Vagueness in Constitutional Courts as a Tool in the Face of Non-Compliance
A Comparative Study of the Supreme Court of Canada, the Supreme Court of the United Stated and the Constitutional Court of South Africa
Naomi L.G. Immers S1536141 Master’s Thesis
Public Administration – Public Management and Leadership Track Supervisor: Dr. Andrei Poama
Second Reader: Dr. Alex Ingrams July 12, 2020
Table of Contents
Table of Contents ... 2
Introduction ... 3
Research Question ... 4
Theoretical Framework ... 5
(Non-)Compliance and the Courts ... 5
Judiciary as Policy-seekers ... 6
Challenges for the Judiciary ... 8
Vagueness as a Tool ... 9
Research Design ... 12
Case Selection ... 12
Data collection ... 12
Operationalization ... 13
Reflection on Validity and Reliability ... 17
Results and Analysis ... 18
Primary Results ... 18
Vagueness and Public Support ... 20
Vagueness and the Legislative and Executive Branches ... 21
Intra-Judicial Vagueness ... 24
Statistical Significance of the Results ... 26
Policy Implications ... 27
Conclusion ... 28
Discussion and limitations ... 29
References ... 31
Appendix I ... 36
Assessment of Courts Against Sternberg’s (2018) Criteria ... 36
Appendix II ... 45
Introduction
Vagueness by a constitutional court may result in the presence of laws or practices that are anti-constitutional. Vague language in court decisions give the legislative and executive branches of government leeway to (partly) refuse or delay compliance with the court’s orders, because that court might not have explicitly stated any sharp boundaries, demands or requirements. In 1954, the Supreme Court of the United States (SCOTUS) ruled that school segregation was unconstitutional and all US schools had to be integrated at ‘all deliberate speed’ (Brown v. Board of Education, 347 U.S. 483 (1954)). The integration faced mass resistance and it took decades before schools were integrated. In 2014, the German Federal Constitutional Court (GFCC) ruled that current inheritance tax law was unconstitutional and had to be revised. The court left the details to the legislators, and as of now, there is yet a revised law to be presented. In both cases, vague language by the court has left the playing field wide open for the legislative policymakers to do as they please. This lack of specificity, certainty, or distinctness in the judgment’s language is what is regarded as ‘vagueness’ in this thesis. Vagueness can occur through the presence of words that indicate uncertainty, such as ‘maybe’, ‘unsure’ or ‘anytime’, or through the absence of certain words. Words like ‘some’ can change the meaning of other words and entire sentences by erasing the boundaries of meaning. Vagueness in court judgments can have far-reaching consequences for ordinary citizens. In constitutional law regarding criminalisation of certain acts, fair trial, human rights and other fundamentals, this could lead to individuals being unconstitutionally treated in trials, treatment, or sentencing. Vagueness can lead to a significant delay of the repeal or revision of a law or act. In the meantime, preservation of the previous laws may have serious consequences for individuals that might for instance receive a different punishment or even be exonerated under a revised law. If vagueness has such an impact on legislative compliance and such potentially large consequences, why do courts use vagueness in their judgments?
Sternberg (2018) has empirically tested the game-theoretic model by Staton and Vanberg (2008). In this model, judicial vagueness increases with a higher degree of judicial uncertainty about the topic and decreases with a higher degree of preference divergence between judges and policymakers. The motivation for the use of vague language in decisions, however, is dependent on the public support a court has. In his research, Sternberg (2018) found support for this model and confirmed that public support and confidence in a court as well as its institutional legitimacy is a key factor in determining the degree to which the courts use vagueness in their opinions. For this research, he compared judgments by the GFCC and the
French Conseil Constitutionnel, two courts of which have historically enjoyed high and low public support, respectively.
