Experimental Investigation of Matrix and Relative Weak Crossover Effects in Japanese

Centrala begrepp

Weak crossover effects differ in nature between matrix and relative clauses in Japanese. While matrix weak crossover configurations are consistently rejected, relative weak crossover configurations are frequently accepted, suggesting that the relevant distinction is structural and not based solely on linear precedence.

Sammanfattning

This paper presents an experimental investigation of weak crossover effects in Japanese, building on previous work in English. The key findings are:

Matrix weak crossover (M-WCO) configurations are consistently rejected, replicating previous results. This suggests that the subject-object asymmetry observed in matrix clauses extends to weak crossover effects.

Relative weak crossover (R-WCO) configurations, on the other hand, are frequently accepted, in contrast to M-WCO. This pattern persists even after controlling for potential non-structural sources of bound variable anaphora.

The difference between M-WCO and R-WCO cannot be attributed solely to linear precedence, as the Japanese data shows the same contrast despite the different word order compared to English.

The authors conclude that the structural distinction between matrix and relative clauses plays a key role in determining the availability of weak crossover effects, supporting theories that treat relative clauses as structurally distinct from their matrix clause counterparts.

Statistik

Three car companies criticized their presidents.
Three car companies criticized two managers.
A car company criticized its president.

Citat

"While such embedding can be direct, with clauses as "arguments" of the verbs in higher clauses, e.g., [clause John said [clause Mary thinks [clause ...]]], it can also be indirect, with embedded clauses occurring within other elements themselves embedded in the clause. Such is the case of relative clauses, where the clauses are embedded in nominal phrases within higher clauses, e.g., [clause John met [nominal a woman [clause who knows [nominal a man [clause who ...]]]]]]."
"Following Wasow (1972), we may term the somewhat surprising unacceptability of (2) a "weak-crossover" effect. While there are many possible theories that can account for the existence of such effects, the robust detection of analogous patterns of judgement across languages imposes constraints on the possible theories of human linguistic competence, particularly as concerns the generation and interpretation of matrix clauses and their directly embedded counterparts."

Viktiga insikter från

Matrix and Relative Weak Crossover in Japanese: An Experimental Investigation

by Haruka Fukus... på arxiv.org 10-04-2024

https://arxiv.org/pdf/2410.02149.pdf

Djupare frågor

What other structural properties of relative clauses, beyond weak crossover, might differentiate them from matrix clauses?

Relative clauses exhibit several structural properties that differentiate them from matrix clauses, in addition to weak crossover effects. One significant property is the position of the head noun, which is modified by the relative clause. In relative clauses, the head noun typically precedes the relative clause, creating a dependency that is not present in matrix clauses. This structural arrangement can lead to differences in how binding and scope are interpreted.
Another property is the nature of the arguments within the clauses. In matrix clauses, subjects and objects are treated asymmetrically due to their hierarchical positions, which affects binding relations and the interpretation of anaphora. In contrast, relative clauses often allow for more flexible interpretations of missing elements, such as subjects or objects, which can lead to different acceptability patterns for bound variable anaphora (BVA).
Additionally, relative clauses can exhibit different syntactic behaviors, such as the presence of resumptive pronouns or the ability to undergo extraction processes, which are less common in matrix clauses. These structural distinctions contribute to the overall complexity of relative clauses and their interaction with binding phenomena, suggesting that they are not merely a subset of matrix clauses but rather possess unique syntactic characteristics that warrant separate analysis.

How might the proposed ambiguity in the analysis of object relative clauses be tested and distinguished from a true structural difference between matrix and relative clauses?

To test the proposed ambiguity in the analysis of object relative clauses (ORCs) and distinguish it from a true structural difference between matrix and relative clauses, researchers could employ a combination of experimental methodologies and syntactic manipulations. One approach would be to design experiments that systematically vary the structural configurations of ORCs while controlling for potential confounding factors such as linear precedence and thematic roles.
For instance, researchers could create sentences that manipulate the presence of overt subjects or objects in both ORCs and their matrix counterparts. By comparing participants' acceptability judgments across these variations, it would be possible to assess whether the observed patterns are consistent with a structural ambiguity or if they align more closely with a distinct syntactic structure for ORCs.
Additionally, eye-tracking or self-paced reading tasks could be utilized to measure processing difficulty associated with different interpretations of ORCs. If participants exhibit longer reading times or increased eye movements when encountering ambiguous structures, this could indicate that the ambiguity arises from processing complexity rather than a fundamental structural difference.
Finally, incorporating neurocognitive measures, such as event-related potentials (ERPs), could provide insights into the timing and nature of cognitive processes involved in interpreting ORCs. By examining the neural responses to different structural configurations, researchers could further elucidate whether the observed effects stem from ambiguity or true structural distinctions.

Could the observed contrast between subject and object relative clauses in terms of bound variable anaphora be explained by factors beyond just structural differences, such as processing complexity or pragmatic factors?

Yes, the observed contrast between subject relative clauses (SRCs) and object relative clauses (ORCs) in terms of bound variable anaphora (BVA) can indeed be explained by factors beyond mere structural differences. Processing complexity is a significant factor that may influence how speakers interpret BVA in these clauses.
SRCs typically present a more straightforward processing path, as the subject of the relative clause is often more accessible for binding. In contrast, ORCs may introduce additional cognitive load due to the need to maintain the relationship between the head noun and the embedded object, which can complicate the binding process. This increased complexity may lead to lower acceptance rates for BVA interpretations in ORCs compared to SRCs, even if both structures are theoretically capable of supporting BVA.
Pragmatic factors also play a crucial role in shaping the interpretation of BVA. The context in which a sentence is presented can influence how speakers perceive the relationships between binders and bindees. For example, if the context suggests a strong association between the head noun and the subject of the relative clause, speakers may be more inclined to accept BVA interpretations in SRCs. Conversely, in ORCs, the pragmatic context may not support the same level of association, leading to reduced acceptability of BVA.
In summary, while structural differences are important in understanding the behavior of SRCs and ORCs, processing complexity and pragmatic factors are also critical in explaining the observed contrasts in BVA interpretations. Future research should continue to explore these dimensions to provide a more comprehensive understanding of how different types of relative clauses function in language.