Off-line measures suggest that the hierarchy rule strategy applies to the processing of free relative clauses (RCs) in German, suggesting that case matches between covert head and RC-pronoun are preferred, whereas mismatches are dissociated based on their (non)conformity with the case hierarchy. We asked whether on-line sentence processing supports this pattern, and investigated free RCs with combinations of nominative, accusative and dative case in the self-paced reading paradigm. Our results are captured best by assuming two further strategies: the parallel syntactic function strategy and, especially, the subject-first strategy. Based on these strategies, we provide an Optimality Theory-style analysis that explains prior findings and particularities of incremental sentence processing, and that additionally accounts for the exceptional behaviour of accusative-nominative mismatches.
This study investigates filler integration by adult native speakers of Greek. Object Relative Clauses (RCs) including ditransitives were tested in order to address (a) whether Greek adults show reactivation of the filler at the gap as proposed by the Trace Reactivation Hypothesis (TRH), and (b) the psychological reality of two theoretical analyses of ditransitives in Greek. The first analysis proposes that the base-generated order is direct object (DO)–indirect object (IO), where the DO c-commands the IO, and the second analysis proposes that the basic word order is IO–DO, where the IO c-commands the DO. Cross-Modal Priming (CMP) (Swinney et al., 1979) and Probe-Classification During Reading aloud (Dekydtspotter et al., 2010) were employed. Trace reactivation effects confirm the TRH only in the CMP task and only when adopting the IO–DO order as basic.
In Italian, relative clauses are syntactically ambiguous between a subject and an object reading when the subject and the object have the same number. In the absence of disambiguating cues, the parser analyzes the sentence as a subject relative clause, as subject relative clauses are easier to process than object relative clauses. However, the object reading can be triggered by morpho-syntactic and syntactic cues, such as agreement and word order. In two self-paced reading experiments, we investigate the effectiveness of these cues in triggering an object reanalysis by comparing cues that appear at an early processing stage (inside the relative clause) and cues that appear at a late processing stage (after the relative clause). Results show that reanalysis is attained more effortlessly for early cues, while late cues appears to trigger little or no reanalysis, suggesting that the longer the processor has been committed to the incorrect analysis, the harder is to undo the initial commitment (digging-in effect; Tabor & Hutchins, 2004; see also Ferreira & Henderson, 1991). From these results, we argue that one critical factor that affects ease of parsing is the processing stage at which a cue is made available in the input. We conclude with a discussion on the self-organized sentence processing model (SOSP), which provides a framework to capture these effects.
We report the results from two experiments on the processing of filler-gap dependencies in German using event-related potentials (ERPs). Our aim was to identify and isolate brain responses linked to semantic vs. syntactic integration processes. Using maximally parallel stimulus materials, we examined ERPs at the subcategorising verb in sentences containing either a semantically implausible direct object filler or a filled direct object gap. Whilst the processing difficulty associated with trying to integrate an implausible filler with its subcategoriser was reflected in an enhanced N400 response, encountering a filled preverbal object gap elicited a P600 response at the verb but no negativity. These results confirm that the semantic and syntactic integration of displaced constituents are dissociable and qualitatively distinct processes.
PRO is a theoretically attractive empty category proposed to fill the subject position in non-finite clauses. This chapter is concerned with the empirical evidence for PRO. We investigate reactivation patterns in Control sentences in Norwegian with data from two reaction time experiments. We use a picture recognition task to measure the time and position of PRO antecedent reactivation during sentence processing. The data analysis explores a baseline that has the ability to separate syntactic priming (correct PRO antecedent reactivation) and repetition priming. We find a significant effect of position and an effect of syntactic priming for Subject Control using mixed effects models. We discuss possible theoretical implications and necessary future research.
We present an experimental study on paradigmatic gaps in the neuter gender of Norwegian adjectives. The formation of neuter gender forms is typically not complicated, and yet people hesitate to use a predictable set of adjectives. We investigate the effect of priming in order to determine if the missing neuter forms will facilitate lexical decisions for their common gender form, possibly through lemma activation. A clear difference was found between regular and problematic adjectives. This shows that the intuition of speakers is reflected in detectable differences in processing.
It has been proposed that language processing invokes extra-grammatical heuristics in addition to, or instead of the computational system (e.g., Townsend & Bever, 2001; Karimi & Ferreira, 2016). The outputs of these extra-grammatical heuristics are called “good-enough” representations. These representations lack (syntactic) detail and are incomplete (Karimi & Ferreira, 2016). This paper evaluates this claim by investigating one extra-grammatical processing heuristic in particular: the NV(N)-strategy. Two experiments prove that (i) interpretations that would result from application of the NV(N)-strategy are sometimes difficult to generate and (ii) listeners compute (syntactic) representations for sentences that are more detailed than the NV(N)-strategy would predict. This gives rise to the question whether “good-enough” representations are computed at all.
