Periodic Reporting for period 1 - ProHeritageSpeakers (Processing and Production in Heritage Speakers: The Role of Disambiguating Cues)
Reporting period: 2021-09-01 to 2023-08-31
This project contributed to research on heritage language/HL bilingualism by bringing new insight into the processing mechanisms underlying real-time language comprehension in child and adult heritage speakers/HSs (early bilinguals who acquire their first language/L1 in a minority setting under reduced input conditions and under the influence of the dominant language/L2 of the society they live in).
The overall objectives were to a. uncover whether child/adult HSs process HL input in real time similarly to monolingual L1 speakers; b. examine whether child/adult HSs are sensitive to grammatical cues like case-marking and number agreement in their HL; c. compare HSs’ ability to exploit these cues for comprehension to their ability to use them in production; (iv) determine whether the presence/absence of similar cues in the dominant L2 lead to differences in HL processing and/or production.
The project brought Romanian, a less investigated language in bilingualism research, to the spotlight and exploited some of its language specific properties to better understand how children (and adults) with a migration background develop their HL and how this may be affected by contact with the L2. Romanian provides various grammatical cues that speakers can use to interpret complex syntactic structures like object "which"-questions. For example, a Differential Object Marker/DOM, "pe" comes before an object wh-word and the subject agrees in number with the verb. ("Pe care băieți îi îmbrățișeaza bunicul?/Which boys is the grandfather hugging?"). To uncover whether child/adult Romanian HSs can use disambiguating cues like DOM and number agreement when processing input in their HL in real time and whether the presence/absence of such cues in their societal language affects the processing and production of "which"-questions in the HL, the project focused on Romanian HSs with L2 German. The project thus investigated whether number agreement, present both in Romanian and in German, represents a stronger cue for HSs in disambiguating between subject and object "which"-questions, than DOM only, which is present in the HL, but not in the L2.
The conclusions of the project are that the presence of DOM on the wh-word in Romanian does not eliminate the subject-object asymmetry found cross-linguistically for the comprehension of "which"-questions and that this asymmetry is stronger in the younger children, as reflected by the effect of age on offline comprehension in monolingual and heritage children alike. The results also suggest that the cue from number agreement guides online processing of object "which"-questions more than DOM on its own, in monolingual children and all groups of adults. The online data also reflect lower rates of correct responses and qualitatively similar, but slower processing in children relative to adults. When comparing the monolingual and HL children, we see that, on a group level, the HS children have more difficulties with the comprehension and production of object "which"-questions relative to the monolingual children. Although number agreement does not seem to impact offline comprehension in either group, the online data suggest that this mismatch guides monolingual children’s online processing more than DOM on its own. The fact that the heritage children do not seem to recruit the number information in the online processing of a short syntactic dependency could indicate more protracted processing, i.e. the number information is processed after the end of the sentence because of slower processing speed. This also suggests that the presence of number agreement in the societal language (German) does not seem to impact the use of this cue for HL processing.
The overall objectives were to a. uncover whether child/adult HSs process HL input in real time similarly to monolingual L1 speakers; b. examine whether child/adult HSs are sensitive to grammatical cues like case-marking and number agreement in their HL; c. compare HSs’ ability to exploit these cues for comprehension to their ability to use them in production; (iv) determine whether the presence/absence of similar cues in the dominant L2 lead to differences in HL processing and/or production.
The project brought Romanian, a less investigated language in bilingualism research, to the spotlight and exploited some of its language specific properties to better understand how children (and adults) with a migration background develop their HL and how this may be affected by contact with the L2. Romanian provides various grammatical cues that speakers can use to interpret complex syntactic structures like object "which"-questions. For example, a Differential Object Marker/DOM, "pe" comes before an object wh-word and the subject agrees in number with the verb. ("Pe care băieți îi îmbrățișeaza bunicul?/Which boys is the grandfather hugging?"). To uncover whether child/adult Romanian HSs can use disambiguating cues like DOM and number agreement when processing input in their HL in real time and whether the presence/absence of such cues in their societal language affects the processing and production of "which"-questions in the HL, the project focused on Romanian HSs with L2 German. The project thus investigated whether number agreement, present both in Romanian and in German, represents a stronger cue for HSs in disambiguating between subject and object "which"-questions, than DOM only, which is present in the HL, but not in the L2.
