disturbed. These errors are divided into phonological errors, omissions, substitutions, and deletions. As in spoken languages, errors in an SRT conducted in a sign language also constitute a reliable indicator of normal and delayed acquisition at linguistic levels of analysis (phonology, morphology, syntax, lexicon). Therefore, the SRT is viewed as a valuable screening tool for language disabilities. Moreover, it has several advantages: it is quick and easy to administer; it tests explicit, precisely specified linguistic structures; and it is not too time-consuming for professionals who have to assess and score in addition to teaching or providing therapy.

Although the SRT has proven to be a good tool for detecting delays or disorders in language development, no SRT study had previously been done on French Sign Language (LSF) development in deaf children. To fill this gap, we designed a behavioral experiment in LSF using this task with deaf children. Our goal was to study language development in LSF at different levels of analysis (phonological and morphosyntactic development, lexical knowledge) as a function of age of SL acquisition (AOA) and chronological age (CA) using an SRT. This study constitutes an innovative contribution to the sign language literature as it is the first attempt to investigate language development in LSF with a large cohort (N = 62) of deaf children. In addition to this developmental issue, our study introduces a new task to assess linguistic knowledge in children in an understudied language, namely LSF, the language used by deaf people in France. As in all sign languages that adapt to the specific visuo-gestural modality, LSF signers simultaneously use manual and non-manual articulators (hands, arms, chest, face, gaze) to produce sublexical and lexical signed units, and locate these signs in the signing space to set up the spatial grammar, that is, organize the syntactic relations between signs. The linguistic analysis of LSF is relatively young, and few grammars have been published (for a review paper, see [11]).

The SRT in sign languages

To date, only a few studies have been carried out on sentence repetition ability in sign languages with deaf people [10,12-14]. Most such studies aimed to between distinguish deaf adults who had and had not acquired sign language as a native language and showed that the SRT constitutes a valuable tool to highlight error rates that vary according to signer group [13,15]. These inter-group distinctions are important because they show how the lack of early exposure to sign language can influence language development and the mastery of complex structures. Moreover, because the SRT involves both receptive and expressive skills, it is also known to be a good proxy for general linguistic abilities.

Mayberry and Fischer [16] ran the first SRT study with American Sign Language (ASL) adult signers, whose AOA varied. The authors observed that AOA had a strong impact on repetition skills, with the native signers recalling more sentences more accurately than the less sign-exposed deaf adults. In addition, the types of errors differed according to AOA: while native signers produced more semantic errors, non-native signers produced more phonological errors, as they paid more attention to the surface structure (phonological shape) of the stimuli. Despite long experience (around 15 years) with sign language, non-native signers underperformed in recall accuracy and exhibited less accurate lexical and phonological skills, indicating the great importance of native exposure. Hauser et al. [15] used an SRT with an ASL sample to provide a test of ASL proficiency that would allow researchers to collect a full range of data from the heterogeneous community of deaf signers. The task comprised 40 ASL sentences varying in length and morphosyntactic complexity. Sentences were repeated by adolescents (12.5 to 14.1 years) and adults who were ASL natives or non-natives. The authors observed a strong effect of developmental age and of ASL AOA on general SRT scores. After this pilot study, they chose the 20 items that were most sensitive to native proficiency to build the final SRT version. This final task was used in Supalla et al.’s [14] study. Once again, AOA and length of exposure were found to influenced repetition performance. The most striking results concerned the types of errors and the strategies used when repeating. While fluent signers tended to preserve the semantics of the sentence, less fluent signers tended instead to repeat the surface structure. Cormier et al. [13] adapted the ASL-SRT to British Sign Language (BSL) and administered it to a group of 20 adults composed of 10 native signers, 5 early learners, and 5 non-native learners of BSL. The authors observed that the earlier the AOA, the more accurate the repetition skills. Marshall et al. [10] used the same task to assess repetition abilities in deaf children with Specific Language Impairment (SLI) in comparison to deaf controls who were non-native learners of BSL. They found that deaf children with SLI presented poorer repetition skills, were less accurate in sentence repetition (modifying the meaning or the order of signs), frequently omitted spatial morphology markers, and rarely used facial expressions. An interesting point was the pattern of errors, which was very similar to that observed with hearing children with SLI.

Recently, Rinaldi et al. [12] reported the results of an SRT with native and non-native deaf children and adults, who had all received intensive exposure to Italian Sign Language. They observed effects of both chronological development and AOA on repetition skills, and emphasized the difficulty the children had in acquiring and mastering the non-manual components.

Does the SRT assess only language skills?

There is a lively debate concerning what abilities are needed to recall a sentence and what kind of capacities a repetition task assesses, and in particular whether memory or language processes are most involved in an SRT. Several researchers contend that memory skills underlie language skills and have demonstrated the role of working memory in language skills, in both neurotypical people and populations with disabilities [6,17-20]. Acheson and MacDonald [21,22] argued that linguistic knowledge and memory are intertwined, and that working memory cannot be separated from language comprehension and production. Along the same lines, Vargo and Black [23] ran an SRT with aphasic adults whose recall skills were well preserved. Their results showed that memory ability is involved but only slightly. The same results were obtained with children with SLI [24]. Devescovi and Caselli [25] investigated the relation between the SRT and working memory in preschool children and reported contradictory results; they argued that, even if the SRT partly relies on memory, strong linguistic abilities are necessary to recall sentences.

Several arguments have been proposed to show that SRT ability depends as much on language as on memory. The first concerns the evidence that all language levels are involved in the repetition process [26,27]. As soon as a speaker hears/sees the sentence to be repeated, he/she builds a conceptual representation. This conceptual representation relies on several memory and linguistic processes to recall the sentence: sensorimotor processes (speech-sign perception and speech-sign production), phonological representations, lexical knowledge and grammatical encoding. Another argument in favor of the strong involvement of language skills concerns the complexity of the structures assessed: if the sentence reaches a certain length or structural complexity threshold, mere imitation is not sufficient, and robust linguistic representations are crucial [28]. In addition, some researchers have pointed out, repetition ability also depends on the speaker or signer’s familiarity with the assessed language. Repetition of a two-sign sentence is easy for young signer children, non-fluent signers, and even non-signers, because repetition can be based on mimicking gestural components and not just on linguistic components. But repeating a syntactically complex sentence in sign language requires robust linguistic representations and good phonological and working memory abilities [2,10,21].

Natalicio [29] raised other critical issues regarding the naturalness of the repetition task. This task is often criticized for not being able to assess true language skills since repetition is ecologically irrelevant. By merely repeating a sentence, speakers do not experience a spontaneous interaction. To rehabilitate this task, Natalicio pointed out that most language-testing situations in a laboratory are not ecological, due to the presence of the experimenter, the recording equipment, the place of testing, the response and time constraints, and all the other variables that are absent in spontaneous interactions. Whatever the task, testing is not a natural event, but the linguistic behaviors identified by assessment still allow researchers to highlight a real language competence.

Another interesting point that we take into account here concerns the use of the gestural modality and the impact in terms of sensorimotor processing: cognitive abilities are very dependent on the gestural modality. Boutla et al. [30] showed that deaf adult signers have a shorter phonological memory span than hearing adults. This could be the consequence of the inherent processing abilities of SL: because they use a visual system, signers are likely more efficient at retaining spatial information than temporal information. However, it is temporality that is most frequently measured in phonological short-term memory trials, even when the assessed language is an SL. We can extend this observation to linguistic skills and we must be extremely careful not to confuse SL characteristics that may influence deaf signers’ abilities with those abilities themselves. However, it should be noted that, as soon as linguistic structures or world knowledge become available, equivalent memory skills are expected irrespective of the language modality.

Native sign language acquisition

In the current state of knowledge, it is assumed that typical sign language acquisition and development follow a similar trajectory to those of a spoken language [31-34]. When a child benefits from full exposure to an SL from birth, this is referred to as native acquisition. Several psycholinguistic studies have reported that signing children of deaf parents (i.e., native signers) show a linguistic development course similar to that of hearing children of hearing parents. For example, a lexical burst is observed around 16 to 20 months of age in SL development [35]. SL development starts at around 6 to 9 months with manual babbling, and production of the first signs is characterized by sign simplifications, substitutions, and reduplications depending on the children’s motor limitations and the phonotactic constraints of their SL [36-38]. Like their speech-exposed counterparts, native signers produce their first signs around 10 to 12 months and also seem to organize their lexicon around semantic categories [35,39-41].

Regarding the acquisition of syntax, it is difficult to compare speech- and sign-exposed children because the SL-specific spatial syntax is typologically distinct from spoken languages. Mastery of spatial components in SL syntax that have no vocal counterparts makes it difficult to compare acquisition in the two modalities [42,43]. Furthermore, some linguistic specificities related to the visuo-gestural modality have been reported in SL development, such as strong eye contact in infancy, manual babbling, absence of simultaneous production of gestures and signs (i.e., language with gesture), and a higher rate of production of predicates.

While the native acquisition of different SL is relatively well understood thanks to a rich empirical literature that is widely disseminated in the SL scientific community, to our knowledge, no study has been done on the developmental stages of LSF acquisition. Some research has been carried out on the acquisition of specific abilities such as pointing functions in an LSF signer deaf child [44], referential processes in children’s narrations [45], LSF-French child bilingualism [46], gesture-sign development in a longitudinal study of four children [47], and morphosyntactic skills in children [37].

Consequences of non-native sign language acquisition for language abilities

Native exposure to SL is far from the most common situation: 95% of deaf children are born to hearing parents who are not able to speak an SL with their child at birth and in the early years [48, for US statistics: 49], while others are born to deaf parents whose knowledge of sign language is incomplete or deficient [50-53]. This late exposure to sign language has consequences that are far from trivial, as language development is known to differ in native and non-native signers. While native signers, either deaf or hearing, with deaf parents grow up and acquire language in an efficient sign language environment, non-native signers usually encounter the gestural modality or sign language at school. Consequently, the linguistic environment has strong impacts on language acquisition and the linguistic skills observed in adulthood. As Cormier et al. [54] claimed, acquisition of a first language from birth will ensure native proficiency, whereas delayed first language acquisition prevents children from achieving complete acquisition. Therefore, most deaf children who learn SL experience language deprivation, the consequence of which is atypical language development.

We know now that lack of exposure to SL from birth may cause significant cognitive and linguistic delays [51,55-57]. Some studies have shown that deaf children exposed to SL late in childhood present unusual phonological development. For example, the first handshapes can be of different levels of complexity, and the production of these handshapes may depend on gestures that non-native learners produced before they were exposed to sign language [56,58]. In this study, it is specifically the impact of AOA on language development that we aim to examine. The same observation as for phonological development has been reported for the development of morphosyntactic structures [56,59,60]. In BSL sentences, Cormier et al. [54] used a grammaticality judgment task and found greater inaccuracy in non-native learners than in native ones; inaccuracy increased as a function of the age of the first exposure to BSL. Poor syntactic ability in sign language production was also reported by Ferjan-Ramirez et al. [51] in deaf teenagers with extremely late language acquisition (around 14 years), who therefore presented atypical sign language development, as indicated by a syntactic deficiency despite a large lexicon. The authors also noted that these teenagers learned signed vocabulary very quickly compared with younger native signers despite the same length of exposure. In contrast, their morphosyntactic skills were poor, with a childlike Mean Length of Utterance (MLU), around 2.08 signs for the most productive teenage signer aged 15;08 years [55]. This MLU is similar to that of a young child aged from 1;08 to 2;0 years [37].

In sum, most researchers report that early exposure to sign language provides benefits, even for second language (L2) literacy [60,61]. Early/native exposure should be distinguished from SL experience: even with 15 to 20 years of practice, non-native adults cannot behave like adults who are native signers [6,53].

