Zero anaphors in Chinese discourse processing

Abstract

In Chinese, zero anaphors occur frequently. This talk discusses several issues in identification of zero anaphors. Syntactic tagging of words in a running text for detection of anaphora requires specification of argument structure for verbal elements. Verbs requiring a subject or an object can be used to locate the position of zero anaphors. Typical argument structures for verbs, however, often do not take into account the expansion of the structure in discourse such as placing the instrument in object position. Once the place of a zero anaphor is located, the discourse processor will attempt to recover the entity. There may be more than one entity in the previous context that can fit in the argument structure of the verb. It has been discussed in the literature that the recovery of zero anaphors requires extra-linguistic knowledge. However, I will discuss two recovery principles that utilize linguistic information as given in the text. The opening statement of a topic continuity normally provides the element for ellipsis in the statements that follow. So the first principle is to return to the opening statement for recovery. Perceptive verbs, on the other hand, often trigger the search of anaphors in the most recent element. This recency principle can be broken by certain adverbial expressions that mark the turn of the event. These principles apply within a topic continuity. The size of a topic continuity is empirically determined to be about 50 words with some variation.

A Preliminary Study of Model-Generated Speech

“…However, a decrease in mutual intelligibility to 0.475 between Beijing and Guangzhou dialects indicates potential communication difficulties [34]. By combining the findings of [33,34], and if the life expectancy difference between two generations is about 100 years, the study extrapolates a mutual intelligibility loss of 0.08 per century, offering insights into historical language dynamics. Notably, this research contributes to our understanding of mid-phonological changes and dialectical relationships.…”

“…Take the implementation of Chinese languages as an example; Refs. [31][32][33][34] have carried out a series of studies on the relationships between Chinese dialects. In [32]'s work, Pearson's correlation coefficients are employed within the statistical analysis to compute the presence of vocabulary and phonetic forms, categorized by the existence of sounds, rhymes, and tones.…”

“…Furthermore, ref. [34] inferred that the maximum mutual intelligibility between old and young generations, based on the index derived by [33], was typically 0.92, suggesting a generational discrepancy of up to 0.08. This discrepancy serves as a benchmark for mutual intelligibility.…”