Syntactic Comprehension of Relative Clauses and Center Embedding Using Pseudowords

Abstract

Relative clause (RC) formation and center embedding (CE) are two primary syntactic operations fundamental for creating and understanding complex sentences. Ample evidence from previous cross-linguistic studies has revealed several similarities and differences between RC and CE. However, it is not easy to investigate the effect of pure syntactic constraints for RC and CE without the interference of semantic and pragmatic interactions. Here, we show how readers process CE and RC using a self-paced reading task in Korean. More interestingly, we adopted a novel self-paced pseudoword reading task to exploit syntactic operations of the RC and CE, eliminating the semantic and pragmatic interference in sentence comprehension. Our results showed that the main effects of RC and CE conform to previous studies. Furthermore, we found a facilitation effect of sentence comprehension when we combined an RC and CE in a complex sentence. Our study provides a valuable insight into how the purely syntactic processing of RC and CE assists comprehension of complex sentences.

Keywords: embedding; merge; self-paced reading; syntactic hierarchy.

Figure 1

Four different conditions of Experiment 1. Blue and orange shades represent the main clause and the subordinate clause, respectively. Ending markers (an AEM or CEM) are shown as “Ending” only after the first verb for easy explanation here. Case markers attached to each subject, object, and verb were shown throughout the actual experiment. Dotted arrows indicate the formation of an RC between clauses. Brackets surrounding subordinate clauses indicate CE. Condition 1 has an RC in two clauses, with the first clause modifying the subject of the second clause (Paul-NOM). Condition 2 has two clauses conjoined by a CEM without an RC or CE. Condition 3 has both an RC and CE. In this case, the subordinate clause (a structure shaded in orange) is merged with the following main object (box-ACC). Condition 4 has the only CE with a CEM. Here we have provided English names to aid understanding of the subjects; Korean names were used in the actual experiment. Abbreviations: RC (relative clause), CE (center embedding), AEM (adnominalizing ending marker), CEM (conjunctive ending marker), NOM (nominative case marker), ACC (accusative case marker).

Figure 2

Reading-time data from the self-paced reading task in Experiment 1. The mean reading times with standard deviations are depicted for all the conditions. The x-axis indicates the position of words presented, and the y-axis denotes the reading times in milliseconds (ms). The triangular data points depict the RC condition, while the circular data points depict the non-RC (NonRC) condition. Blue and orange lines indicate non-CE (NonCE) and CE conditions, respectively. The significant effects from two-by-two linear mixed-effect analyses are shown at the top of each word position (For details, see Table 3). Exemplar words labeled with four conditions (i.e., RC/NonCE, NonRC/NonCE, RC/CE, NonRC/CE) are presented at the bottom. Abbreviations: RC (the main effect of a relative clause), CE (the main effect of center embedding), RC × CE (an interaction between the relative clause and center embedding), AEM (adnominalizing ending marker), CEM (conjunctive ending marker), NOM (nominative case marker), ACC (accusative case marker). (* = p < 0.05, ** = p < 0.01, *** = p < 0.001).

Figure 3

The pseudoword version of stimuli in Experiment 2. Blue and orange shades represent the main clause and the subordinate clause, respectively. Ending markers (an AEM or a CEM) are shown as “Ending” only after the first verb for easy explanation here. Case markers attached to each subject, object, and verb were shown throughout the actual experiment. Dotted arrows indicate the RC. Brackets surrounding the subordinate clause indicate the CE. Construction of every condition was identical to that of Figure 2. Pseudowords did not provide participants with cues for a constituent—whether the word was a noun or a verb. Therefore, they could only infer the syntactic role of the pseudowords by case markers or ending markers attached to the pseudowords. Here we have provided English pseudowords with four letters to aid understanding of the stimuli. In the actual experiment, we used Korean two-syllable pseudowords. Abbreviations: RC (relative clause), CE (center embedding), AEM (adnominalizing ending marker), CEM (conjunctive ending marker), NOM (nominative case marker), ACC (accusative case marker).

Figure 4

Reading time data from the self-paced pseudoword reading task in Experiment 2. The mean reading times with standard deviations are depicted for all the conditions from Experiment 2. The x-axis indicates the position of pseudowords presented, and the y-axis denotes the reading times in milliseconds (ms). The triangular data points depict the RC condition, while the circular data points depict the non-RC (NonRC) condition. Blue and orange lines indicate non-CE (NonCE) and CE conditions, respectively. The significant effects from two-by-two linear mixed-effect analyses were depicted at the top of each word position (for details see Table 5). Exemplar pseudowords labeled with the four conditions (i.e., RC/NonCE, NonRC/NonCE, RC/CE, NonRC/CE) are presented at the bottom. Abbreviations: RC (the main effect of relative clauses), CE (the main effect of center embedding), RC × CE (an interaction between the relative clause and center embedding), AEM (adnominalizing ending marker), CEM (conjunctive ending marker), NOM (nominative case marker), ACC (accusative case marker). (* = p < 0.05, ** = p < 0.01, *** = p < 0.001).

Figure 5

The partial structure frequencies of the stimuli. The number at the right end of the color gradient indicates the frequency of the partial structure (i.e., a set of words covered by the length of the color gradient) occurring in the corpus. For example, “83” in Condition 1 indicates that the structure of (subject + object + verb-AEM) appeared 83 times in the corpus [60]. The first values for Condition 2 and 4 are not identical because morphologically different endings were used. Not every partial structure was used to determine the relative difference in surprisal because only the structures with the same starting position can be compared. The Sejong Semantically Tagged Corpus of the Modern Korean with 797,251 sentences were analyzed by Hanmaru 2.0, a concordance program developed together with the corpus itself, to calculate partial structures of the sentence stimuli. Abbreviations: AEM (adnominalizing ending marker), CEM (conjunctive ending marker), NOM (nominative case marker), ACC (accusative case marker).