Assessing sign language and the development of signed abilities: A long road to success

It is necessary to provide benchmarks for the developmental trajectory of SL ability. To succeed in this, we must be able to assess the developmental trajectory. Unfortunately, as mentioned above, there is a lack of assessment tools for sign language development, and in particular for LSF. While several tests have been created to assess different linguistic levels and populations, it is not yet possible to perfectly describe the linguistic development of signing deaf people in its all aspects ([62] and http://www.signlang-assessment.info). Several attempts have been made to create tests in LSF but none of them are systematically used because of their lack of reliability. For example, the TELSF (Test for French Sign Language [63,64]) is considered to be too long to run and extremely difficult to score; furthermore, no successful adaptation of the LSF Receptive Skills Test has been developed [65]. Consequently, there is no commonly accepted test to measure LSF performance and proficiency. These tests did, however, enable us to highlight the constraints and difficulties encountered when adapting a language test [66]. Creating a tool that measures sign language skills is not an easy task. First, gestural languages require some technical adaptations to their physical and structural properties, and it is crucial to consider technical issues such as the presentation of visual stimuli or the recording of participant responses. In addition, to build an SRT, sign language linguists need to collaborate with deaf native signers to select suitable linguistic items and record them. We must also ensure that the assessment tool is easy to administer and score for all the professionals who will use it (i.e., speech therapists, researchers, teachers and specialist teachers, sign language teachers) and in all test situations (i.e., school, speech therapy, L2 sign language courses). Finally, we need to provide a reliable test that can rank a child on a clear, accurate language developmental scale [40].

This study

This study examined language development at different levels of analysis (phonology, morphology, lexicon, syntax) in deaf children using French sign language, taking into account the AOA and CA, two determining factors in spoken language development. To assess sign language abilities in LSF-using children, we designed a first LSF Sentence Repetition Task, composed of 20 sentences. The SRT is known to assess linguistic abilities reliably and quickly. We administered our SRT to a cohort of 62 children comprising 34 native signers (6;09–12 years) and 28 non-native signers (6;08–12;08 years).

In this experimental study, in order to maintain some variety in our sentences, as in natural production, the children’s ability to repeat signed sentences was assessed with 20 LSF sentences varying in linguistic complexity (i.e., length [2 to 6 signs], type of morphosyntactic units [e.g., absence/presence of classifiers, mouth gestures]). In accordance with previous sign language studies on AOA and SRT, non-native signers were expected to achieve a lower performance level than native signers on our SRT, and to make more lexical-semantic and phonological errors. In addition, older children were expected to perform better than younger ones. Finally, we expected that, at the same age, native signers would outperform non-native signers.

Materials and method

Participants

Sixty-two deaf children (38 girls) aged from 6;01 to 12;09 years old (mean age = 8;08 years, SD = 1;06 years) participated in the study. They had varied experience in LSF and different AOA, length of exposure to LSF, and quantity and quality of LSF input. (Table 1 shows the biographical characteristics of the participants.) Thirty-four children were native signers (mean age = 8;75 years, SD = 1;06 years) who grew up in deaf families and had strong exposure to LSF from birth, while 28 children grew up in hearing families and were therefore considered non-native signers (mean age = 8;84 years, SD = 1;8 years). We conducted a t-test that failed to show a significant difference between the groups for CA (t(60) = 0.16, p > .05). Native and non-native signers received a bilingual education (written French and LSF), learned LSF at school, and used LSF as their preferred language. Note that all native children attended the same school, which ensured consistency with respect to the language environment/experience for all of them. According to the parental questionnaires and teacher reports, none of the children had other cognitive and/or social impairments. All children were given information about the goal of the study and informed that they could cease to participate in the experiment at any time if they wished. Before the experiment, their parents signed a consent form and were also informed about the purpose of the study and the exact nature of the experimental task. Before we ran the experiment with deaf children, 10 deaf adults who were all native signers of LSF were tested to ensure the validity of our SRT. The adults aged from 25 to 45 (mean age = 29;9 years; 3 males and 7 females) and grew up in different areas of France, ensuring real LSF expertise. Moreover, participants were matched for socioeconomic status. Before taking part in the study, the participants were individually informed about the experimental protocol and the data storage and anonymization procedure. They then gave their written consent. The data collected were anonymized by applying the European Data FAIR principle [67] in collaboration with HumaNum for the management of the experimental data. The study was approved by the local ethics committee of the Department of Psychology at Paris Nanterre University (SignMET. Project Number 543264-LLP-1-2013-1-IT-KA2-KA2MP) and was performed in accordance with the Declaration of Helsinki. The individuals appearing in the pictures in this article gave written informed consent (as outlined in the PLOS consent form) to publish these case details.

Table 1

Sociodemographic data on deaf children: 34 native signers and 28 non-native signers assessed with the SRT.

CA	AOA	Type of Hearing Aid	Length of exposure to LSF and / or French	Hearing Status of parents	Home Linguistic Environment
06;02	native	-	LSF family & LSF school	deaf—deaf	LSF
06;04	native	-	LSF family & LSF school	deaf—deaf	LSF
06;05	native	hearing device	LSF family & LSF school	deaf—hearing	LSF—Vocal French
06;09	native	-	LSF family & LSF school	deaf—deaf	LSF
07;00	native	-	LSF family & LSF school	deaf—deaf	LSF
07;01	native	-	LSF family & LSF school	deaf—deaf	LSF
07;04	native	-	LSF family & LSF school	deaf—deaf	LSF
07;04	native	-	LSF family & LSF school	deaf—deaf	LSF
07;05	native	-	LSF family & LSF school	deaf—deaf	LSF
07;06	native	-	LSF family & LSF school	deaf—deaf	LSF
07;07	native	-	LSF family & LSF school	deaf—deaf	LSF
07;10	native	CI	LSF—French family & LSF school	deaf—deaf	LSF
08;01	native	-	LSF family & LSF school	deaf—deaf	LSF
08;03	native	-	LSF family & LSF school	deaf—deaf	LSF
08;03	native	-	LSF family & LSF school	deaf—deaf	LSF
08;08	native	-	LSF family & LSF school	deaf—deaf	LSF
08;08	native	-	SL family & LSF school	deaf—deaf	Romanian SL *
08;10	native	-	LSF family & LSF school	deaf—deaf	LSF
08;11	native	-	LSF family & LSF school	deaf—deaf	LSF
09;00	native	-	LSF family & LSF school	deaf—deaf	LSF
09;01	native	-	LSF family & LSF school	deaf—deaf	LSF
09;05	native	-	LSF—GSL family & LSF school	deaf—deaf	Greek SL—LSF *
09;10	native	-	LSF family & LSF school	deaf—deaf	LSF
10;00	native	-	LSF family & LSF school	deaf—deaf	LSF
10;02	native	-	LSF family & LSF school	deaf—deaf	LSF
10;03	native	-	LSF family & LSF school	deaf—deaf	LSF
10;03	native	-	LSF family & LSF school	deaf—deaf	LSF
10;04	native	-	LSF family & LSF school	deaf—deaf	LSF
10;06	native	-	LSF family & LSF school	deaf—deaf	LSF
10;07	native	-	LSF family & LSF school	deaf—deaf	LSF
10;07	native	-	LSF family & LSF school	deaf—deaf	LSF
10;07	native	-	LSF family & LSF school	deaf—deaf	LSF
11;02	native	-	LSF family & LSF school	deaf—deaf	LSF
12;00	native	-	LSF family & LSF school	deaf—deaf	LSF
06;01	late	CI	kindergarten	hearing—hearing	French
06;06	late	CI	LSF non-dominant language	hearing—hearing	French
06;09	late	CI	LSF non-dominant language	hearing—hearing	French
06;09	late	CI	AOA 03;01—experience 03;08	hearing—hearing	French
06;10	late	-	LSF non-dominant language	hearing—hearing	French
07;00	late	hearing device	AOA 03;04—experience 03;08	hearing—hearing	French
07;01	late	CI	LSF dominant language	hearing—hearing	French
07;02	late	hearing device	not known	hearing—hearing	French & Berber
07;03	late	CI	not known	hearing—hearing	French
07;05	late	CI	LSF non-dominant language	hearing—hearing	French
08;00	late	2 CI	AOA 04;06—experience 03;06	hearing—hearing	French
08;00	late	-	kindergarden	hearing—hearing	French
08;04	late	hearing device	not known	hearing—hearing	French
08;11	late	CI	not known	hearing—hearing	French
09;01	late	CI	not known	hearing—hearing	French
09;07	late	CI	not known	hearing—hearing	French
09;08	late	-	not known	hearing—hearing	French
09;10	late	hearing device	LSF non-dominant language	hearing—hearing	French
09;11	late	CI	not known	hearing—hearing	French
09;11	late	post linguistic deaf	elementary school	hearing—hearing	French & Swiss German
10;01	late	bi-CI	LSF non-dominant language	hearing—hearing	French
10;05	late	CI	not known	hearing—hearing	French
10;08	late	-	kindergarden	hearing—hearing	French
10;08	late	CI	elementary school	hearing—hearing	Kurdish & French L2
10;09	late	-	not known	hearing—hearing	French
11;00	late	-	LSF non-dominant language	hearing—hearing	French
11;01	late	CI	not known	hearing—hearing	French
12;09	late	CI	not known	hearing—hearing	French

The children’s parents were asked to specify their hearing status (deaf or hearing), which language(s) they spoke or signed daily with their child, and what kind of exposure their child had to LSF and spoken French. This information allowed us to set up AOA groups.

* These children have LSF as the dominant language, as they have been educated in LSF-French bilingual school since the age of 3 years. In addition, deaf teachers assured us that these children are highly proficient in LSF. CA: chronological age; AOA: age of acquisition; CI: cochlear implant; LSF: French Sign Language.

Sentence repetition task in LSF: Stimuli

The SRT in LSF consisted of 20 sentences, which were selected from a pilot study in which two deaf adults assessed the initial pool of 35 LSF sentences based on their naturalness. Four criteria were taken into consideration in satisfying naturalness: (1) grammaticality, (2) plausibility, (3) age-matched semantic content, and (4) saliency of signs for display on screen. To create our 35 LSF sentences, we initially relied on Rinaldi et al.’s [12] corpus. We adapted the Italian version of the SRT thanks to deaf native signers in order to create a culturally and linguistically well-designed pool of LSF sentences. The sentences varied in length and syntactic complexity. (Fig 1 and S1–S4 Videos shows some examples; the complete materials are presented in S1 Table). Given that length is not the sole marker of structural complexity in sign language (as in spoken language), several short sentences included complex morphosyntactic structures such as (1) classifiers (a category of signs with a non-specific meaning, expressed by a particular handshape that specifies—classifies—a referent with a particular property, such as two-legged, vehicles, etc. [68]), (2) dual predications, and (3) use of the non-dominant hand to maintain reference in the signing space. Note that long sentences may contain only simple morphosyntactic structures.

Fig 1

Sentences with four levels of complexity.

Sentences of each level of syntactic complexity were presented in the Supporting Information (S1 Video: easy sentence; S2 Video: intermediate easy sentence; S3 Video: intermediate difficult sentence; S4 Video: complex sentence). All the sentences presented in the SRT are listed in the (S1 Table).

The lexical items chosen were those that any deaf child was likely to be familiar with. The sign rate was designed to be adapted to children’s perception and recognition, and facial expressions were present but not strongly emphasized. All the sentences were checked in our team composed of deaf speakers and deaf and hearing linguists. One LSF speaker signed the instructions that were given prior to the experiment as well as the 20 sentences constituting the SRT. Post-experiment interviews revealed that the task was not considered too easy or too difficult.

Procedure

The children were tested by deaf native signers or fluent hearing signers. Before starting the SRT, the experimenter engaged in a short interview with the children in order to make them feel comfortable for the following task. Testing was carried out in the school library. The SRT lasted only 10 minutes. The instructions were previously recorded and were presented on a laptop. The experimenter verified that the children had understood the task correctly, and if not, repeated the instructions a second time. The children were told to repeat verbatim each sentence of the LSF-SRT presented on the laptop. The children’s repetitions were video-recorded. Children were instructed to repeat in front of the camera to ensure that their repetition was properly recorded. They could ask to see the sentence a second time if necessary and make self-corrections.

