Poor writing, not specialized concepts, drives processing difficulty in legal language

Poor writing, not specialized concepts, drives processing difficulty in legal language. Eric Martínez, Francis Mollic, Edward Gibson. Cognition, Volume 224, July 2022, 105070. https://doi.org/10.1016/j.cognition.2022.105070

Abstract: Despite their ever-increasing presence in everyday life, contracts remain notoriously inaccessible to laypeople. Why? Here, a corpus analysis (n ≈10 million words) revealed that contracts contain startlingly high proportions of certain difficult-to-process features–including low-frequency jargon, center-embedded clauses (leading to long-distance syntactic dependencies), passive voice structures, and non-standard capitalization–relative to nine other baseline genres of written and spoken English. Two experiments (N=184) further revealed that excerpts containing these features were recalled and comprehended at lower rates than excerpts without these features, even for experienced readers, and that center-embedded clauses inhibited recall more-so than other features. These findings (a) undermine the specialized concepts account of legal theory, according to which law is a system built upon expert knowledge of technical concepts; (b) suggest such processing difficulties result largely from working-memory limitations imposed by long-distance syntactic dependencies (i.e., poor writing) as opposed to a mere lack of specialized legal knowledge; and (c) suggest editing out problematic features of legal texts would be tractable and beneficial for society at-large.

4. Discussion

Our study aimed to better understand the reason why legal texts can be difficult to understand for laypeople by assessing to what extent: (a) difficult-to-process features that are reportedly common in contracts are in fact present in contracts relative to normal texts, and (b) such features–insofar as they are present–cause processing difficulties for laypeople of different reading levels. Here we discuss in turn the extent to which our results successfully answer these questions, as well as the implications of our results from both a scientific and policy perspective.

With regard to (a), our corpus analysis revealed that features such as center embedding, low-frequency jargon, passive voice and non-standard capitalization–all associated with processing difficulty–were more prevalent in contracts relative to all other texts genres that we looked at. In most cases, this difference was striking. Prior to our study, there had been long-standing speculation and anecdotal accounts of the presence of these features in legal texts, and more recent studies had to some degree identified the prevalence of passive voice (Goźdź-Roszkowski, 2011) and non-standard capitalization (Arbel & Toler, 2020) in legal contracts, either on a smaller scale or with regard to specific types of contracts. Our study provides the first large-scale systematic account of the presence of all of these features in legal texts, both overall and relative to a baseline.

With regard to (b), our experimental study revealed that contracts drafted with all of these features were more difficult to both comprehend and recall than contracts drafted without all of these features, while our analyses of individual linguistic structures revealed that some of the features–such as center-embedding and low-frequency words–present greater difficulties in the context of recall than others, such as passive voice. Although language experience–as measured by ART–predicted comprehension performance, there was no correlation between ART and recall performance, nor was there a significant interaction between register and performance on ART in predicting recall or comprehension. Taken together, these results suggest that these features collectively present processing difficulties for readers of all levels of experience.

From a cognitive science perspective, our results provide insight into the long-puzzling issue of why contracts and other legal texts appear so difficult to understand for laypeople. Some legal theorists have taken the position that “law is a system built upon expert knowledge of technical concepts,” such as habeas corpus, promissory estoppel, and voir dire (Tobia, 2019). As a result, the processing difficulty of legal texts is simply a natural result of not knowing specialized legal concepts. Others have argued that “law is a system built upon ordinary concepts,” such as cause, consent, and best interest (Tobia, 2019; Tobia, 2021). In which case, processing difficulty could be explained by psycholinguistic factors.

Our findings better align with an ordinary concepts account of legal language. Previous work in the general psycholinguistics literature has suggested that center-embedded clauses are difficult to process due to the working memory constraints they impose on readers. Correspondingly, the fact that center-embedded clauses were more than twice as prevalent per sentence in the contract corpus than in the standard-English corpus, and inhibited recall to a greater degree than other features in our experimental study suggests that the cause of the processing difficulty of legal texts may be largely related to working memory costs as opposed to a mere lack of understanding of specialized legal concepts.