In my research, I utilized Staton and Vanberg’s model, as well as Sternberg’s findings, and tested them against other constitutional courts in a comparative study of three courts. My research question, hypotheses and methods were greatly based on Sternberg’s (2018) research. Aside from that, this thesis is an empirical, innovative study that expands existing theory through application on different and different kinds of cases. I applied Sternberg’s methods to different courts that vary in in their degree of public support: the Supreme Court of Canada or SCC (high support), the Constitutional Court of South Africa or CCSA (low support), and the Supreme Court of the United States or SCOTUS (medium support). I have assessed the degree of public support for these courts based on (inter)national surveys and polls that tested for the awareness of, confidence and trust in, and overall support for these courts. To test the generalizability of the model, I selected courts that apply a different law system than the European civil-law tradition, as well as lie in different areas of the world. The rule of law in common law states is derived from judicial decisions, and each statement or decision can affect future law in the jurisdiction. This should not affect the need for courts to be more of less vague, as the right for judicial review is the essential component here, not the law system. By selecting courts operating in different law traditions than the ones already tested, different areas of the world and a different language than the courts tested by Sternberg (2018) I tested the generalizability of his findings.
Research Question
How does the degree of public support for a court affect the degree of vague language in that court’s decisions?
In order to answer my research question and test my hypotheses, I collected court case judgments by the three courts that rule on constitutional review cases. I compiled a list of phrases and words that indicate vagueness, leeway, or other uncertainty or lack of specificity in the English language and ran the cases through textual analysis programme Linguistic Inquiry and Word Count (LIWC). I then analysed the data within the context of Staton and Vanberg’s (2008) model. I measured the degree of vagueness in relation to non-compliance risk and most importantly the public standing of the court. My findings confirm public support is a condition for a rise or a decline of the degree of vagueness in a court’s judgments, as courts with a lower degree of public support use increasingly vaguer language in their judgments when faced with
sufficient risk of non-compliance, while courts with higher public support use more specific wording when faced the same risk. However, the statistical significance of the results is not unilateral, which is discussed in the results section of this thesis.
Theoretical Framework
To be able to answer my research question a few claims must be substantiated. First of all, that the degree of public support is essential for the court and its individual judges in a way that it influences the effectiveness of the institution. Secondly, that judges are active policy-seekers that deliberately and strategically shape legislature. Thirdly, that vagueness is a tool that the courts may use – deliberately or not – to balance compliance and public support. In this theoretical framework, I will explain if and how a court is reliant on support by the public and how this may affect the way the court functions and gives their judgments through the current research.
(Non-)Compliance and the Courts
In order to understand the need for public support by the courts, one must realise that courts are inherently vulnerable institutions (Caldeira and Gibson, 1992; Mondak and Smithey, 1997). Although their rulings must be adhered to according to the law, the courts have no way of enforcing their rulings and are reliant on the executive and legislative in doing so. If the legislative does not comply in changing or removing legislature that is ruled unconstitutional, the courts have little means to change this.
They therefore also rely on the public: the public can pressure and influence the legislative into following the court’s rulings. This works both ways: the public can also pressure the legislative into dismissal of the ruling or do nothing at all (Mondak and Smithey, 1997). Compliance and public support are difficult to balance. Each branch of the government needs the public on their side. For the executive and legislative, this is a very straightforward concept: politicians need their constituents to vote for them in election times and therefore need to ensure that the public agrees with their actions. The judiciary is not always elected by the public, and in this case building a strong institutional legitimacy becomes increasingly harder (Caldeira and Gibson 1992). Additionally, politicians who feel pressure from constituents may refuse compliance in order to retain votes. However, because the other branches are so reliant of the public, the judiciary can utilize this reliance in their favour.
All three branches must make difficult choices, and there will always be parts of the public that disagree with the current course. It is vital for each that they find balance in their policies and course of action and that they build a sound reserve of goodwill to cushion a potential blow (Caldeira and Gibson, 1992). If a court goes too far against public opinion by being too progressive or conservative, they risk losing support from a large portion of the public. If they are too careful, however, they might not address important questions that the judiciary is needed for (Mondak and Smithey, 1997)1.