Scoring

Fluent signers annotated all the data twice. All annotations were done in an Excel spreadsheet (Table 2). Because the task consisted in repeating exactly the same sentence as the one produced by the LSF speaker, we coded the presence or absence of linguistic variation between the model and the child’s repetition.

Table 2

Scoring grid for all signs.

GLOSS	Is the sign REPEATED?	If the sign is repeated, is it DIFFERENT from the model?	If the sign is DIFFERENT, which level is the difference located at?
CHILD (UL)	YES or NO	YES or NO	Linguistic criteria		Difference
			Repeated sign	Substitution _ regionalism*
				Substitution _ another sign
				Variant of the target sign
			UL—Manual Parameters	Handshape
				Movement
				Orientation
				Location
			CL Dominant Hand	CL-DH size
				Handshape
				Movement
				Orientation
				CL-DH not held (reference not maintained)
				Inflection
				Wrong CL-DH location and structure maintained
				Wrong CL-DH location and structure not maintained
			CL Non-Dominant Hand	CL-NDH size
				Handshape
				Movement
				Orientation
				CL-NDH not held (reference not maintained)
				Inflection
				Wrong CL-NDH location and structure maintained
				Wrong CL-NDH location and structure not maintained
			Laterality	Dominant Hand—Non-Dominant Hand relation
			Facial Expression	Lexical-semantic
				Grammatical + CL
			Eye gaze	Eye gaze
			Mouth actions	Mouthing
				Mouth gestures
			Chest	Lexical-semantic
				Grammatical

CL: classifier; DH: dominant hand; NDH: non-dominant hand.

* Regionalisms represented a marginal proportion of repetitions. Non-manual parameters (laterality, facial expression, eye gaze, mouth actions and chest) have been coded but have not been taken in account for this paper. See S2 Fig to take notice of the target sign CHILD.

The steps in our analysis were as follows: (1) Was a sign repeated or not? (2) If it was repeated, then was it different from the model? (3) If it was different from the model, then what kind of lexical variation was observed: (a) substitution_regionalism; (b) substitution_other sign; (c) variant of the target sign? (4) If the repeated sign was a variant of the target sign (called “phonological errors”), then what kind of manual parameter—that is, the phonological unit in LSF: handshape, movement, location, or orientation—was incorrectly reproduced?

As Table 2 shows, after viewing each sentence, the annotators reported for each sign in the sentence whether it was produced or not (YES or NO). If the sign was produced, coders reported whether the repetition of the sign was similar to or different from the model’s (YES, the sign is different, or NO, the sign is similar). If the sign produced was different, the coders reported the linguistic level at which the sign varied. We decided to code each formal parameter that composed a sign. For this purpose, we separated manual parameters (lexical signs, classifiers produced with the dominant hand or non-dominant hand) from non-manual ones (facial expression, eye gaze, mouth actions, and chest posture).

Once the formal analyses were completed, we analyzed the repetition, taking into consideration two linguistic levels of analysis: phonological and lexico-semantic. First of all, we noted that the children clearly had difficulties repeating the non-manual parameters. We therefore decided to ignore non-manual variations as several factors in the experimental situation and the video-recording may have affected repetition performance: the children had to look at the experimenter whereas the sentence structure required them to direct the eye gaze to the hand, they were too young to spot fine actions such as mouthing, they were too young to master space perfectly, etc. At the end of scoring, for each child, our matrix displayed four columns, including (1) the number of signs produced, (2) the number of lexical errors among the produced signs, (3) the number of phonological errors, and (4) the types of phonological errors. Because sign language annotation is time-consuming, we employed two coders to score the children’s repetition abilities. All the repetition material (62 recordings in all) was coded by two independent coders. Both coders were fluent in sign language and received the same annotation instructions with specific rules predetermined by the authors. The intercoder comparisons did not show reliable differences for any comparisons (Student t-test p > .05).

Results

Sign and sentence repetition

Two analyses of variance (ANOVAs) were run with AOA and CA as between-subject factors on the percentages of repeated signs (dependent variable 1, DV1) and of errors in sign repetition among the repeated signs (DV2). Moreover, the type of error (i.e., target sign, sign substitution and regionalism) was analyzed. The analyses run for each type of error were also conducted with AOA and CA as between-subject factors. Finally, additional analyses of repeated signs and errors among repeated signs were also run in deaf adults (control group). We first present the children’s results followed by those of the adult control group. A descriptive comparison of children’s and adults’ performance is shown in Fig 2.

Fig 2

Percentage of repeated signs and percentage of errors in repeated signs according to age of acquisition and chronological age.

Adults are considered as the control group (*** p < .001; ** p < .01; * p < .05; n.s. p > .05).

Percentage of repeated signs

The ANOVA revealed that the main effect of AOA was significant (F(1,56) = 17.96, p < .001, ηp2 = .240; Fig 2). This indicates that, on average, the native signers repeated more signs (94.1%, SD = 5.9%) than the non-native ones (84.4%, SD = 13.7%). Moreover, the main effect of CA was also significant (F(2,56) = 11.77, p < .001, ηp2 = .296). Planned comparisons revealed that on average the youngest children (6–7 years) repeated fewer signs (82.9%, SD = 15.2%) than the two older groups (for 8–9 and 10–12 years, respectively: 92.5%, SD = 6.33%, and 94.8%, SD = 4.56%; F(1,56) = 23.23, p < .001). In contrast, no significant difference was found between the two older groups (F < 1). The AOA × CA interaction was not significant (F(2,56) = 1.96, p > .10).

Percentage of errors in repeated signs

The ANOVA revealed a significant main effect of AOA (F(1,56) = 29.18, p < .001, ηp2 = .343, Fig 2). This effect indicates that, on average, native signers were more accurate (34.5%, SD = 11.6%) when they repeated the LSF sentences than non-native signers (51.7%, SD = 15.2%). Furthermore, the main effect of CA also achieved significance (F(2,56) = 5.89, p < .01, ηp2 = .174). Planned comparisons showed that the mean percentage of errors was significantly higher in the youngest group of children (49.7%, SD = 16.2%) in comparison with the two older groups of children together (8–9 years: 38.8%, SD = 12.9%; 10–12 years: 37.5%, SD = 15.5%; F(1,56) = 11.79, p < .001). No significant difference was found between the two older groups (F < 1). Finally, the interaction between AOA and CA was not significant (F < 1).

Percentage of incorrect sign repetition according to sign category

Incorrect sign repetition was classified into three types: (1) variant of the target sign (the repeated sign is easily recognizable as the one produced by the model but the child’s repetition contains one or more errors); (2) substitution_other sign (i.e., another LSF sign has been substituted for the target sign); or (3) substitution_regionalism (i.e., the local sign has been substituted for the target sign). The percentage of type of errors was analyzed (Fig 3). The ANOVA showed that only the main effect of AOA was significant for regionalisms (F(1,56) = 5.43, p < .05, ηp2 = .80), indicating that native signers used regionalisms more often to substitute for target signs (2.84%, SD = 2.92%) than non-native signers (1.28%, SD = 2.04%). No effect of CA (F(2,56) = 1.79, p > .10) or AOA × CA interaction (F < 1) was reported. Likewise, ANOVAs failed to reveal any significant effect for target signs (AOA: F < 1; CA: F < 1; AOA × CA interaction: F < 1) or for sign substitutions (AOA: F(1,56) = 2.00, p > .10; CA: F < 1; AOA × CA interaction: F < 1).

Fig 3

Percentage of errors according to sign category (substitution_other sign, substitution_regionalism or variant of the target sign) with respect to age of acquisition and chronological age.

Adults are considered as the control group (*** p < .001; ** p < .01; * p < .05; n.s. p > .05).

Phonological errors in sign repetition

A 2 × 3 ANOVA with AOA (2 levels: native signers and non-native signers) and CA (3 levels: 6–7, 8–9 and 10–12 years) as between-subject factors was run on the number of phonological errors. Sign category (3 levels: Lexical Sign, Dominant-Hand Classifier and Non-Dominant-Hand Classifier) and Phonological parameters (3 levels: Handshape, Movement, Location) were considered as within-subject factors.

Number of phonological errors

The ANOVA showed that the main effect of AOA was significant (F(1,56) = 37.03, p < .001, ηp2 = .398; Fig 4): made fewer phonological errors on average (M: 26 errors, SD = 12) than non-native signers (M: 50 errors, SD = 18). Neither the main effect of CA (F < 1) nor the AOA × CA interaction reached significance (Fs < 1).

Fig 4

Number of phonological errors with respect to age of acquisition and chronological age.

Adults are considered as the control group (*** p < .001; ** p < .01; * p < .05; n.s. p > .05).

Number of phonological errors according to sign category

Additional ANOVAs were conducted separately on the three different types of signs (Lexical Sign, Dominant-Hand Classifier and Non-Dominant-Hand Classifier). For Dominant-Hand Classifiers, a significant effect of AOA was found (F(1,56) = 12.12, p < .001, ηp2 = .178; Fig 5), as well as a significant effect for the main factor CA (F(2,56) = 4.71, p < .05, ηp2 = .144).

Fig 5

Effect of age of acquisition and chronological age on phonological errors according to the sign category repeated: Lexical Sign, Dominant-Hand Classifier, and Non-Dominant-Hand Classifier.

Adults are considered as the control group (*** p < .001; ** p < .01; * p < .05; n.s. p > .05).

Planned comparisons showed that younger children made more phonological errors on average (28%, SD = 8.2%) on Dominant-Hand Classifiers than the two older groups (F(1,56) = 9.65, p < .01). In contrast, the two older groups made the same proportion of phonological errors on Dominant-Hand Classifiers (8–9 years: 23%, SD = 7.2%; 10–12 years: 20.6%, SD = 9.2%; F < 1). In addition, the AOA × CA interaction was not significant (F(2,56) = 1.07, p > .10).

For Lexical Signs, neither AOA (F < 1), CA (F < 1) nor the AOA × AC interaction was significant (F(2,56) = 1.21, p > .10). This indicates that all children made phonological errors involving Lexical Signs in the same proportion. The same conclusion applies to Non-Dominant-Hand Classifiers: no effect of AOA or CA and no AOA × CA interaction was observed (respectively, F(1,56) = 2.28, p > .10; F < 1; F < 1).

Proportion of errors according to type of phonological parameter

The ANOVA revealed that the main effect of Parameter was significant, suggesting that children did not repeat manual parameters in the same way (F(2,112) = 112.57, p < .001, ηp2 = .06; Fig 6). Planned comparisons showed that location was repeated more accurately than movement or handshape (F(1,56) = 440.23, p < .001; % of errors for location: 7.81%, SD = 7.33%; movement: 34.84%, SD = 11.96%; handshape: 35.58%, SD = 9.86%). In contrast, there was no difference in the proportion of errors for the movement and handshape parameters (F(1,56) = 14.51, p > .10). The main effect of AOA was not significant (F(1,56) = 3.23, p > .05), nor were the other effects (CA: F < 1; AOA × CA interaction: F < 1; Parameter × AOA interaction: F(2,112) = 1.21, p > .05; Parameter × CA interaction: F < 1; Parameter × CA × AOA interaction: F < 1).

Fig 6

Proportion of phonological errors made deaf signing children according to age of acquisition, chronological age and type of manual parameter.

Adults are considered as the control group (*** p < .001; ** p < .01; * p < .05; n.s. p > .05).

Adults’ phonological abilities

The ANOVA showed a main effect of Sign category (F(2,22) = 9.42, p < .001, ηp2 = .460), suggesting that adults made, on average, more phonological errors in Lexical Signs than in Classifiers (respectively 46%, SD = 10%; Dominant-Hand Classifier 26%, SD = 8%; Non-Dominant-Hand classifier: 27%, SD = 8%; Figs 2 to 6). A significant main effect was also found for Phonological parameter (F(2,22) = 14.28, p < .001, ηp2 = .560). Planned comparisons showed that location was repeated more accurately than movement and handshape (respectively, F(1,1) = 11.86, p < .01 and F(1,11) = 23.07, p < .001; % of errors for location: 5%, SD = 4%; movement: 27%, SD = 15%; handshape: 54%, SD = 20%). Contrary to children, there was a difference in the proportion of errors for the movement and handshape parameters (F(1,11) = 7.89, p < .01).