Furthermore, if certain concepts are not known by those without expert legal training, then one would not expect to find many words to describe those concepts aside from the low-frequency jargon used by legal experts (just as there are no higher-frequency synonyms for terms such as quark or electron in physics, for example). Consequently, the fact that our corpus analysis revealed that contracts contained even more cases of words with high-frequency synonyms than standard English texts undercuts the view that processing difficulty is driven merely by lack of specialized knowledge. Although it is conceivable that specialized concepts contribute to the perceived processing difficulty of legal texts, our results suggest that insofar as low-frequency legal terminology presents processing difficulty for laypeople, this often results not from unfamiliarity with the concept underlying that terminology but with the terminology itself (such as the phrases ab initio and ex post facto, which in many cases respectively can be simplified to “from the start” and “after the fact”).

From a policy perspective, these findings also provide insight into the long-standing issue of how to ease the processing difficulty of legal texts for laypeople. Efforts to simplify legal language over the last 50 years have focused largely on public legal documents, despite the fact that contracts and other private legal documents are more commonly encountered by laypeople–and increasingly so with the rise of the internet and online terms of service agreements. The fact that contracts contain a stunningly high proportion of features that incur processing difficulty in laypeople that can be feasibly replaced with easier-to-process alternatives underscores the importance for efforts to simplify legal language to not neglect private legal documents. Moreover, the fact that certain features that are common to legal texts–such as center embedding and low-frequency words–appear to inhibit recall to a greater degree than others, such as passive voice, suggests that lawyers interested in simplifying legal texts for the benefit of readers ought to prioritize unpacking clauses into separate sentences and opting for higher frequency synonyms when possible.

The main effect of language experience on comprehension performance suggests that those with less language experience have a harder time understanding legal texts. Given that those with less reading experience as a group tend to be of lower socioeconomic status (Bradley & Corwyn, 2002; Kieffer, 2010), and those of lower SES face greater disenfranchisement from the legal system (Legal Services Corporation, 2017), this suggests that simplifying contracts may have non-trivial access to justice implications, particularly as their prevalence increases. At the same time, the fact that those with higher reading experience also struggled to comprehend and understand contracts written in legalese suggests that redrafting texts into a simpler register would have beneficial effects for those of all reading levels.

To better understand how to integrate these findings, we should aim to understand why lawyers choose to write in such an esoteric manner in the first place. One possibility is that legal language must be written so as to maintain communicative precision. This possibility is undercut by our results and previous findings that show comprehension of legal content with a simplified register (e.g., Masson & Waldron, 1994). While it seems entirely plausible that certain legal jargon is inevitable, our results suggest that in many instances such jargon can be replaced with simpler alternatives that increase recall and comprehension while preserving meaning.

Another possibility is that lawyers choose to write in a complex manner to convey their priorities. For example, if a lawyer prioritizes the user's responsibilities they may focus on making them clear at the expense of other content (e.g., company's obligations). If the lawyer's priorities differ from the reader's priorities they may even do this implicitly as opposed to engaging in an outright “conspiracy of gobbledegook” (Mellinkoff, 2004).

Lastly, lawyers may not choose to write in an esoteric manner. Similar to the “curse of knowledge” (Hinds, 1999; Nickerson, 1999), they may not realize that their language is too complicated for the average reader to understand (Azuelos-Atias, 2018). This hypothesis appears to be supported by previous findings that show an effect of features such as prior knowledge and reading skill on the processing of specialized texts (Cain, Oakhill, & Lemmon, 2004; Kendeou & Van Den Broek, 2007; Long, Prat, Johns, Morris, & Jonathan, 2008; Noordman & Vonk, 1992; Ozuru, Dempsey, & McNamara, 2009). Similarly, one might predict that lawyers would be equally likely to comprehend contracts if they were drafted in an esoteric style as they would if they were drafted in a simpler register, which may render them less able to appreciate the difficulty of these features for those without legal training. Further work into the plausibility of these hypotheses could yield insight into how best to persuade lawyers to integrate the findings of our and similar studies and help alleviate the growing mismatch between the ubiquity and impenetrability of legal texts in the modern era.