This paper investigates the processing of pronoun gender by bilingual children. Prior research shows that Dutch–Russian bilinguals below age 7 often make gender agreement errors in Russian anaphoric pronouns, whereas monolingual children are target-like by age 4. This paper aims to establish whether the frequent production errors in the speech of bilinguals are due to lacking knowledge of grammatical gender or due to incidental performance breakdowns. The results of an eye-tracking experiment demonstrate that 5–6-year-old Dutch-Russian bilinguals are sensitive to gender cues, but are slower than Russian monolingual peers and adults in pronoun resolution. The findings support the view that bilinguals possess abstract grammatical representations, but are less efficient in gender production and processing due to competition cost in bilingualism.
In this paper, we describe three case studies of a syntactic intervention given to three Italian-speaking children with cochlear implants (CIs) to improve proficiency in the production and comprehension of relative clauses, and also their narrative skills. The methodology adopted for the syntactic intervention follows previous studies on aphasic patients with agrammatism and children with developmental language disorders (DLD). Indeed, these studies have shown that an explicit teaching of syntactic rules helps in the recovery and improvement of complex structures derived by syntactic movement. Results showed a general improvement in the production and comprehension of relative clauses, and generalization effects to untrained structures. Moreover, also narrative skills improved. Results were maintained several months after the end of the syntactic intervention.
This study deals with the case of Emma, an Italian 8-year-old girl affected by a sex chromosome anomaly called Trisomy X and diagnosed with Expressive Language Impairment. Language assessment of both comprehension and production was conducted through standardized and non-standardized, experimental tests. Passive, cleft, wh-sentences, and relative clauses with manipulation of number features were tested, in addition to clitic pronouns. Her spontaneous production was also assessed. Emma has problems with functional words realizing the highest portion of nominal structure: articles, prepositions, and clitic pronouns. Given her very good cognitive abilities, Emma was administered a tailor-made language intervention focused on functional elements, namely determiners and prepositions, aimed at enhancing her language awareness. During language intervention, use of Italian Sign Language (LIS) was crucial to boost lexical retrieval of functional prepositions and pluri-syllabic lexical words.
While language comprehension tends to be well preserved in older adults, the processing and comprehension of syntactically complex sentences might be influenced by age-related changes in Working Memory (WM) and Interference Control (IC). Further, aging can be accompanied by cognitive decline caused by neurological conditions such as Mild Cognitive Impairment (MCI), but its impact on on-line sentence processing has rarely been studied. We report a study of word-by-word reading times and comprehension of sentences with different syntactic complexity in young adults (n = 69) and two subgroups of older adults: healthy older adults (n = 32) and older adults with single- and multi-domain amnestic MCI (n = 21). The experimental protocol was based on a self-paced reading task and a variety of neuropsychological measures including Operation Span (WM) and Stroop (IC) tasks. Syntactic complexity was induced using Spanish embedded relative clauses varying subject- versus object-extraction of the antecedent noun phrase (canonical or non-canonical word order, respectively). Moreover, within non-canonical sentences, we distinguished between those that did or did not contain long-distance dependencies between the extracted object and embedded verb. All these manipulations were expected to lead to a gradual recruitment of IC and WM based on the complexity of the sentence structure. Comprehension was similar across groups, with differences explained by WM capacity. In both subgroups of older adults, the on-line processing of object extracted sentences was modulated by their available IC and WM resources, although older adults with MCI seem to recruit WM to a lesser extent. In conclusion, results suggest that IC and WM have a modulatory role in the processing and comprehension of syntactically complex sentences in older adults. Moreover, older adults with MCI seem to be particularly overwhelmed by WM demands during sentence processing and comprehension.
Children with Down syndrome (DS) have difficulty comprehending complex syntactic structures. Ring and Clahsen (2005) attribute these problems to a specific syntactic deficit that affects A(=argument) movement. We investigated the comprehension of passive sentences (A-movement) and wh-questions (wh-movement) in 22 German children and adolescents with DS compared to mental-age matched typically developing children. The results suggest that an impairment in the syntactic development affecting also wh-movement structures occurs in a substantial proportion of children with DS and exceeds their general cognitive disabilities. The error patterns of both tests reveal problems with non-canonical clauses and the use of the Agent-first strategy as in early stages of typical acquisition. Moreover, limitations in phonological working memory might contribute to syntactic processing problems in individuals with DS.