Discussion

This study aimed to assess LSF abilities in children to trace their developmental trajectory. For this purpose, different levels of analysis (phonological and morphosyntactic development, lexical knowledge) were investigated as a function of AOA and CA using an SRT. The choice to use an SRT was motivated by the fact that this task constitutes a reliable indicator of normal or delayed acquisition that can examine phonological, morphological or lexical processing. Therefore, the SRT is viewed as a valuable screening tool for language disabilities. Our work is the first contribution to investigate the development of LSF in a large cohort of deaf children. In addition to this developmental issue, the strength of our study was that it introduced a new task to assess sign language knowledge in children who use an understudied language, namely LSF. To do this this, 34 native signers (6;09–12 years) and 28 non-native signers (6;08–12;08 years) in addition to 10 deaf adult controls were tested. As usual in developmental studies, we wanted to replicate the CA effect for LSF. The systematic variation of AOA allowed us to better understand the extent to which extend sign language development is affected by late acquisition. Children’s abilities to repeat signed sentences were assessed with 20 LSF sentences that were carefully designed to vary in linguistic complexity (i.e., length [2 to 6 signs], types of morphosyntactic units [e.g., classifiers, mouth gestures]).

Regarding general repetition abilities, we were able to show that native signers generally succeeded better in repeating the sentences than non-native signers. This is attested by native signers’ significantly better repetition abilities in comparison with non-native ones. Furthermore, with respect to CA, we found that, on average, younger children (6–7 years) made more errors in the SRT than the older ones (8–9 and 10–12 years). Surprisingly, the two older groups of signers performed equally well. One possible interpretation of this latter result is that at age 8 or 9 children already have good repetition abilities. An additional argument is that adults presented similar SRT error rates to the older children (see Fig 3).

At the lexical level, neither AOA nor CA had significant effects. The only effect found was that native signers produced more regionalisms than non-native signers did. One possible explanation of this effect could be the geographic area where we tested native signers. We went to the south of France, where native signers tend to produce regionalisms. As in Rinaldi et al.’s study [12], very few substitutions were found in our study. This result may be the consequence of the immediacy of the SRT recall task. Unlike a delayed repetition task [9], the immediate recall task does not involve basing the repetition on semantic grounds but rather on a “surface” mechanism such as mimicry, without integrating semantic content. If the task had been a delayed recall task, with a delay between the perception and repetition of the sentences, signers would have needed a semantic representation of the sentence in order to repeat it. In addition, delayed repetition involves a higher working memory load to keep the information active. This might have resulted in the repetition of a correct sentence favoring the semantic content, rather than the formal characteristics produced by the model.

At the morphosyntactic level, like Rinaldi et al. [12], we found no differences in the effects of AOA and CA on morphosyntactic abilities. The SRT does not seem to be an appropriate task to assess the development of the syntactic and morphosyntactic aspects of language. It measures these aspects globally, but not accurately. Like Polišenská et al. [27], we assume that immediate SRTs test lexical phonology and morphosyntax more than semantic conceptual understanding. Although an SRT cannot perfectly test morphosyntactic aspects, it can capture the influences of CA and AOA, and thus it can be considered a screening task to measure overall SL ability. A usage-based explanation can be suggested for this finding: structures with which speakers/signers have a lot of experience may be repeated better [69], explaining the better skills of native signers and older children. Several factors are known to affect the age of acquisition of the signs, vocabulary size and overall lexical development in the deaf population, such as the parents’ hearing status, the age of identification of deafness, and the age of first exposure with sign language [70].

At the phonological level, unlike Rinaldi et al. [12], we decided to focus on manual parameters since our recordings did not allow us to do a fine-grained analysis of the non-manual components. Several children looked at the experimenter more than the camera, with the unfavorable consequence that we were unable to assess facial expression correctly; other children were intimidated by the assessment situation and did not produce typical facial expressions, although they did produce them during the interview.

We are aware that linguists may find it questionable to invoke phonology in the domain of signed languages. However, since the first structural studies of ASL [71,72], sign language linguists consider phonology to constitute not just the organization of speech sounds but rather the organization of minimal units in all languages, even gestural ones. Some attempts have already been made to describe sign languages phonologically [73-76] and claimed that signs can be segmented into meaningful sublexical elements. These sublexical elements, currently called parameters (i.e., handshape, movement, location and orientation), can be considered as phonological units. Parameter substitutions in a given sign can be compared to phoneme substitutions in a word, and switching from one parameter to another makes it possible to oppose minimal pairs in all sign languages: for example, the sign CAKE, described as an “S hand” for handshape, with a “double short tapping movement” on the “cheek” for movement and location, can be opposed to the sign SOAP if one substitutes an “oscillating” movement for the “double short tapping movement” (see S1 Fig). An important issue in sign language studies concerns the role of modality in sign language acquisition, and more specifically whether the phonological development is sign-specific or not. Some researchers suggest that phonological processing is multimodal—or amodal—and involves representations that go beyond the sensory modalities [50,77,78]. Several sign language studies suggest that this language modality involves similar neural systems to those that support spoken language. While we do not claim that speech and sign processing are identical, this research indicates that sign language production and comprehension seem to rely on a similar left-lateralized neural network, whatever the level of processing involved, phonological, semantic, or syntactic [79-83].

Our results showed a significant effect of AOA on phonological skills, as indicated by a lower rate of phonological errors (i.e., handshape, movement, and location) in native signers than in non-native signers. However, both native and non-native signers exhibited the same pattern of results (handshape = movement > location). Interestingly, we also found that, regardless of AOA and CA, location was the best mastered phonological parameter. In contrast, the pattern of results observed in adults was slightly different from that in children (handshape > movement > location). In sum, analysis of the adults’ phonological skills confirmed that the task was sensitive enough to capture phonological differences between the three parameters.

To summarize, our data highlight that the SRT in LSF appears to be a reliable task for detecting differences in phonological abilities as a function of AOA and CA.

As with spoken languages, early signed-phonological development depends on ease of production but also on the frequency of occurrence of both sign and parameter. While several studies have focused on early sign language development [84-88], no benchmarks concerning phonological development in older children are available in the literature. The handshape parameter—the shape of the hand when producing the sign—is the most difficult one to acquire. In a sign, different handshapes involve different numbers of features, which determine the markedness of particular handshapes. Marked handshapes are phonologically more complex (i.e., contain the greatest number of selected features) and motorically more difficult than unmarked ones. Unmarked handshapes involve the fewest selected features. Four stages in handshape development have been observed in ASL [89]. These stages have also been reported in another sign language [BSL: 90, 91], but no study has yet investigated this issue in LSF. Several studies have provided evidence that frequent handshapes in early signs are among the easiest to produce [88], and that unmarked handshapes are acquired and mastered early [85,89,92,93]. Unmarked handshapes are also often made with the non-dominant hand because they are the simplest to articulate. These unmarked handshapes are preserved in sign language aphasia [91]. Marked handshapes are more phonologically complex and are acquired later.

The movement parameter can be classified in two categories: path movement, which involves movement of the hand and the arm, and internal movement, which involves movement of the wrist and fingers. The initiation, type and temporal aspects of the movements performed by these four body parts are phonological: oscillating, waving, arc, straight line, curved, repeated, accelerated, etc. [94,95]. Because of its dynamic aspects and the more or less fine-grained mobilization of different body parts, the mastery of movement is driven by general motor skills (for a review, see [34]). Finally, location is the place in the signing space where the signer produces the sign: head and face, torso, shoulder, arm and neutral space. Because of its strong perceptual saliency and ease of articulation, this parameter tends to be the most accurately reproduced in the early stages of development [34,43,84-87,96].

Consistent with the previous literature, we observed that movement and handshape are the most complex phonological parameters to acquire and that location is mastered early [84-86,96]. Like Mann et al. [88], we showed that handshape and movement are still not accurately mastered by older children, whether native or non-native signers. As other researchers have already suggested, these repetition errors can be attributed to the motor aspects of these parameters, which are undoubtedly more complex [64,82]. In addition, the large inventory of handshapes and movements, in contrast to the location parameter, may influence the frequency of occurrence of each one, causing poorer mastery. In contrast, location is acquired faster because it does not require a fine-grained motor demand, and the anatomical understanding and cognitive representation of the body mean that it can be mastered more easily [96]. Furthermore, Caselli and Cohen-Goldberg [97] reported that location is a more robust parameter owing to its perceptual saliency and higher sublexical frequency.

Regarding the repetition of classifiers, native signers made fewer errors on classifiers executed with the dominant hand. They repeated them better, probably because they use this type of unit more frequently. For non-dominant-hand classifiers, there were too few tokens in our material to reach a conclusion about the influence of AOA or CA on correct repetition. We observed that only dominant-hand classifiers were affected by AOA and, to a lesser extent, CA: native signers and older children repeated these units better. We suggest that this result is a consequence of frequency of use. Native and older signers produce dominant-hand classifiers more frequently—evidence of their morphosyntactic development and sign language mastery. Consequently, when they have to repeat them in an SRT, the phonological pattern of these units is more accurate.

Another cue that allowed us to observe the poorer quality of sign language among non-native signers concerns the sign stream. In non-native signers, the sign rate and pace were slow and non-fluent, with many incomplete movements. These characteristics were not taken into account in the scoring; the coders were asked to give a score from 0 to 5 in order to judge the quality of repetition, regardless of the signers’ lexical and phonological skills. Coders did not know whether the participants were native signers or not. As expected, non-native signers were rated much lower, with a mean score of 2.9/5, whereas native signers had a mean score of 3.5/5. The intercoder comparisons did not show reliable differences for qualitative judgments (Student t-test p > .05).

To conclude, we were pleased to see that our adaptation of the SRT for LSF proved successful. The task highlighted differences in repetition abilities between native and non-native signers of LSF. With time, our study will provide a new LSF screening tool for the clinical, educative and scientific communities. We must ensure that the assessment tool is easy to administer and score for all the professionals who will use it (i.e., speech therapists, researchers, teachers and specialist teachers, sign language teachers) in all test situations (i.e., school, speech therapy, L2 sign language courses). Last, we need to provide a reliable test that can rank children on a clear and reliable language developmental scale [66].

Future work

Our results provide partial support for the validity of the LSF-SRT, as reflected by the different significant effects of CA and AOA. However, we will not formulate firm conclusions about the validity issue until additional data from further studies are at hand. Beyond the standardization and calibration of the new SRT screening tool in LSF, the first version of which was presented in this article, further work should add qualitative analyses. The first analysis could be a phonetic analysis, as a measure of the phonetic complexity of unsuccessfully copied signs. Another analysis could involve measuring the sign stream (number of signs per minute), given that non-native signers seem to sign more slowly. In addition, to improve the quality of assessment, we could increase the complexity of sentences and the types of morphosyntactic structures used in the test. The SRT must be complemented with more accurate assessments to measure language ability or diagnose a language disorder in sign language.

We do not yet have an accurate representation of what kinds of linguistic abilities are mastered at a given age, or which levels can be expected to be achieved successfully in language acquisition and instruction. To the best of our knowledge, there are no reliable benchmarks of the developmental stages of language acquisition in deaf children who use LSF. This study is a first attempt to answer this question, but we need additional tools to evaluate deaf children’s efficiency level in different linguistic domains. In future, we should be able to provide tools that make it possible to plot language development curves, especially during crucial periods of language acquisition, in order to quickly detect disorders. Another fruitful perspective would consist in testing children with dysfunctional acquisition such as non-native signers and children with SLI, for whom sign language data are rare and very heterogeneous. We have to determine whether sign repetition abilities may predict other language skills. Further investigations will combine these repetition data with narrative production based on a cartoon (Caët & Blondel, in prep.) to assess to what extent production skills can be predicted from repetition abilities.

Easy sentence.

(TXT)

Click here for additional data file.

Intermediate easy sentence.

(TXT)

Click here for additional data file.

Intermediate difficult sentence.

(TXT)

Click here for additional data file.

