Language, a dynamic complex system, adapts to its environment. The Menzerath-Altmann (MA) law, stating that "The longer a language construct, the shorter its components," models this complexity by examining the relationship between a linguistic construct's length and its constituents' lengths. This study uses the MA law to analyze Chinese political discourse, exploring the interaction between diachronic changes (changes over time) and synchronic variations (variations at a specific point in time) in Hong Kong and Mainland China. The two regions provide a unique context due to Hong Kong's transition from British colonial rule to a Special Administrative Region (SAR) of China in 1997. Analyzing these corpora allows for controlled comparisons of language change and variation. The aim is to demonstrate that the MA law applies not only to historical changes or regional variations independently, but also to their interplay, offering a model of language as a self-organizing complex system.

Existing research on political discourse often focuses either on synchronic or diachronic aspects. Synchronic studies analyze variations within a specific time period, such as Savoy's (2018) examination of communication styles in the 2016 US presidential elections. Diachronic studies examine changes over time, like Burgers and Ahrens' (2020) analysis of metaphors in US State of the Union addresses. However, few studies integrate both perspectives. While some argue that diachronic change inherently leads to variation (Thomason, 1997), others suggest that variation can lead to diachronic change (Xu et al., 2022). The current study addresses this gap by examining both diachronic change and synchronic variation simultaneously within a complex systems framework. Previous applications of the MA law have focused on either diachronic changes or synchronic variations but not both concurrently. This research fills this gap by investigating both in a controlled manner.

The study uses the HKBU Corpus of Political Speeches, including addresses from Hong Kong colonial governors (1984-1996) and SAR Chief Executives (1997-2014), and reports on the Work of the Government by PRC Premiers (1984-2013). Data was extracted from three five-year periods approximately ten years apart: pre-handover (1984-1988), early SAR period (1997-2001), and post-handover (2010-2014). The analysis focuses on Chinese-language speeches. Sentences were operationally defined as text segments separated by punctuation. Clause length was determined by counting commas, semicolons, or colons within sentences; headlines were treated differently depending on region and analysis. Word length was measured by syllable count (equivalent to character count in formal written Chinese). The relationship between linguistic constructs and their immediate constituents (sentence-clause and clause-word) was analyzed using the MA law's formula: y = ax^b. This nonlinear function was transformed into a linear function (Y = bX + ln(a)) for fitting, with R² used to assess model fit. Statistical tests (t-tests) and hierarchical clustering were used to compare parameters across regions and time periods.

The study found that the MA law successfully models the relationship between sentence length and clause length in PRC political speeches, except for some speeches in the 2009-2013 period. The fitted parameters (a and b) differentiated PRC speeches from different periods. In Hong Kong, the MA law fit well after excluding headlines as independent sentences. Parameter values across the three Hong Kong periods were similar, suggesting limited diachronic change. However, a significant difference existed between the latter two PRC periods and Hong Kong speeches, highlighting the divergence of the two varieties over time, largely driven by changes in the PRC. Analysis of the clause-word relationship also fit the MA law. Parameters differentiated PRC and Hong Kong speeches and showed diachronic changes within each region. Interestingly, both regions showed similar diachronic tendencies in their parameters, despite the lack of significant diachronic change in Hong Kong’s sentence-clause data. Clustering analysis generally confirmed the distinction between PRC and HK political speeches based on the fitted parameters of the clause-word relationship.

The findings demonstrate that the MA law provides a robust model for analyzing the complex interplay of diachronic changes and synchronic variations in language. The different rates of diachronic change in PRC and HK political discourse are clearly reflected in the model's parameters. The results support the hypothesis that the MA law captures the characteristics of language as a self-organizing system. The sensitivity of the clause-word level correlation offers more detailed information than the sentence-clause level correlation, consistent with some lexicalist theories. This study highlights the importance of considering both diachronic and synchronic aspects of language evolution, along with the influence of linguistic levels in modeling language complexity.

This study successfully applied the MA law to model the combined effects of diachronic change and synchronic variation in Chinese political discourse. The results showed clear distinctions between PRC and HK political speeches, driven primarily by diachronic changes in PRC discourse. The similar diachronic tendencies observed in both regions suggest shared evolutionary patterns. Future research could investigate these patterns in other languages and explore the potential correlation between the fitted parameters across different linguistic levels.

The study's limitations include the operational definitions of sentences and clauses, which might influence the results. The focus on political speeches from specific sources may limit generalizability to other genres or communicative contexts. Further investigation is needed to fully explore the correlation between the fitted parameters, a and b, particularly the negative correlation observed between them. The relatively small number of sentences in some individual speeches (e.g., 1985 PRC political speech) may also have impacted the model's fit for those speeches.