Off-line measures suggest that the hierarchy rule strategy applies to the processing of free relative clauses (RCs) in German, suggesting that case matches between covert head and RC-pronoun are preferred, whereas mismatches are dissociated based on their (non)conformity with the case hierarchy. We asked whether on-line sentence processing supports this pattern, and investigated free RCs with combinations of nominative, accusative and dative case in the self-paced reading paradigm. Our results are captured best by assuming two further strategies: the parallel syntactic function strategy and, especially, the subject-first strategy. Based on these strategies, we provide an Optimality Theory-style analysis that explains prior findings and particularities of incremental sentence processing, and that additionally accounts for the exceptional behaviour of accusative-nominative mismatches.
This study investigates filler integration by adult native speakers of Greek. Object Relative Clauses (RCs) including ditransitives were tested in order to address (a) whether Greek adults show reactivation of the filler at the gap as proposed by the Trace Reactivation Hypothesis (TRH), and (b) the psychological reality of two theoretical analyses of ditransitives in Greek. The first analysis proposes that the base-generated order is direct object (DO)–indirect object (IO), where the DO c-commands the IO, and the second analysis proposes that the basic word order is IO–DO, where the IO c-commands the DO. Cross-Modal Priming (CMP) (Swinney et al., 1979) and Probe-Classification During Reading aloud (Dekydtspotter et al., 2010) were employed. Trace reactivation effects confirm the TRH only in the CMP task and only when adopting the IO–DO order as basic.
In Italian, relative clauses are syntactically ambiguous between a subject and an object reading when the subject and the object have the same number. In the absence of disambiguating cues, the parser analyzes the sentence as a subject relative clause, as subject relative clauses are easier to process than object relative clauses. However, the object reading can be triggered by morpho-syntactic and syntactic cues, such as agreement and word order. In two self-paced reading experiments, we investigate the effectiveness of these cues in triggering an object reanalysis by comparing cues that appear at an early processing stage (inside the relative clause) and cues that appear at a late processing stage (after the relative clause). Results show that reanalysis is attained more effortlessly for early cues, while late cues appears to trigger little or no reanalysis, suggesting that the longer the processor has been committed to the incorrect analysis, the harder is to undo the initial commitment (digging-in effect; Tabor & Hutchins, 2004; see also Ferreira & Henderson, 1991). From these results, we argue that one critical factor that affects ease of parsing is the processing stage at which a cue is made available in the input. We conclude with a discussion on the self-organized sentence processing model (SOSP), which provides a framework to capture these effects.
We report the results from two experiments on the processing of filler-gap dependencies in German using event-related potentials (ERPs). Our aim was to identify and isolate brain responses linked to semantic vs. syntactic integration processes. Using maximally parallel stimulus materials, we examined ERPs at the subcategorising verb in sentences containing either a semantically implausible direct object filler or a filled direct object gap. Whilst the processing difficulty associated with trying to integrate an implausible filler with its subcategoriser was reflected in an enhanced N400 response, encountering a filled preverbal object gap elicited a P600 response at the verb but no negativity. These results confirm that the semantic and syntactic integration of displaced constituents are dissociable and qualitatively distinct processes.
PRO is a theoretically attractive empty category proposed to fill the subject position in non-finite clauses. This chapter is concerned with the empirical evidence for PRO. We investigate reactivation patterns in Control sentences in Norwegian with data from two reaction time experiments. We use a picture recognition task to measure the time and position of PRO antecedent reactivation during sentence processing. The data analysis explores a baseline that has the ability to separate syntactic priming (correct PRO antecedent reactivation) and repetition priming. We find a significant effect of position and an effect of syntactic priming for Subject Control using mixed effects models. We discuss possible theoretical implications and necessary future research.
We present an experimental study on paradigmatic gaps in the neuter gender of Norwegian adjectives. The formation of neuter gender forms is typically not complicated, and yet people hesitate to use a predictable set of adjectives. We investigate the effect of priming in order to determine if the missing neuter forms will facilitate lexical decisions for their common gender form, possibly through lemma activation. A clear difference was found between regular and problematic adjectives. This shows that the intuition of speakers is reflected in detectable differences in processing.
It has been proposed that language processing invokes extra-grammatical heuristics in addition to, or instead of the computational system (e.g., Townsend & Bever, 2001; Karimi & Ferreira, 2016). The outputs of these extra-grammatical heuristics are called “good-enough” representations. These representations lack (syntactic) detail and are incomplete (Karimi & Ferreira, 2016). This paper evaluates this claim by investigating one extra-grammatical processing heuristic in particular: the NV(N)-strategy. Two experiments prove that (i) interpretations that would result from application of the NV(N)-strategy are sometimes difficult to generate and (ii) listeners compute (syntactic) representations for sentences that are more detailed than the NV(N)-strategy would predict. This gives rise to the question whether “good-enough” representations are computed at all.