Complex sentence.

(TXT)

Click here for additional data file.

All sentences.

SASS. Size and Shape Specifiers.

(DOCX)

Click here for additional data file.

Minimal pairs in LSF in which changing the internal movement (a. CAKE/SOAP), the handshape (b. GIRL/NICE) or the location (c. IDEA/YELLOW) changes the meaning.

(TIFF)

Click here for additional data file.

Target sign /CHILD/. Coders had to compare this sign to the sign produced by participants.

(TIF)

Click here for additional data file.

24 Feb 2020

Submitted filename: 2. PlosOne-Response-to-reviewers -R2.docx

Click here for additional data file.

27 Apr 2020

PONE-D-20-04317

Language experience in LSF development: Behavioural evidence from a sentence repetition task

PLOS ONE

Dear Mrs Bogliotti,

Thank you for submitting your manuscript to PLOS ONE. After careful consideration, we feel that it has merit but does not fully meet PLOS ONE’s publication criteria as it currently stands. Therefore, we invite you to submit a revised version of the manuscript that addresses the points raised during the review process.

==============================

Many of the points raised are very straightforward, such as ensuring that the English is proofread by a native speaker and that stylistic conventions (such as the use of abbreviations, and using decimal points rather than commas (for example, 7.45, not 7,45) are obeyed. The remaining points mostly ask for additional information and explanation about the methods, together with recommendations for some additional statistical tests. I am sure that none of these changes will pose any problems and look forward to receiving your revised version. We would appreciate receiving your revised manuscript by Jun 11 2020 11:59PM. When you are ready to submit your revision, log on to https://www.editorialmanager.com/pone/ and select the 'Submissions Needing Revision' folder to locate your manuscript file.

If you would like to make changes to your financial disclosure, please include your updated statement in your cover letter.

To enhance the reproducibility of your results, we recommend that if applicable you deposit your laboratory protocols in protocols.io, where a protocol can be assigned its own identifier (DOI) such that it can be cited independently in the future. For instructions see: http://journals.plos.org/plosone/s/submission-guidelines#loc-laboratory-protocols

Please include the following items when submitting your revised manuscript:

A rebuttal letter that responds to each point raised by the academic editor and reviewer(s). This letter should be uploaded as separate file and labeled 'Response to Reviewers'.

A marked-up copy of your manuscript that highlights changes made to the original version. This file should be uploaded as separate file and labeled 'Revised Manuscript with Track Changes'.

An unmarked version of your revised paper without tracked changes. This file should be uploaded as separate file and labeled 'Manuscript'.

Please note while forming your response, if your article is accepted, you may have the opportunity to make the peer review history publicly available. The record will include editor decision letters (with reviews) and your responses to reviewer comments. If eligible, we will contact you to opt in or out.

We look forward to receiving your revised manuscript.

Kind regards,

Bencie Woll, PhD

Academic Editor

PLOS ONE

Additional Editor Comments:

The development of an SRT for French Sign Language is an imprtant step both for researchers of LSF and for practitioners seeking a suitable assessment tool for use in educational settings where children are acquiring LSF. The authors have made considerable changes in content and focus from the original drafts and this has led to a much-improved paper, with very positive reviews. Both reviewers, however, ask for a number of changes and additions of information before the paper can be finally accepted. All of these are very straightforward. I look forward to receiving the final revised version and to its publication.

Journal requirements:

When submitting your revision, we need you to address these additional requirements.

1. Please ensure that your manuscript meets PLOS ONE's style requirements, including those for file naming. The PLOS ONE style templates can be found at

https://journals.plos.org/plosone/s/file?id=wjVg/PLOSOne_formatting_sample_main_body.pdf and

https://journals.plos.org/plosone/s/file?id=ba62/PLOSOne_formatting_sample_title_authors_affiliations.pdf

2. Thank you for stating the following beneath the Acknowledgments Section of your manuscript:

'Funding

This research was partially supported by the Education, Audiovisual and Culture Executive Agency – EACEA (543264-LLP- 1-2013-1-ITKA2-KA2MP “SignMET” – Principal Investigator: Pasquale Rinaldi). Sign Language coders were paid by the EVASIGNE Project (Paris Lumière University Funding) (PI: Caroline Bogliotti & Marion Blondel)'

We note that you have provided funding information that is not currently declared in your Funding Statement. However, funding information should not appear in the Acknowledgments section or other areas of your manuscript. We will only publish funding information present in the Funding Statement section of the online submission form.

Please remove any funding-related text from the manuscript and let us know how you would like to update your Funding Statement. Currently, your Funding Statement reads as follows:

'The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript'

Please clarify the sources of funding (financial or material support) for your study. List the grants or organizations that supported your study, including funding received from your institution.

State what role the funders took in the study. If the funders had no role in your study, please state: “The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.”

If any authors received a salary from any of your funders, please state which authors and which funders.

If you did not receive any funding for this study, please state: “The authors received no specific funding for this work.”

Please include your amended statements within your cover letter; we will change the online submission form on your behalf.

3. Your ethics statement must appear in the Methods section of your manuscript. If your ethics statement is written in any section besides the Methods, please move it to the Methods section and delete it from any other section. Please also ensure that your ethics statement is included in your manuscript, as the ethics section of your online submission will not be published alongside your manuscript.

4. We note that Figures 2 and 7 and Table 2 include an image of a person.

As per the PLOS ONE policy (http://journals.plos.org/plosone/s/submission-guidelines#loc-human-subjects-research) on papers that include identifying, or potentially identifying, information, the individual(s) or parent(s)/guardian(s) must be informed of the terms of the PLOS open-access (CC-BY) license and provide specific permission for publication of these details under the terms of this license. Please download the Consent Form for Publication in a PLOS Journal (http://journals.plos.org/plosone/s/file?id=8ce6/plos-consent-form-english.pdf). The signed consent form should not be submitted with the manuscript, but should be securely filed in the individual's case notes. Please amend the methods section and ethics statement of the manuscript to explicitly state that the patient/participant has provided consent for publication: “The individual in this manuscript has given written informed consent (as outlined in PLOS consent form) to publish these case details”.

If you are unable to obtain consent from the subject of the photograph, you will need to remove the figure and any other textual identifying information or case descriptions for this individual.

[Note: HTML markup is below. Please do not edit.]

Reviewers' comments:

Reviewer's Responses to Questions

Comments to the Author

1. Is the manuscript technically sound, and do the data support the conclusions?

The manuscript must describe a technically sound piece of scientific research with data that supports the conclusions. Experiments must have been conducted rigorously, with appropriate controls, replication, and sample sizes. The conclusions must be drawn appropriately based on the data presented.

Reviewer #1: Partly

Reviewer #2: Yes

**********

2. Has the statistical analysis been performed appropriately and rigorously?

Reviewer #1: Yes

Reviewer #2: Yes

**********

3. Have the authors made all data underlying the findings in their manuscript fully available?

The PLOS Data policy requires authors to make all data underlying the findings described in their manuscript fully available without restriction, with rare exception (please refer to the Data Availability Statement in the manuscript PDF file). The data should be provided as part of the manuscript or its supporting information, or deposited to a public repository. For example, in addition to summary statistics, the data points behind means, medians and variance measures should be available. If there are restrictions on publicly sharing data—e.g. participant privacy or use of data from a third party—those must be specified.

Reviewer #1: Yes

Reviewer #2: Yes

**********

4. Is the manuscript presented in an intelligible fashion and written in standard English?

PLOS ONE does not copyedit accepted manuscripts, so the language in submitted articles must be clear, correct, and unambiguous. Any typographical or grammatical errors should be corrected at revision, so please note any specific errors here.

Reviewer #1: Yes

Reviewer #2: No

**********

5. Review Comments to the Author

Please use the space provided to explain your answers to the questions above. You may also include additional comments for the author, including concerns about dual publication, research ethics, or publication ethics. (Please upload your review as an attachment if it exceeds 20,000 characters)

Reviewer #1: The project team developed a sentence repetition task (SRT) in French Sign Language (LSF) and tested the psychometric soundness of the LSF-SRT. The authors found that children who are native signers made significantly more repetitions than non-native signers and that younger children made more errors than older children. These finds are consistent with the findings of other language fluency tests.

The introduction was well written with an appropriate literature review. The authors adequately and thoroughly described past research on the use of the SRT in signed languages and the cognitive processes that the test measures including how these tests are more sensitive to language fluency than working memory skills.

While the age ranges of the native and late signers appear similar, please compute and share the results of a t-test to show that the two groups did not differ in their chronological ages.

The actual AOA of the subjects were not listed in Table 1 except “native” or “late. It would be better to include the actual AOA. Table 1 seems to suggest that at least one of the “native” signers has deaf parents that sign Romanian SL at home. It is unclear why this child was included in this study as a native LSF signer. And, another subject has one parent that uses Greek Sign Language and the other uses SLF. Similarly, three of the children in the “non-native” group are from spoken language bilingual families. At the present, we do not know if the sign language acquisition trajectory is the same for children who grow up in a home that does not use the same spoken/written language used at school compared to those who grow up in a family that uses the same spoken/written language used at school. If all of these children were removed from the study, are the results the same?

How was the initial pool of 35 LSF sentences created? What were the qualifications of the two deaf adults that repeated the initial pool of 35 LSF sentences and provided judgements? Were they native signers? Do they have psychology or linguistic backgrounds? What was taken into consideration in the selection of the “best” or most natural 20 sentences?

The reviewer is happy that the authors included Table 2 to give readers an idea of how repetitions were evaluated. This is helpful for other teams who wish to develop an SRT in their local sign language.

A description of the coders’ backgrounds would be helpful if another team wants to replicate the study. Were the coders deaf and native SLF signers? Since each subjects’ repetitions were coded by two coders, please provide the inter-rater reliability between coders. This is an important piece of the test’s psychometric soundness and the authors have the data available.

The reviewer got confused when reading the percentage of repeated signs then the percentage of errors in repeated signs. The percentage of repeated signs analyses included signs that were repeated incorrectly but the reviewer was expecting the first analysis to be on correct repetitions. The authors did not appear to analyze the AOA and CA effects on correct repetitions without errors. This analysis needs to be conducted.

When the number of phonological errors was reported, was that the total sample number of errors (in case of native signers, 26 errors, SD = 12) or the mean number of errors per native signing subject? Please provide the mean but it was not clear if it was the mean or sum.

It was noted that the native signers used regional variations in their repetitions. It is difficult to develop an SRT with no possible regional variations. If the observed regional variations were the same among a number of native signers then the team might want to consider it an appropriate/correct response rather than an error. This makes the reviewer wonder about the test instructions—were the test takers told to make exact repetitions and given an example of a variation and told why it is wrong and necessary to give an exact repetition? Some other sign language SRTs do this to minimize variations in repetitions. Please describe your test instructions.

Since chronological age (CA) is a continuous variable but was transformed to a grouping variable for the ANOVA results. It might be worthwhile to include the correlation between CA and percent correct repetitions so the results could be compared to other SRT studies.

The operational definition of “frozen signs” needs to be provided and the significance of analyzing dominant and non-dominant hand classifiers. Since this is being analyzed, it would be appropriate to report how many of these subjects were right-hand dominant.

Adults were included in the study and were called “control.” But, they did not really serve as a control group. If anything, the native signers were the control group for the late signers. The data from the adult group could be kept or omitted for a different manuscript as it did not really help the authors document the psychometric soundness of the LSF-SRT for children unless the authors wanted to compare the results of young native signers with adult native signers.

The use of “late” for non-native signers has been criticized in the past because “late” sounds more like a judgement compared to “non-native,” please consider changing the terms.

Similar to the introduction, the discussion was thorough and helpful.

The authors suggest that the LSF-SRT is a tool that clinicians and educators could use in the future. But, the authors did not thoroughly describe who their coders were. If their coders were deaf native signers with linguistics knowledge, then it might not be possible for hearing professionals or non-native signers to code results similar to the coders in this study.