Non-compliance by the legislative branch can have unwanted effects on public image for both the legislative and the judiciary. The legislative risks electoral consequences when publicly defying a court’s rulings, especially if that court is held in high regard by the other branches and the public (Staton, 2005; Staton and Vanberg, 2008; Sternberg, 2018; Vanberg, 2006). The judiciary risks loss of face in the case of public defiance, which in turn may lead to more defiance by other policymakers as their influence as policymakers erodes (Carrubba 2005; Staton and Vanberg 2008). Non-compliance and lack of public support enforce one another and can lead to a downwards spiral (Carrubba, 2005; Staton and Vanberg, 2008). A court that does enjoy public support, however, can use this support to ensure compliance. The public, who are aware of any non-compliance, would lose support for the legislative instead. Therefore, the other branches would not dare to defy the court in order to keep voters in their court which then leads to more support for the court as they have now successfully enforced their ruling (Staton and Vanberg, 2008).
Judiciary as Policy-seekers
Non-compliance and public support would not be relevant for the court if they were not actively trying to actively create or shape legislation. Traditionally, courts were viewed as external, passive actors in the policy-making sphere, but recently research has pointed out that they are active participants in the ‘game’ in which they play (Volcansek, 2001). The judiciary
1 In R.G. & G.R. Harris Funeral Homes Inc. v. Equal Employment Opportunity Commission (590 U.S. ___ (2020)), the question was raised whether termination of employment on the basis of sexuality or gender identity is constitutional. After oral arguments, Justice Gorsuch stated that the workers had good arguments, but that the court would have to take into account the possible “massive social upheaval” following a certain judgement. This comment sheds light on the possibility that certain orders may not be made by the courts in order to appease the public.
will actively try to interpret law according to their own best interest as well as try to eliminate the possibility of their decisions being overturned (Tsebelis, 2011). Constitutional courts and their judges have been described as the ‘third chambers of parliament’ (Hönnige, 2009) and ‘veto-players’ (Tsebelis, 2011; Volcansek, 2001) within the political system. They are key players in creating, shaping and especially restraining the legislative process. Hönnige (2009) has tested the degree in which the composition of courts matters in the outcome of their decisions, and concludes that indeed, judges are active policy-seekers.
Courts with a common law system and courts with a law system based on civil code are often compared concerning their influence on policy. States with common law systems (the United Kingdom and their former colonies and commonwealth constituents) value precedent as the most important basis of the law (Tsebelis, 2011). Laws in common law systems are largely an accumulation of court decisions instead of parliamentary decisions while civil law systems are mostly based on Germanic or Napoleonic Code, which consist of structured codes and legislature on which statutes are built (Tsebelis, 2011). While the common law tradition is said to create, apply and interpret the law, civil law system is explained as only interpreting the law in their decisions (Tsebelis, 2011). Although it would seem that judges in common law courts have more influence as policymakers, Tsebilis (2011) argues otherwise. The courts in civil and common law tradition countries have similar influences the creation and shaping of legislation and policy. What matters is whether the court has the possibility and capability to actively influence legislature in the country – something that courts with the right of constitutional review have by default – not their law tradition (Tsebilis, 2002; Tsebilis, 2011). Constitutional courts review legislature against a states’ constitution to determine whether that law can be applied. These courts have the right to judicial review, which is the power the judiciary has order to check the executive and legislative branches of governments. As courts and judges are active policy-seekers, they may actively and strategically review the laws that pass their review in constitutional court.