This paper investigates the processing of pronoun gender by bilingual children. Prior research shows that Dutch–Russian bilinguals below age 7 often make gender agreement errors in Russian anaphoric pronouns, whereas monolingual children are target-like by age 4. This paper aims to establish whether the frequent production errors in the speech of bilinguals are due to lacking knowledge of grammatical gender or due to incidental performance breakdowns. The results of an eye-tracking experiment demonstrate that 5–6-year-old Dutch-Russian bilinguals are sensitive to gender cues, but are slower than Russian monolingual peers and adults in pronoun resolution. The findings support the view that bilinguals possess abstract grammatical representations, but are less efficient in gender production and processing due to competition cost in bilingualism.
In this paper, we describe three case studies of a syntactic intervention given to three Italian-speaking children with cochlear implants (CIs) to improve proficiency in the production and comprehension of relative clauses, and also their narrative skills. The methodology adopted for the syntactic intervention follows previous studies on aphasic patients with agrammatism and children with developmental language disorders (DLD). Indeed, these studies have shown that an explicit teaching of syntactic rules helps in the recovery and improvement of complex structures derived by syntactic movement. Results showed a general improvement in the production and comprehension of relative clauses, and generalization effects to untrained structures. Moreover, also narrative skills improved. Results were maintained several months after the end of the syntactic intervention.
This study deals with the case of Emma, an Italian 8-year-old girl affected by a sex chromosome anomaly called Trisomy X and diagnosed with Expressive Language Impairment. Language assessment of both comprehension and production was conducted through standardized and non-standardized, experimental tests. Passive, cleft, wh-sentences, and relative clauses with manipulation of number features were tested, in addition to clitic pronouns. Her spontaneous production was also assessed. Emma has problems with functional words realizing the highest portion of nominal structure: articles, prepositions, and clitic pronouns. Given her very good cognitive abilities, Emma was administered a tailor-made language intervention focused on functional elements, namely determiners and prepositions, aimed at enhancing her language awareness. During language intervention, use of Italian Sign Language (LIS) was crucial to boost lexical retrieval of functional prepositions and pluri-syllabic lexical words.
While language comprehension tends to be well preserved in older adults, the processing and comprehension of syntactically complex sentences might be influenced by age-related changes in Working Memory (WM) and Interference Control (IC). Further, aging can be accompanied by cognitive decline caused by neurological conditions such as Mild Cognitive Impairment (MCI), but its impact on on-line sentence processing has rarely been studied. We report a study of word-by-word reading times and comprehension of sentences with different syntactic complexity in young adults (n = 69) and two subgroups of older adults: healthy older adults (n = 32) and older adults with single- and multi-domain amnestic MCI (n = 21). The experimental protocol was based on a self-paced reading task and a variety of neuropsychological measures including Operation Span (WM) and Stroop (IC) tasks. Syntactic complexity was induced using Spanish embedded relative clauses varying subject- versus object-extraction of the antecedent noun phrase (canonical or non-canonical word order, respectively). Moreover, within non-canonical sentences, we distinguished between those that did or did not contain long-distance dependencies between the extracted object and embedded verb. All these manipulations were expected to lead to a gradual recruitment of IC and WM based on the complexity of the sentence structure. Comprehension was similar across groups, with differences explained by WM capacity. In both subgroups of older adults, the on-line processing of object extracted sentences was modulated by their available IC and WM resources, although older adults with MCI seem to recruit WM to a lesser extent. In conclusion, results suggest that IC and WM have a modulatory role in the processing and comprehension of syntactically complex sentences in older adults. Moreover, older adults with MCI seem to be particularly overwhelmed by WM demands during sentence processing and comprehension.
Children with Down syndrome (DS) have difficulty comprehending complex syntactic structures. Ring and Clahsen (2005) attribute these problems to a specific syntactic deficit that affects A(=argument) movement. We investigated the comprehension of passive sentences (A-movement) and wh-questions (wh-movement) in 22 German children and adolescents with DS compared to mental-age matched typically developing children. The results suggest that an impairment in the syntactic development affecting also wh-movement structures occurs in a substantial proportion of children with DS and exceeds their general cognitive disabilities. The error patterns of both tests reveal problems with non-canonical clauses and the use of the Agent-first strategy as in early stages of typical acquisition. Moreover, limitations in phonological working memory might contribute to syntactic processing problems in individuals with DS.