Editorial feedback: Line 111, SLI was mentioned but it should be spelled out the first time used for the naïve reader. And, 132, “SLI children” should be changed to “children with SLI” (also, see line 112). Specific Language Impairment was spelled out (without its abbreviation) in the final section, Future Work. Please consider revising.

Claims made the sentence that starts with line 190 should be cited.

The reviewer did not review earlier submissions of the manuscript but read the authors’ responses to the past reviewers’ comments. It appears that the authors addressed most of the past reviewers’ concerns. The manuscript does appear to be more focused on test development and validity than phonology. The authors do not talk about validity at all although their results provide partial support on the validity of the LSF-SRT that they developed.

Reviewer #2: The manuscript has the potential to offer a valuable addition to research into LSF, and the use of a sentence repetition task as an assessment tool. It contributes to and confirms previous findings, whilst helping to bring LSF study in-line with studies in other sign languages. The notable number of native LSF participants from deaf families included in the study is also commendable.

However, I believe there are several issues that need to be addressed by the authors before it would be suitable for publication.

General comments:

• I propose that the article be proofread and corrected by a native English speaker. Whilst the English is mostly clear and understandable, it is not currently at a suitable standard for publishing in PLOS ONE, with many errors and omitted words, so substantial editing is needed.

• Care needs to be taken with consistency in using abbreviations. e.g. LSF, SL and SLI are frequently used in both long and short form. e.g. line 180-181 “SL development” is immediately followed by “sign language development”

• Sections of the manuscript appear to be repetitious. Eg. the start of the discussion section, whilst intended as a recap for the reader, is very similar to the abstract.

Introduction

• Line 111 is the first mention of ‘SLI’ so please correct to “specific language impairment (SLI)” for clarity.

• Line 88 - Likewise, explain ‘ASL’ as ‘American Sign Language’ in the first instance for the reader, unless it can be assumed the readers will be familiar with the abbreviation.

Methodology

• Little information is given as to how the sentences for the SRT were created. Did the authors adapt sentences from other SRTs, or did they create new sentences? A pilot study is mentioned (line 313) from which 20 sentences were selected for their “naturalness”. I am unclear as to what is meant by this and think it requires some clarification. Given that the focus of the manuscript is the utilisation of an SRT, more information about the creation of the sentences would be useful. The description provided of the linguistic content of sentences, along with complexity level is good.

• Participants - the inclusion of 10 deaf native signing adults is a useful addition to the study for comparison, however limited information is given about them other than their age range. As they are being used as a comparative group, please provide more detail on their ages (and mean age), and sex etc.

• Procedure - line 337 states testing took place in the school library, suggesting that all of the participants attended the same school. If this is the case, it may be a relevant point to make for the study as it offers some control to the language environment/experience of all of the children whilst they are at school. It may be a strength to the study to make this clear, if it is the case.

• If the children do all attend the same school, Table 1 on page 14 detailing children’s linguistic environment should clarify that this refers to their “home language environment”.

• Line 380-381 First states there are 6 coders, then states there are 2 coders - I assume the 2 independent coders were used for inter-rater reliability, but this is slightly unclear, please rephrase and add the reliability rating if available.

Results

• The asterisks indicating significance levels appear to be missing from the figures.

• Lines 431 and 432 contain commas instead of decimal points (2,84% instead of 2.84%). Proofreading needed.

Discussion

• Lines 671-674 appear to state that there are no known developmental stages of language acquisition in deaf children and that the current study is a first attempt to address this issue. This statement is not true. I assume that the authors intend this statement to relate to LSF, but it is unclear to the reader so needs to be clarified or rewritten.

• If lines 671-674 are, in fact, in relation to LSF, I am not convinced that this study addresses the question of developmental stages of language acquisition in LSF, beyond the finding that native or early LSF acquisition results in higher accuracy on the task.

• Line 650 - the participants’ fluency was rated by coders on a scale of 0-5. Were the coders blind to which group the participants were in (native or late learners), and was any inter-rater reliability undertaken? Only the average score for early signers is provided in the text, please provide the rest of the data for this measurement.

• Lines 539-540 regarding regionalisms - were participants all tested in the same region, or from different areas of France? if participants were all from the same area (or school) would the authors predict differing regionalisms in their signing?

**********

6. PLOS authors have the option to publish the peer review history of their article (what does this mean?). If published, this will include your full peer review and any attached files.

If you choose “no”, your identity will remain anonymous but your review may still be made public.

Do you want your identity to be public for this peer review? For information about this choice, including consent withdrawal, please see our Privacy Policy.

Reviewer #1: Yes: pH

Reviewer #2: No

[NOTE: If reviewer comments were submitted as an attachment file, they will be attached to this email and accessible via the submission site. Please log into your account, locate the manuscript record, and check for the action link "View Attachments". If this link does not appear, there are no attachment files to be viewed.]

While revising your submission, please upload your figure files to the Preflight Analysis and Conversion Engine (PACE) digital diagnostic tool, https://pacev2.apexcovantage.com/. PACE helps ensure that figures meet PLOS requirements. To use PACE, you must first register as a user. Registration is free. Then, login and navigate to the UPLOAD tab, where you will find detailed instructions on how to use the tool. If you encounter any issues or have any questions when using PACE, please email us at figures@plos.org. Please note that Supporting Information files do not need this step.

29 May 2020

PONE-D-20-04317

Language experience in LSF development: Behavioural evidence from a sentence repetition task

PLOS ONE

Rebuttal Letter

Answers to the Editor request

Comment Editor 1#. Please ensure that your manuscript meets PLOS ONE's style requirements, including those for file naming. The PLOS ONE style templates can be found at

https://journals.plos.org/plosone/s/file?id=wjVg/PLOSOne_formatting_sample_main_body.pdf and

https://journals.plos.org/plosone/s/file?id=ba62/PLOSOne_formatting_sample_title_authors_affiliations.pdf

Response Editor 1# We checked templates once again.

Comment Editor 2# Thank you for stating the following beneath the Acknowledgments Section of your manuscript:

'Funding

This research was partially supported by the Education, Audiovisual and Culture Executive Agency – EACEA (543264-LLP- 1-2013-1-ITKA2-KA2MP “SignMET” – Principal Investigator: Pasquale Rinaldi). Sign Language coders were paid by the EVASIGNE Project (Paris Lumière University Funding) (PI: Caroline Bogliotti & Marion Blondel)'

We note that you have provided funding information that is not currently declared in your Funding Statement. However, funding information should not appear in the Acknowledgments section or other areas of your manuscript. We will only publish funding information present in the Funding Statement section of the online submission form.

Please remove any funding-related text from the manuscript and let us know how you would like to update your Funding Statement. Currently, your Funding Statement reads as follows:

'The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript'

a. Please clarify the sources of funding (financial or material support) for your study. List the grants or organizations that supported your study, including funding received from your institution.

b. State what role the funders took in the study. If the funders had no role in your study, please state: “The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.”

c. If any authors received a salary from any of your funders, please state which authors and which funders.

d. If you did not receive any funding for this study, please state: “The authors received no specific funding for this work.”

Please include your amended statements within your cover letter; we will change the online submission form on your behalf.

Response E2 # Thank you for this point. We now suppress all information related to the funding in the manuscript. We will put them directly in the Funding Statement section of the online submission form. > XXX

Comment E3#. Your ethics statement must appear in the Methods section of your manuscript. If your ethics statement is written in any section besides the Methods, please move it to the Methods section and delete it from any other section. Please also ensure that your ethics statement is included in your manuscript, as the ethics section of your online submission will not be published alongside your manuscript.

Response E2# We did this change.

Comment E3 # 4. We note that Figures 2 and 7 and Table 2 include an image of a person.

As per the PLOS ONE policy (http://journals.plos.org/plosone/s/submission-guidelines#loc-human-subjects-research) on papers that include identifying, or potentially identifying, information, the individual(s) or parent(s)/guardian(s) must be informed of the terms of the PLOS open-access (CC-BY) license and provide specific permission for publication of these details under the terms of this license. Please download the Consent Form for Publication in a PLOS Journal (http://journals.plos.org/plosone/s/file?id=8ce6/plos-consent-form-english.pdf). The signed consent form should not be submitted with the manuscript, but should be securely filed in the individual's case notes. Please amend the methods section and ethics statement of the manuscript to explicitly state that the patient/participant has provided consent for publication: “The individual in this manuscript has given written informed consent (as outlined in PLOS consent form) to publish these case details”.

If you are unable to obtain consent from the subject of the photograph, you will need to remove the figure and any other textual identifying information or case descriptions for this individual.

Response E3# The individuals appearing in the figure 2 is Hatice Aken, coauthor of the paper, and she had already signed the Consent For. She’s the model in the SRT. The individual appearing in figure 7 is Philomène Perin and provided a Consent Form too. These two forms were already uploaded on the PLOS platform. Moreover, we added, as you suggested, this information in the Method section.

Answers to the Reviewer 1 request

Reviewer #1: The project team developed a sentence repetition task (SRT) in French Sign Language (LSF) and tested the psychometric soundness of the LSF-SRT. The authors found that children who are native signers made significantly more repetitions than non-native signers and that younger children made more errors than older children. These finds are consistent with the findings of other language fluency tests.

The introduction was well written with an appropriate literature review. The authors adequately and thoroughly described past research on the use of the SRT in signed languages and the cognitive processes that the test measures including how these tests are more sensitive to language fluency than working memory skills.

Comment 1# While the age ranges of the native and late signers appear similar, please compute and share the results of a t-test to show that the two groups did not differ in their chronological ages.

Response 1# Thank you for this remark. We conducted a t-test which shows no difference between the groups for CA (t(60) = 0.16, p > .05).

Comment 2# The actual AOA of the subjects were not listed in Table 1 except “native” or “late”. It would be better to include the actual AOA.

Response 2# We understand it would have been interesting to provide this information. However, for native, it is not necessary because they have LSF from birth. In addition, hearing parents of late signers have difficulties to estimate the real onset of exposure due to different learning context. Consequently, this information was not available for us.

Comment 3# Table 1 seems to suggest that at least one of the “native” signers has deaf parents that sign Romanian SL at home. It is unclear why this child was included in this study as a native LSF signer. And, another subject has one parent that uses Greek Sign Language and the other uses SLF.

Response 3 # Thank you for pointing out this fact. We were aware about these two cases. However, we decided to include them in our sample for the following reasons. These children have LSF as dominant language, as they are schooled in LSF – French bilingual school since the age of 3 years. In addition, deaf teachers ensured us that these children are highly proficient in LSF.

Furthermore, one of them has a native LSF input (French Sign Language mother with Greek Sign Language father). We add commentaries in the legend of the table in order to disambiguate this issue.

Comment 4 # Similarly, three of the children in the “non-native” group are from spoken language bilingual families. At the present, we do not know if the sign language acquisition trajectory is the same for children who grow up in a home that does not use the same spoken/written language used at school compared to those who grow up in a family that uses the same spoken/written language used at school. If all of these children were removed from the study, are the results the same?

Response 4 # Thank you for this commentary. However, the three mentioned children, due to their native deafness, did not access to spoken languages of their parents. Consequently, we reasonably assumed that they had a similar sign language acquisition trajectory than the other children with no bilingualism at home.

Comment 5# How was the initial pool of 35 LSF sentences created?

Response 5 # To create our 35 LSF sentences, we initially relied on the corpus of Rinaldi et al. (2018). We adapted the Italian version of the SRT thanks to deaf native signers in order to create a well-designed LSF culturally and linguistically pool of sentences. We added this information in the Stimuli section of the revised version : “The SRT in LSF consisted of 20 sentences, which were selected from a pilot study in which two deaf adults assessed the initial pool of 35 LSF sentences based on their naturalness” (line 329)

Rinaldi, P., Caselli, M. C., Lucioli, T., Lamano, L., & Volterra, V. (2018). Sign Language Skills Assessed Through a Sentence Reproduction Task. The Journal of Deaf Studies and Deaf Education, 23(4), 408 421.

Comment 6# What were the qualifications of the two deaf adults that repeated the initial pool of 35 LSF sentences and provided judgements? Were they native signers? Do they have psychology or linguistic backgrounds?

