In the latest blog from our Legislatures in Uncertain Times conference, Matt Williams, University of Oxford, discusses differing approaches to legislative language.
Introduction
How do uncertain times affect the language of legislation? In uncertain times, legislatures can struggle to frame policy in clear language. But, legislatures constrained as to how and what legislation may be made, will respond to uncertain times with largely generic legislation. Such legislation conforms to linguistic norms (genre), and is unambitious in content. Legislatures that are less encumbered by rules will, when guided by a powerful executive, manage uncertainty through adaptable legislation. Such legislation will deviate from generic forms. The meaning of legislation is thereby more easily determined by executive discretion, according to the perceived requirements of changeable contexts. In other words, weak legislatures with strong executives face uncertainty by yielding control of their primary resource — language.
This enables efficient policy change, but undermines representativeness and accountability (Strøm et al 2003). If the meaning of legislative language is controlled by a narrow elite, there is a diminished ‘logic of communication’ between government and governed (Schmidt 2008). A logic of communication — whereby institutions of government clearly enunciate the hows, whats, and whys of policy — is enhanced by institutional and partisan constraints that encourage clarity from policy elites. The principal consequence of a diminished logic of communication is that context determines text, rather than text guiding acceptable responses to context. Policy is developed without prior permission, but in the expectation of forgiveness. This approach to statecraft has proven difficult to manage in the UK, and has contributed to distrust of politicians.
It may seem trite to argue that uncertainty begets unclear legislation. But it is surprising how far institutions and parties can mitigate any loosening of language, and how quickly new legislatures can develop a clear mode of communication. These claims were tested using a text mining algorithm, that analysed all primary legislation enacted by the UK Parliament, Scottish Parliament, and Northern Ireland Assembly from 2000-2015. That is 7,725,930 words from Westminster, 2,431,028 words from Holyrood, and 749,682 words from Stormont. Wales was excluded because Acts of the National Assembly only date back to 2012. And, it must be noted that data for Northern Ireland are broken up by the 2002-07 suspension of devolution. Nonetheless, there is significant evidence of lexical and grammatical clarity from the newer legislatures. Westminster has, conversely, responded to uncertainty with rapid increases in the indeterminacy of its language.
The Genre of Legislation
Legislation has for some time been recognised as a genre of language (Kurzon 1997). In that, legislation relies on generic words, grammar, and even page layouts to maximise comprehension and interpretability. Indeed, democracy is a fight over language. Swirling social forces combine, and recombine, around individuals and groups. These then propose theories of power to inform policy. These policies are then reified into language to guide action. Language of law is essential to the functioning of political institutions. Law mediates how institutions coordinate and communicate their ideas (Schmidt 2008).
What variables affect legislative genre? One of the most sustained problems in constructing clear language is uncertainty. The financial crisis of 2007 created the greatest policy uncertainties of recent years. Of course, Brexit and terrorism also figure. But, policy uncertainty grows when resources are unusually scarce, and when future revenues are indeterminate. Under such circumstances there is especial disagreement as to what can and should be said. In the UK, the financial crisis created major political ructions, with changes in government and party systems at Westminster and Holyrood. Post-crash uncertainty has not been felt evenly across the three legislatures, it is worth noting. So comparisons as to how each have dealt with uncertainty must be read cautiously. My aim here is modest. I wish to share findings that speak to how three legislatures have sought to describe, prescribe, proscribe, disable, and enable in recognisably authoritative language.
Whilst legal genres cannot easily be classified, consistencies and inconsistencies in language can be observed with entity extraction software. My approach was not to measure what words mean, because meaning is unfixed. Instead, certain lexical and grammatical forms generally undermine meaning. Therefore, I measure indeterminate parts of speech in law that tend towards meaninglessness. Specifically, noun and verb-qualifying adjectives and adverbs can muddy the meaning of concepts. Besides these, conjunctions will extend the possible contexts within which a text can be used. And, finally, verbs affect interpretation. Notable in this regard are enabling verbs that delegate authority (make, amend) to agents of the legislature. And, certain auxiliary modal verbs will also extend the possible interpretations of language (may/might as opposed to will/shall).