The way in which the judiciary is structured and the way it interacts with the legislative as well as executive actors ensures that its “independence cannot be assumed” (Whittington, 2003). The political affiliation and personal preferences of the courts are pivotal in court decisions (Hönnige, 2009; Whittington, 2003) which leads to the opinions of individual judges affecting the adoption or rebuttal of laws or policy and shape the legislature of the state they operate in. This is regardless of the system through which judges are appointed (Hönnige, 2009). Beneath all, judges are people who operate within their own frame of reference,
influenced by their culture, political affiliation, religion, and other individual aspects. These biases are more often than not implicit, but can still hold great consequences, such as increased length of a sentence (Biwer, 2019; Sorensen et.al., 2012). Recent studies have reiterated that bias is still present in courts today, whether racial (Biwer, 2019), socioeconomic (Neitz, 2013), or for example previous criminal records. Additionally, judges in certain systems have to adhere to political influences. In the Unites States, for example, Supreme Court Justices are appointed by the President and are often affiliated with that President’s party. The appointment process is highly politicized and afterwards these Justices are expected to rule in accordance to party policy (Epstein et.al., 2006). Hönnige (2009) examined whether a judge is that is appointed through elected politicians (e.g. SCOTUS) or through a panel of their peers (e.g. the GFCC) would be more influenced by their political affiliation, but found that judges appointed through any method are influenced by this. This means that individuals, their ideologies and their (implicit) biases can have far-reaching effects on citizens, considering that constitutions contain the fundamental principles on which a country is built, as well as basic human rights such as free speech, equality, and fair trial.
Challenges for the Judiciary
The influence that the judiciary has is not self-evident nor entirely comprehensive. In practice, there are many challenges for judges in their role as policymakers. Firstly, legislature and policy often require highly technical and specific knowledge about certain issues (Staton and Vanberg, 2008; Vermuele, 2000). The courts and their judges have many different areas of law and policy in their docket for review, and it is impossible for them to be experts in every single case. The courts often know what outcome they desire, but not necessarily how to achieve that outcome (Gilligan and Krehbiel, 1990; Staton and Vanberg, 2008). In the case of judicial review this means that courts and their judges can assess whether a law is unconstitutional but lack the expertise to explain what a suitable alternative would be. Staton and Vanberg (2008) give an example concerning racial school segregation, in which the court ordered that schools would be integrated but did not advise on the way this should be done or give a timeframe in which integration should be complete. The court did so because they did not have the expertise to judge what timeframe or what measures were reasonably achievable by the schools. Although the non-precise judgment was given to give the schools some discretion, it also paved the way for a lot of resistance by school districts (Staton and Vanberg, 2008). Judicial decisions are not self-enforcing, and “compliance cannot be taken for granted” (Staton and Vanberg, 2008). In order to avoid non-compliance, courts may decide to issue very specific rulings. However, in
combination with a lack of expertise in the policy area, courts risk ‘locking in’ inappropriate policy that has undesired or even detrimental effects (Sternberg, 2018). The courts must find a balance between these risks of non-compliance as well as locking in unsuitable policy.
Vagueness as a Tool
Very specific rulings and opinions ensure a higher visibility of non-compliance by policymakers (Staton and Vanberg, 2008). The public has the possibility to notice that the other branches ignore the court’s judgment. If the wording is vague, however, it cannot be said that the legislative explicitly ignored the judiciary’s opinions. Non-compliance is less visible when judicial opinions contain high degrees of vagueness, which reduces the risk of losing face for the legislative policymakers. As illustrated with the school segregation example, non-specific rulings can cause non-compliance by the legislative as well as executive branches. Although some policymakers defy rulings despite their specificity, vagueness increases the chances. This has long been argued by scholars such as Baum (1976) and Johnson (1979), and again stressed by Staton and Vanberg (2008) and Sternberg (2018). If vagueness increases the risk of non-compliance, why do judges utilize vague language in their rulings?