Response 6# The two signers involved in the above-mentioned pilot study were native signers. They were recruited in our sign language linguistics department or work as assistant-engineer in a research lab.

Comment 7# What was taken into consideration in the selection of the “best” or most natural 20 sentences?

Response 7# Four criteria were taken into consideration for satisfying naturalness: 1) grammaticality, 2) plausibility, 3) age-matched semantic content, 4) saliency of signs for display on screen. We added this sentence in the revised version line 332.

Comment 8#The reviewer is happy that the authors included Table 2 to give readers an idea of how repetitions were evaluated. This is helpful for other teams who wish to develop an SRT in their local sign language.

Response 8# Thank you very much for this encouraging comment. The coding grid has been several times modified which occasioned a huge work.

Comment 9# A description of the coders’ backgrounds would be helpful if another team wants to replicate the study. Were the coders deaf and native SLF signers?

Response 9# We thanks the reviewer for this suggestion. Coders were hearing fluent signers who study in sign language linguistics department. When coders need information about a sign or a regionalism, we ask to deaf signers of the team to help them to disambiguate signs difficult to perceive. These latter cases were rare.

Comment 10# Since each subjects’ repetitions were coded by two coders, please provide the inter-rater reliability between coders. This is an important piece of the test’s psychometric soundness and the authors have the data available.

Response 10# we agree this is a relevant information to add.

Analysis with Student t-test shows that they were no significant cross-coder differences for the following comparisons:

- % of repeated signs (coder 1: 89.5% vs. coder 2: 90%), t(122) = 0.2, p > .05

- % of errors in repeated signs (coder 1: 41.8% vs. coder 2: 42.6%) t(122) = 0.24 p > .05

- Number of phonological errors (coder 1: 23.3% vs. coder 2: 25.6%), t(122) = 0.56, p > .0

- % errors of handshape (coder 1: 11.6% vs. coder 2: 12.2%), t(122) = 0.66, p > .05

- % errors of movement (coder 1: 11.1 % vs. coder 2: 12.1%), t(122) = 1.11, p > .05

- % errors of location (coder 1: 8.8%. vs. coder 2: 6.7%), t(122)=1.10, p > .05

Comment 11# The reviewer got confused when reading the percentage of repeated signs then the percentage of errors in repeated signs. The percentage of repeated signs analyses included signs that were repeated incorrectly but the reviewer was expecting the first analysis to be on correct repetitions. The authors did not appear to analyze the AOA and CA effects on correct repetitions without errors. This analysis needs to be conducted.

Response 11# Sorry for this confusing and we will try now to explain our methodological choice. For our study, what is critical is to target which kind of phonological errors (handshape, movement, location) will occur most frequently in deaf children. For this purpose, we decided to focus on percentage of repeated sign presenting errors. Among these wrong repeated signs, we have classified in the three type of phonological errors.

Comment 12 # When the number of phonological errors was reported, was that the total sample number of errors (in case of native signers, 26 errors, SD = 12) or the mean number of errors per native signing subject? Please provide the mean but it was not clear if it was the mean or sum.

Response 12# Statistically, the first analysis consisted to compute for each participant the sum of phonological errors. The second analysis lead us to calculate the ratio of each phonological error types (handshape, movement, location).

Comment 13# It was noted that the native signers used regional variations in their repetitions. It is difficult to develop an SRT with no possible regional variations. If the observed regional variations were the same among a number of native signers then the team might want to consider it an appropriate/correct response rather than an error.

Response 13# Thank you for this remark. We separated regionalisms from the correct responses. However, we did not judge these regionalisms as errors as they were found in different repetitions.

To explain our rationale to the reviewer: as our scoring grid was built by taking into account the criterion of an ‘exact repetition’, then only when the target sign was produced with a phonological variation in comparison to the model, it was considered as an error (i.e. variant of the target sign; Figure 3). Consequently, regionalism did not satisfy the “criterion error”.

Comment 14# This makes the reviewer wonder about the test instructions—were the test takers told to make exact repetitions and given an example of a variation and told why it is wrong and necessary to give an exact repetition? Some other sign language SRTs do this to minimize variations in repetitions. Please describe your test instructions.

Response 14# In our study, instructions given to the participants were clearly formulated: the participant have to look the sentence and must to wait the signer has finished to repeat as precisely as possible according the model. One hypothesis for explaining the emergence of regionalism is the frequency of some regionalism that frequency could probably not inhibit despite the instruction.

Comment 15# Since chronological age (CA) is a continuous variable but was transformed to a grouping variable for the ANOVA results. It might be worthwhile to include the correlation between CA and percent correct repetitions so the results could be compared to other SRT studies.

Response 15 # We ran the correlation between CA and percent of repeated signs with error. A correlation rp= -0.2 was found indicating that higher the CA, the lower the percentage of errors. However, this correlation was only marginally significant (.05 < p <.10)

Comment 16# The operational definition of “frozen signs” needs to be provided

Response 16# We agree and to avoid misunderstanding of the reader, we replace frozen by lexical. Indeed, it is well established now that SL have two types of lexicon: the core native lexicon, in which are classified the frozen sign, also called lexical signs, that refer to the permanent lexicon in sign languages. This lexicon includes signs that are highly stable, standardized in form and meaning, and frequently used in the sign language. By contrast, the non-native core lexicon in which we found item as depicting signs (classifiers, role shift) that are not stabilized in the sign language, highly variable and weakly lexicalized. The non-core native signs may move into the core lexicon over time (reference).

Comment 17# and the significance of analyzing dominant and non-dominant hand classifiers. Since this is being analyzed, it would be appropriate to report how many of these subjects were right-hand dominant.

Response 17# the rationale to separate Dominant and Non-Dominant Hand classifier was to study at which level of complexity the children were able to repeat. Note that the dominant hand classifiers are more frequent than the non-dominant hand classifier. Moreover, in general classifiers are units emerging during the LSF language development. Most of participants were right-hand dominant.

Comment 18# Adults were included in the study and were called “control.” But, they did not really serve as a control group. If anything, the native signers were the control group for the late signers. The data from the adult group could be kept or omitted for a different manuscript as it did not really help the authors document the psychometric soundness of the LSF-SRT for children unless the authors wanted to compare the results of young native signers with adult native signers.

Response 18 # We agree with reviewer that native signers could be the control group. That is the rationale we followed in the previous version of the manuscript. However, we were asked by a reviewer to add this control group. We found this suggestion fruitful as a reference group for this task. Moreover, it is usual in clinical linguistics to add such adult data in order to assess the validity of test.

Comment 19# The use of “late” for non-native signers has been criticized in the past because “late” sounds more like a judgement compared to “non-native,” please consider changing the terms.

Response 19# We totally agree and we thank the reviewer. We now in the revised version, we replace ‘late’ by ‘non-native’ signers in opposition to native.

Similar to the introduction, the discussion was thorough and helpful.

Comment 20# The authors suggest that the LSF-SRT is a tool that clinicians and educators could use in the future. But the authors did not thoroughly describe who their coders were. If their coders were deaf native signers with linguistics knowledge, then it might not be possible for hearing professionals or non-native signers to code results similar to the coders in this study.

Response 20# This point is very relevant. The scoring has been thought in such a way that hearing signers, speech therapists, schoolteachers could score without help of native signers. As we said previously, coders were all hearing signers. Moreover we created a scoring booklet in which we explain the manner how we scored and which relevant linguistic facts the coder have to focus on.

Comment 21# Editorial feedback: Line 111, SLI was mentioned but it should be spelled out the first time used for the naïve reader.

And, 132, “SLI children” should be changed to “children with SLI” (also, see line 112).

Specific Language Impairment was spelled out (without its abbreviation) in the final section, Future Work. Please consider revising.

Response 21# All changes were done.

Comment 22# Claims made the sentence that starts with line 190 should be cited.

Response 22# we went back to line 190 the reference initially mentioned at line 195.

Comment 23#The reviewer did not review earlier submissions of the manuscript but read the authors’ responses to the past reviewers’ comments. It appears that the authors addressed most of the past reviewers’ concerns. The manuscript does appear to be more focused on test development and validity than phonology. The authors do not talk about validity at all although their results provide partial support on the validity of the LSF-SRT that they developed.

Response 23# We agree that our results partially support the validity of the LSF SRT as reflected by main effect of CA and AOA. However, we preferred refrain firm conclusion on validity issue before to have more data in further research.

We add the following sentence “Our results provide partial support of LSF SRT validity as reflected by different significant effect of CA and AOA. However, we refrain to formulate firm conclusions on the validity issue before to have at hand additional data in further research.” Line 683.

Answers to the Reviewer 2 request

Reviewer #2: The manuscript has the potential to offer a valuable addition to research into LSF, and the use of a sentence repetition task as an assessment tool. It contributes to and confirms previous findings, whilst helping to bring LSF study in-line with studies in other sign languages. The notable number of native LSF participants from deaf families included in the study is also commendable.

However, I believe there are several issues that need to be addressed by the authors before it would be suitable for publication.

General comments:

Comment 24#I propose that the article be proofread and corrected by a native English speaker. Whilst the English is mostly clear and understandable, it is not currently at a suitable standard for publishing in PLOS ONE, with many errors and omitted words, so substantial editing is needed.

Response 24# We agree and the revised version was proofread and corrected by a native speaker of English.

Comment 25# Care needs to be taken with consistency in using abbreviations. e.g. LSF, SL and SLI are frequently used in both long and short form. e.g. line 180-181 “SL development” is immediately followed by “sign language development”

Response 25# We decided to use long form for sign language and short from for SRT, SLI and all acronyms for sign languages of each countries (ASL, LSF, etc.). We checked for consistency along the whole manuscript.

Comment 26# Sections of the manuscript appear to be repetitious. Eg. the start of the discussion section, whilst intended as a recap for the reader, is very similar to the abstract.

Response 26# We agree that the first sentence of the discussion was similar with the sentence in the present study section. We reformulated this sentence in order to avoid repetition for readers. We replace by the following sentence in the revised sentence:

Line 527 “This study aimed to assess LSF abilities in children to trace their developmental trajectory. For this purpose, different levels of analysis (phonological and morphosyntactic development, lexical knowledge) were investigated as a function of AOA and CA using an SRT”

Introduction

Comment 27# Line 111 is the first mention of ‘SLI’ so please correct to “specific language impairment (SLI)” for clarity.

Response 27# Thank you for this remark. We made the change accordingly.

Comment 28# Line 88 - Likewise, explain ‘ASL’ as ‘American Sign Language’ in the first instance for the reader, unless it can be assumed the readers will be familiar with the abbreviation.

Response 28# Thank you for this remark. We made the change accordingly

Methodology

Comment 29# Little information is given as to how the sentences for the SRT were created. Did the authors adapt sentences from other SRTs, or did they create new sentences?

A pilot study is mentioned (line 313) from which 20 sentences were selected for their “naturalness”. I am unclear as to what is meant by this and think it requires some clarification. Given that the focus of the manuscript is the utilisation of an SRT, more information about the creation of the sentences would be useful. The description provided of the linguistic content of sentences, along with complexity level is good.

Response 29 # Thank you for this fruitful commentary which was also made by the other reviewer. Here, the answer proposed to the request information: “To create our 35 LSF sentences, we initially relied on the corpus of Rinaldi et al. (2018). We adapted the Italian version of the SRT thanks to deaf native signers in order to create a well-designed LSF culturally and linguistically pool of sentences. We added this information in the Stimuli section of the revised version: “These 20 sentences were selected from a pilot study in which 2 deaf adults assessed the initial pool of 35 LSF sentences based on their naturalness’. Line 329.

Concerning the naturalness, four criteria were taken into consideration: 1) grammaticality, 2) plausibility, 3) age-matched semantic content, 4) saliency of signs for display on screen. We added this sentence in the revised version” line 331

Comment 30# Participants - the inclusion of 10 deaf native signing adults is a useful addition to the study for comparison, however limited information is given about them other than their age range. As they are being used as a comparative group, please provide more detail on their ages (and mean age), and sex etc.