Homelessness law offers some examples. Each of the UK’s four nations have different linguistic constructions to define ‘priority need’ for housing those fleeing violence (Bate 2017). Parts of speech measured for this essay are highlighted:
The English and Scottish approaches differ particularly sharply. Scotland even abolished the notion of ‘priority need’ in December 2012. But before this step, Scotland’s law on homelessness was the clearest of any regime in the UK. There are very few indeterminate parts of speech, such that the law employs a binary — if x then y — construction. England’s approach is analogue, rather than binary — if x is vulnerable and violence is likely then y. England’s approach relies heavily on discretionary interpretation of text in context.
What, therefore, are the hypotheses? I expect that the financial crisis and ensuing political uncertainty after the 2010 election caused a significant change in Westminster’s use of language, towards greater reliance on indeterminate parts of speech. Conversely, I expect that Scotland’s and Northern Ireland’s legislatures used language more consistently, and moderately reduced the indeterminacy of language in response to the financial crisis. This is because institutional and partisan constraints on what could be legislated began to bite.
Evidence
The evidence presented is not based on any sampling methodology, because every single word legislated was analysed. The first two graphs describe changes in use of adjectives and conjunctions, the latter two graphs look to use of indeterminate modal and enabling verbs. Each graph describes annual changes in the average per section incidence of each of the four parts of speech.
As can be seen, the UK Parliament utilised significantly more indeterminate language after the financial crash, and especially after the 2010 election. Westminster was rocked by the expenses scandal of 2009 and the Wright reforms gave Parliament some enhanced tools for scrutiny. But, despite these, the exigencies of austerity and the political ideology of the Conservative-Liberal Democrat coalition favoured pragmatic, and therefore indeterminate legislative language (Williams 2017). The changes in use of noun/verb-qualifiers and conjunctions is especially stark.
In Scotland, we observe more consistent use of language across the period, with moderately reduced indeterminacy. This can be explained by multi-party governance up to 2011, followed by the SNP’s preference for clear statements of policy over broad delegations of power. There was a modest uptick in indeterminacy at the end of the period under study (2015), after the independence referendum and the publication of the Smith Report. The effects of the Scotland Act 2016 remain to be seen.
In Northern Ireland, there was a modest peak in indeterminate language during the second Executive of 2007-2011. This came after suspension of devolution and direct rule from 2002-2007, and with the devolution of policing and justice powers in April 2010. But, in 2011-2016 austerity hit, leading to more restrained use of language. The 2006 St. Andrew’s Agreement required cross-community approval of key matters, but this was not forthcoming for HM Treasury’s welfare reforms imposed in February 2013. In September 2014 Peter Robinson wrote in the Belfast Telegraph that the Executive was ‘no longer fit for purpose’.
Conclusion
How do uncertain times affect the language of legislation? If a legislature has limited institutional and partisan constraints on framing language, one can expect significant changes in legislative genre. This is observable as rapid and sustained increases in linguistic indeterminacy. If, conversely, a legislature has constraints on what it can legislate, and there are ideological disagreements between powerful parties, there will be greater clarity in legislative language.
Westminster’s rapid changes to its use of language in response to uncertainty is notable. Since the 2017 election, there is no discernible majority in the Commons in favour of any coherent response to Brexit. One can therefore expect Parliament to be more assertive against the executive and ‘take back control’ of its language. A prominent example of this will no doubt be the EU (Withdrawal) Bill. Its language has already been blasted for its indeterminacies. If democracy is a war over words, then battle has assuredly recommenced by the Thames.
Notes
Dr Matt Williams is Access and Career Development Fellow at Jesus College, University of Oxford, where he analyses the language of politics, creating new data that describes how the language of legislation has changed over the past century, and considers the effects of these changes on litigation strategies and public administration. Find out more on his website or follow Matt on Twitter: @magl_williams