What exactly causes wording to be vague? Before determining how vagueness is used, vagueness itself needs to be conceptualized. Vagueness is traditionally described as the absence of clear boundaries (Hyde 2007; Keefe, 2000). The descriptor ‘tall’, for example, is highly depended on context. A Dutch person who measures 1.70m is not considered tall, while a 1.70m Indonesian might be. It has close connection to other non-specificity concepts, such as ambiguity or generality. Often, wording is both vague and ambiguous, or both vague and general (Sorensen, 2018). Ambiguity refers to wording that can be interpreted as different things (e.g. ‘bank’ for both the monetary institution and the side of a river), and generality refers to overarching words that can encompass multiple subgroups (e.g. ‘children’ for ‘boys’ and ‘girls’). Vagueness is different in a sense that it is not fixed. The word ‘bank’ has several meanings, while ‘tall’ does not. Vagueness is an inherent feeling that one gets from the lack of clear pre-set boundaries. It is a lack of ‘logic’ and rules. Most importantly, vagueness is the lack of clear, sharp boundaries (Hyde, 2007; Keefe, 2000; Sorensen, 2018). In other words, if a phrase indicates that something is not certain or leaves room for interpretation, this is vague. This can come in many forms: terms like ‘uncertain’, ‘potential’ and ‘possibly’ indicate vagueness, but so do terms like ‘anytime’, ‘as fast as possible’ and ‘long’. The term vagueness is, of itself, very vague. In recent philosophical and linguistic research, vagueness has been attributed to many types of words and phrases. Words like ‘tall’ or even ‘red’ are not as
semantically determined as they may seem and are open for interpretation. A common example is that of the heap of sand: if you see a heap of sand comprising of thousands of grains of sand, and you take one grain off, the heap is still a heap. You can repeat this indefinitely until there is no sand left. The paradox concerning the heap is at what amount of grains the heap is still a heap, and at what amount it is not yet anymore. There are many more of such paradoxes in the world: From what amount of hairs is a person bald? And at what age are they ‘old’? Some words create vagueness in a phrase, such as ‘being ‘almost’ home’ or ‘being ‘potentially’ dangerous’ (Keefe 2000). Vague statements structure a broad range of interpretation, while concrete statements structure a narrower range of possible interpretations (Eichorst and Lin, 2019; Sternberg, 2018).
One of the reasons for the judiciary to give expert policymakers the discretion of designing and implementing appropriate policy and to avoid locking in unsuitable policy. Staton and Vanberg (2008) introduce another strategy in which courts use increasingly vague language in their game-theoretic model. Especially courts that have a low degree of public support or have not been around for a long enough time to establish legitimacy within the country may opt for the use of vague language in order to protect their image. Alternatively, popular courts with high public support or courts that have built strong institutional legitimacy would be more specific in order to urge the legislative into compliance (Staton and Vanberg, 2008). In short, Staton and Vanberg’s (2008) model entails that if there is a sufficient risk of non-compliance from the legislative and/or executive, unpopular (constitutional) courts will use increasingly vague language, while popular courts will use increasingly specific language. To sum up: in order to protect their public standing, courts that have low degrees of public support will use vaguer language if there is sufficient risk that their orders will not be complied to. On the other hand, courts that do have broad public support will use less vague language in their orders if they expect non-compliance, in order to make the public more aware of their reasonings and put more pressure on the other
branches to comply. Figure 1 offers an elementary illustration of the relationship between non-compliance risk and vagueness and how the degree of public support of a condition for this relationship to be positive or negative.
Vag u en ess Non-compliance risk
Figure 1: an illustration of
the trade-off hypotheses
Court with high degree of public support Court with low degree of public support
Courts must be careful in balancing their use of vagueness: by being too vague, they actively increase the risk of non-compliance by the rest of the government and hurting their reputation with the public. By being very specific, they increase the chance of compliance but risk losing important public support if they are still faced with non-compliance. Sternberg (2018) has tested the model on two European courts: the German Federal Constitutional Court and the French Conseil Constitutionnel. His results support Staton and Vanberg’s (2008) model.
For each case in this thesis, what constitutes a ‘sufficient’ risk of non-compliance is assessed separately on the basis of each country’s policy in which the government can declare their disagreement with the court. In the United States, for example, State and Federal governments can file amicus briefs in support of their laws. Each country has different ways in which the legislative and/or executive government can voice their dissent regarding court cases or support regarding their legislation. The method of assessment in this case can therefore never be completely similar. This could have an effect on the outcomes of this thesis. However, cross-national comparative studies such as these will inherently have some varying approaches, and for each court the most suitable methods of assessment were selected. I go into more detail on this topic in the research design section of this thesis. The term ‘sufficient’ is necessary in the assessment of the non-compliance risk as there always is some – however small – degree of non-compliance risk.