Response 30# We added the following sentences in the revised manuscript : “They are aged from 25 to 45 (29;9 years old, X Male) and grew up in different area from France, ensuring a real LSF expertise. Moreover, participants were match on the socio-economic status”

Comment 31# Procedure - line 337 states testing took place in the school library, suggesting that all of the participants attended the same school. If this is the case, it may be a relevant point to make for the study as it offers some control to the language environment/experience of all of the children whilst they are at school. It may be a strength to the study to make this clear, if it is the case.

Response 31 # Thank you for this suggestion. We added this information about native children in the Participant Section. We added the following sentence: “Note that all native children attended the same school, which ensured consistency with respect to the language environment/experience for all of them” Line 295.

Comment 32#If the children do all attend the same school, Table 1 on page 14 detailing children’s linguistic environment should clarify that this refers to their “home language environment”.

Response 32 # Unfortunately, all the children did not attend the same school. For this reason, we found more appropriate to take as language environment reference the home language environment especially for non-native signers in order to assess LSF exposure.

Comment 33# Line 380-381 First states there are 6 coders, then states there are 2 coders - I assume the 2 independent coders were used for inter-rater reliability, but this is slightly unclear, please rephrase and add the reliability rating if available. Same request was made by the other reviewer concerning the intercoder reliability.

Response 33# Thank you for this remark. It was a mistake for us. That was not six and we replace six by two.

we agree this is a relevant information to add.

Analysis with Student t-test shows that they were no significant cross-coder differences for the following comparisons:

- % of repeated signs (coder 1: 89.5% vs. coder 2: 90%), t(122) = 0.2, p > .05

- % of errors in repeated signs (coder 1: 41.8% vs. coder 2: 42.6%) t(122) = 0.24 p > .05

- Number of phonological errors (coder 1: 23.3% vs. coder 2: 25.6%), t(122) = 0.56, p > .0

- % errors of handshape (coder 1: 11.6% vs. coder 2: 12.2%), t(122) = 0.66, p > .05

- % errors of movement (coder 1: 11.1 % vs. coder 2: 12.1%), t(122) = 1.11, p > .05

- % errors of location (coder 1: 8.8%. vs. coder 2: 6.7%), t(122)=1.10, p > .05

Results

Comment 34# he asterisks indicating significance levels appear to be missing from the figures.

Response 34 # We have considered this suggestion. However, due to huge information in each figure, it will be too unreadable. We propose to add the name of each factor and interaction with the significant level.

Comment 35# Lines 431 and 432 contain commas instead of decimal points (2,84% instead of 2.84%). Proofreading needed.

Response 35 # Thank you for this remark. We made the change accordingly

Discussion

Comment 36# Lines 671-674 appear to state that there are no known developmental stages of language acquisition in deaf children and that the current study is a first attempt to address this issue. This statement is not true. I assume that the authors intend this statement to relate to LSF, but it is unclear to the reader so needs to be clarified or rewritten.

Response 36 # Thank you for this remark. We added in the revised version “”in LSF” to clarify this point.

Comment 37# lines 671-674 are, in fact, in relation to LSF, I am not convinced that this study addresses the question of developmental stages of language acquisition in LSF, beyond the finding that native or early LSF acquisition results in higher accuracy on the task.

Response 37 # We agree that with respect to AOA, it cannot be viewed as a developmental stages study. Nevertheless, the study of AOA allow us t examine to which extent language development may be influence by AOA. However, as we varied the CA, this study constitutes an approach to discuss the developmental trajectory.

Comment 38# Line 650 - the participants’ fluency was rated by coders on a scale of 0-5. Were the coders blind to which group the participants were in (native or late learners), and was any inter-rater reliability undertaken? Only the average score for early signers is provided in the text, please provide the rest of the data for this measurement.

Response 38 # Yes, the coders had no information about demographic data of learners. We added the following sentence “Coders did not know whether participants to rate were native or not.” The inter-rater reliability did not differ significantly. We add the average score for non-native signers in the revised manuscript. t(122)=0.20, p >.05

Mean C1 3,29

Mean C2 3,26

Comment 39# Lines 539-540 regarding regionalisms - were participants all tested in the same region, or from different areas of France? if participants were all from the same area (or school) would the authors predict differing regionalisms in their signing?

Response 39 # Thank you for this remark. No, the participants did not come from same region. Native children came from the same school in Toulouse (south of France), and non-native came from North area: Rouen and Paris

For native group, we predict that the same regionalisms may be used.

Submitted filename: Rebuttal letter_PONEResponse to reviewers_R3.docx

Click here for additional data file.

14 Jul 2020

Language experience in LSF development: Behavioral evidence from a sentence repetition task

PONE-D-20-04317R1

Dear Dr. Bogliotti,

We’re pleased to inform you that your manuscript has been judged scientifically suitable for publication and will be formally accepted for publication once it meets all outstanding technical requirements.

Within one week, you’ll receive an e-mail detailing the required amendments. When these have been addressed, you’ll receive a formal acceptance letter and your manuscript will be scheduled for publication.

An invoice for payment will follow shortly after the formal acceptance. To ensure an efficient process, please log into Editorial Manager at http://www.editorialmanager.com/pone/, click the 'Update My Information' link at the top of the page, and double check that your user information is up-to-date. If you have any billing related questions, please contact our Author Billing department directly at authorbilling@plos.org.

If your institution or institutions have a press office, please notify them about your upcoming paper to help maximize its impact. If they’ll be preparing press materials, please inform our press team as soon as possible -- no later than 48 hours after receiving the formal acceptance. Your manuscript will remain under strict press embargo until 2 pm Eastern Time on the date of publication. For more information, please contact onepress@plos.org.

Kind regards,

Bencie Woll, PhD

Academic Editor

PLOS ONE

Additional Editor Comments (optional):

The authors have dealt with the reviewers’ comments and the paper now reads well and makes an important contributionn to studies of LSF acquisition. There still remain a few minor points made by the reviewers that should to be corrected in the final version.:

A few minor grammatical errors:

Line 22 - should read "10 adult signers were evaluated" instead of "was evaluated"

Line 42 - should read "on a case by case basis"

Line 56 - insert 'be' so the sentence reads "may be disturbed"

Line 65 - should read "no SRT study has..." instead of had.

Inconsistency in the use of ‘sign language’ vs SL.

Reviewer 2’s Comment 36. Please add a brief clarification.

The authors should briefly address Reviewer 1’s comment about the importance of the home spoken language.

The reporting of intercode comparisons should briefly report the correlation.

Reviewers' comments:

Reviewer's Responses to Questions

Comments to the Author

1. If the authors have adequately addressed your comments raised in a previous round of review and you feel that this manuscript is now acceptable for publication, you may indicate that here to bypass the “Comments to the Author” section, enter your conflict of interest statement in the “Confidential to Editor” section, and submit your "Accept" recommendation.

Reviewer #1: All comments have been addressed

Reviewer #2: (No Response)

**********

2. Is the manuscript technically sound, and do the data support the conclusions?

The manuscript must describe a technically sound piece of scientific research with data that supports the conclusions. Experiments must have been conducted rigorously, with appropriate controls, replication, and sample sizes. The conclusions must be drawn appropriately based on the data presented.

Reviewer #1: Yes

Reviewer #2: Yes

**********

3. Has the statistical analysis been performed appropriately and rigorously?

Reviewer #1: Yes

Reviewer #2: Yes

**********

4. Have the authors made all data underlying the findings in their manuscript fully available?

The PLOS Data policy requires authors to make all data underlying the findings described in their manuscript fully available without restriction, with rare exception (please refer to the Data Availability Statement in the manuscript PDF file). The data should be provided as part of the manuscript or its supporting information, or deposited to a public repository. For example, in addition to summary statistics, the data points behind means, medians and variance measures should be available. If there are restrictions on publicly sharing data—e.g. participant privacy or use of data from a third party—those must be specified.

Reviewer #1: Yes

Reviewer #2: Yes

**********

5. Is the manuscript presented in an intelligible fashion and written in standard English?

PLOS ONE does not copyedit accepted manuscripts, so the language in submitted articles must be clear, correct, and unambiguous. Any typographical or grammatical errors should be corrected at revision, so please note any specific errors here.

Reviewer #1: Yes

Reviewer #2: Yes

**********

6. Review Comments to the Author

Please use the space provided to explain your answers to the questions above. You may also include additional comments for the author, including concerns about dual publication, research ethics, or publication ethics. (Please upload your review as an attachment if it exceeds 20,000 characters)

Reviewer #1: The authors developed a sentence repetition task in LSF and provided data on how native and non-native deaf signers performed on the test. They included age of LSF acquisition, chronological age, and error types in the analyses. The results shed light on the LSF fluency of deaf children of different language experiences and provide psychometric support for the use of the test in research and clinical settings.

I had the privilege to review the previous submission and this revision. I am impressed, overall, how much the authors took into consideration the reviewers comments in each manuscript revision. The present manuscript is much better than the previous one and I hardly have any new criticisms of the draft.

I think the sample could be more "clean" (and replicable) if the native signers had the same LSF experience at home and the non-native signers had the same French experience at home. I disagree with the authors' response that since the participants are deaf, the home spoken language is irrelevant. In the USA, deaf children from hearing English-speaking and Spanish-speaking families perform differently in school. While this does not mean that home language itself might have an effect but other factors that are associated with not having the dominant spoken language at home might have an effect on children's overall language development. I believe that if the two samples (native, non-native) could be more homogeneous within group by omitting those with different language experiences than the rest in the group. If such participant exclusion does not change the results of the study then this would improve the SRT replicability and cross-linguistic comparisons in future studies. The authors appear to disagree.

I would like to thank the authors for computing intercoder comparisons. I am more used to seeing the correlation being reported, rather than a t-test, to show the inter-rater reliability. Please provide the correlation as well in the final manuscript.

Reviewer #2: The authors have addressed all of the reviewers comments and cleared up the few previous areas of confusion. I would like to congratulate the authors on an interesting paper which will be a valuable addition to studies in LSF.

There remain only very minor editorial issues to be addressed from the previous review:

Comment 24 - the manuscript has been corrected by a native English speaker and the clarity of the manuscript has been much improved. There are only a few minor grammatical errors that remain:

Line 22 - should read "10 adult signers were evaluated" instead of "was evaluated"

Line 42 - should read "on a case by case basis"

Line 56 - insert 'be' so the sentence reads "may be disturbed"

Line 65 - should read "no SRT study has..." instead of had.

Comment 25 - the authors say that they have chosen to use the long-form of "sign language" (as opposed to the abbreviation SL) thought the manuscript. However, from page 4 onwards, a mixture of both forms are still used throughout.

Comment 36 - I acknowledge and appreciate that the authors have clarified that they are specifically referring to LSF in line 698. However, the previous sentence which begins the paragraph on line 695-697 still needs to be clear that they are referring to LSF and not language in general, which is how the sentence currently reads.

**********

7. PLOS authors have the option to publish the peer review history of their article (what does this mean?). If published, this will include your full peer review and any attached files.

If you choose “no”, your identity will remain anonymous but your review may still be made public.

Do you want your identity to be public for this peer review? For information about this choice, including consent withdrawal, please see our Privacy Policy.

Reviewer #1: Yes: pH

Reviewer #2: No

28 Jul 2020

PONE-D-20-04317R1

Language experience in LSF development: Behavioral evidence from a sentence repetition task

Dear Dr. Bogliotti:

I'm pleased to inform you that your manuscript has been deemed suitable for publication in PLOS ONE. Congratulations! Your manuscript is now with our production department.

If your institution or institutions have a press office, please let them know about your upcoming paper now to help maximize its impact. If they'll be preparing press materials, please inform our press team within the next 48 hours. Your manuscript will remain under strict press embargo until 2 pm Eastern Time on the date of publication. For more information please contact onepress@plos.org.

If we can help with anything else, please email us at plosone@plos.org.

Thank you for submitting your work to PLOS ONE and supporting open access.

Kind regards,

PLOS ONE Editorial Office Staff

on behalf of

Professor Bencie Woll

Academic Editor

PLOS ONE