The conclusions of the project are that the presence of DOM on the wh-word in Romanian does not eliminate the subject-object asymmetry found cross-linguistically for the comprehension of "which"-questions and that this asymmetry is stronger in the younger children, as reflected by the effect of age on offline comprehension in monolingual and heritage children alike. The results also suggest that the cue from number agreement guides online processing of object "which"-questions more than DOM on its own, in monolingual children and all groups of adults. The online data also reflect lower rates of correct responses and qualitatively similar, but slower processing in children relative to adults. When comparing the monolingual and HL children, we see that, on a group level, the HS children have more difficulties with the comprehension and production of object "which"-questions relative to the monolingual children. Although number agreement does not seem to impact offline comprehension in either group, the online data suggest that this mismatch guides monolingual children’s online processing more than DOM on its own. The fact that the heritage children do not seem to recruit the number information in the online processing of a short syntactic dependency could indicate more protracted processing, i.e. the number information is processed after the end of the sentence because of slower processing speed. This also suggests that the presence of number agreement in the societal language (German) does not seem to impact the use of this cue for HL processing.
132 participants (Romanian monolingual and Romanian-German bilingual children and adults) took part in a comprehension and a production task. The online results (measuring participants’ eye-gaze on the screen while listening to sentences) reflect (i) faster processing strategies in adults compared to children; (ii) an overall higher proportion of looks to the correct image in subject questions relative to object questions; (iii) more looks to the correct image in object-questions disambiguated through number mismatch vs. number match. No differences emerged between Romanian monolingual and Romanian-German bilingual adults in the online processing of "which"-questions. Children's online results showed similar processing patterns in monolingual and heritage children for subject questions, which overall displayed more looks to the correct images than object questions. The data also revealed a higher proportion of looks to the target image in object-questions with number mismatch than in questions with number match, but only in the monolingual group.
The offline results (which measured how accurately participants interpret this type of questions) showed that adults were at ceiling with all structures. For children, the offline results revealed (i) significantly better performance with subject- compared to object-questions, (ii) an effect of age in both groups, but also that (iii) monolingual children were significantly more accurate with object questions than HSs. The results of the elicited production task indicate that the monolingual and first-generation immigrant adults produce significantly more target subject and object "which"-questions compared to the other three groups (monolingual children, heritage children and heritage adults) and that the monolingual children produce significantly more target subject and object questions than both heritage children and heritage adults. The project results have been disseminated to scientific audiences at 6 international conferences, 1 workshop organised as part of the project, 1 event entitled "Herkuntssprachen in der Familien". Other activities for the exploitation and dissemination of the project results included 2 presentations for Romanian families across Germany and 1 non-technical summary of results for the participants in the study.
The offline results (which measured how accurately participants interpret this type of questions) showed that adults were at ceiling with all structures. For children, the offline results revealed (i) significantly better performance with subject- compared to object-questions, (ii) an effect of age in both groups, but also that (iii) monolingual children were significantly more accurate with object questions than HSs. The results of the elicited production task indicate that the monolingual and first-generation immigrant adults produce significantly more target subject and object "which"-questions compared to the other three groups (monolingual children, heritage children and heritage adults) and that the monolingual children produce significantly more target subject and object questions than both heritage children and heritage adults. The project results have been disseminated to scientific audiences at 6 international conferences, 1 workshop organised as part of the project, 1 event entitled "Herkuntssprachen in der Familien". Other activities for the exploitation and dissemination of the project results included 2 presentations for Romanian families across Germany and 1 non-technical summary of results for the participants in the study.
This study is the first to compare the online comprehension and production of wh-dependencies in heritage children and adults, and the first to examine the real-time processing of these structures in Romanian monolinguals. The materials developed as part of the project are currently being adapted to German to assess the same bilingual group of children in German, their societal language, in order to examine whether they show similar processing difficulties for these structures in both their languages or whether they only display slower processing in the HL. As far as the wider societal implications are concerned, the results of the project and the discussions and links established with various Romanian organizations across Germany for participant recruitment, as well as with the participants and the parents themselves, have contributed to raising awareness within the Romanian community of the importance of raising the children bilingually and of maintaining the contact with the HL both in the home and outside the home.