The following hypothesis summarizes these findings and will be the guideline of this thesis. Is has been derived from Staton and Vanberg’s (2008) and Sternberg’s (2018) hypotheses. In the following chapter I will operationalize these concepts and explain my methods of measurement and analysis.
Hypothesis 1:
Given that there is a sufficient risk of non-compliance, a court with a high degree of
public support will have a lower degree of vague language in their decisions.
Hypothesis 2:
Given that there is a sufficient risk of non-compliance, a court with a low degree of
Research Design
Case Selection
In order to test the generalizability of Sternberg’s (2018) research and Staton and Vanberg’s (2008) model, the case selection must meet the same criteria that Sternberg’s (2018) cases have. Sternberg (2018) identified three conditions that courts must meet in order to test Staton and Vanberg’s (2008) model. Firstly, all analysed courts must have the right of judicial review. Without the right of judicial review, “the game between court and legislative majority could not be played” (Sternberg 2018, 10). Courts without the right of judicial review cannot review laws or decisions by the legislative and can therefore not have a significant impact on policy. Secondly, the judges of these courts must be regarded as policy seekers. Staton and Vanberg’s (2008) model hypothesizes that the courts actively seek to influence policy and that the degree of vague language in court opinions and judgements may be done intentionally. Thirdly, the courts must vary in their degree of public support in order to test whether this affects the degree of vague language.
For my research, I selected courts based on the same criteria that Sternberg (2018) applied to his case selection. I decided, in order to properly test generalizability, to select courts that meet the same essential criteria but that vary on others. I selected courts in a different language from Sternberg’s (2018) analysed courts (English vs German and French) and in different parts of the world (Africa and North America vs Europe). The law tradition is also different: the courts in this research are from a common law tradition, while Sternberg’s German and French courts come from a civil law tradition. For my research, I have selected the following courts:
◼ Supreme Court of Canada (SCC) – high public support
◼ Constitutional Court of South Africa (ZACC) – low public support ◼ Supreme Court of the United States (SCOTUS) – medium public support These three courts all satisfy the criteria as identified by Sternberg (2018). A detailed evaluation of the three courts against Sternberg’s (2018) criteria can be found in Appendix I.
Data collection
My data set consists of self-collected data retrieved from each country’s legal database. The data set contains all online published decisions from 2009-2019 of the SCC, SCOTUS and ZACC that review the constitutionality of local or national laws. I have chosen for the past ten
years in order to compare the three courts in the most accurate way. The ZACC has only been established in 1994 and its first few years of existence were full of turmoil in post-apartheid South Africa. As Gibson et. al. (1998) explained, a newer court might struggle with establishing institutional legitimacy regardless of the types of cases and the risk of non-compliance. Additionally, although public support for institutions is mostly stable, it does vary over time. By limiting the data to a span of 10 years in which the degree of public support for the courts is the same the results of this research will be more accurate. The data was collected through the websites of the courts and assessed for suitability by reading the judgements as well as other legal sources and databases that provide data on the cases (e.g. SCOTUSBlog, which keeps a record of all supreme court cases, filed briefs and statistical data). Staton and Vanberg’s (2008) model is only applicable to decisions in which a law was rejected and therefore - as Sternberg (2018) did – I omitted the decisions in which laws were deemed constitutional. The final number of decisions in the data set is 153.
Operationalization
Measuring Vagueness
For this thesis, vagueness is defined as the absence of precision, certainty and specificity, and most importantly the absence of boundaries. As explained in the theoretical section of this thesis, when the borders of a term are unsharp or subjective, it is vague (Hyde 2007; Keefe, 2000; Sorensen, 2018). This way, most words and phrases can be assessed as vague, since most words do not have a predetermined set of boundaries or meaning attached to them. In a sense, language can never truly be specific and is always reliant on the context in which it is used. Practically speaking, however, some words do have pre-attached meanings and boundaries within certain groups or domains. In law especially, certain definitions can be predetermined by a court or even a country’s complete legal framework. Additionally, some words can make entire sentences or texts be vague due to their specific nature (Keefe 2000). For this thesis, the focus lies on these types of vagueness-indicators. Words like ‘reasonably’, ‘some’ and ‘either’ indicate that a certain boundary is not clear, regardless of whether specific words do have sharp boundaries. While words like ‘red’, ‘tall’ or ‘heap’ can be vague due to a lack of context or knowledge of the existing boundaries, words like ‘any’, ‘apparent’ or ‘nearly’ always indicate absence of sharp boundaries. Sternberg (2018) focusses on these type of words, that “modify the gradeability of interpretation” of a text. These words are used to indicate that there is a wide range of possible interpretations, and can be adjectives, nouns, verbs or prepositions. It is these types of vague words that this thesis will focus on.
The law and language are intertwined, and a large section of constitutional law judgements comprise of an assessment and interpretation of contended laws and policy as well as the constitution itself. Vague terms lead to a an even wider scope of possible interpretation and therefore increase the risk that a preferred outcome is not reached (e.g. compliance or non-compliance). In order to replicate Sternberg’s (2018) methods as closely as possible, I follow the way in which they operationalized and tested for vagueness. Sternberg (2018) conceptualizes a vague decision as the “strategic use of vague word choices to modify the informative structure of court rulings in order to influence how legislators interpret such terms” (Sternberg, 2018, p. 12). In other words, the presence of vagueness indicator words signals that the decision itself is vague. A higher percentage sentences that contain vague wording means a vaguer decision overall. In order to find vague wording and phrasing, I will run the court decisions through textual analysis programme LIWC (Linguistic Inquiry and Word Count). LIWC is a programme that counts words and types of words using dictionaries. LIWC has its own list of words that indicate vagueness for most languages. Sternberg (2018) pointed out that this list only includes single words and not phrases, and that it might lack domain-specific vocabulary. In order to correct for this, Sternberg (2018) used word-embeddings to find semantically similar words. He specifically uses the word2vec word embedding model that can be implemented in different types of programming. With this model, the words that are semantically closest to the LIWC dictionary list of vague words can be found. The word2vec model makes a vector space, in which similar words are placed in the same vector. This is done by finding the wording that most often surround these vague terms within the data set and then looking for terms that can be found in the same or similar wording. Sternberg (2018) then considered the top 15 closest words for each LIWC dictionary vague term and eliminated any that were not suitable.
Word2vec requires extensive knowledge of programming languages such as Python. Programming does not lie within my scope of knowledge, and thus I have opted for a different method to expand the word list. Since word2vec is a tool that is used to find synonyms, near synonyms and antonyms, I have chosen to expand my wordlist through old-fashioned dictionaries and thesauruses. Although this is a different method to Sternberg’s (2018), it adequate to find fitting words. Additionally, word2vec as a model still requires extensive manual evaluation of the proposed words, as it often proposes antonyms as more similar than synonyms, among other things (Theil et.al. 2009). I have used multiple dictionaries, general thesauruses, lexical databases and a specific legal thesaurus and dictionary in order to assure
domain-specific words are included2. The original word list that LIWC provides in their standard dictionary contained 154 tentative words, and the extended list contains 502 words and phrases. The courts’ judgements are split into sentence-long segments in LIWC and for each segment the programme finds if and how many words of that segment contain words that are in the wordlist. This way, the percentage of sentences that contain vague wording can be determined.
Measuring Non-Compliance Risk
Sternberg stresses that the measurement of the perceived risk of non-compliance must be alike to the assessment that the judges may have made while making their decision (Sternberg, 2018). This means that the non-compliance risk must be measured using tools and information that the judges had available to them at that time. For the German and French courts that Sternberg (2018) tested, he found country-specific ways in which the government can communicate their disagreement with a case before the ruling. In Germany, this is done through government briefs, and in France through press-releases. Each country has different laws, rules, and regulations for dealing with these kinds of cases. For the Unites States, Canada, and South Africa, the ways in which the government can express disagreement varies greatly.
In the United States, direct parties cannot file an amicus brief as is done in Germany, but other parts of the government can. States often file amicus briefs in support of other states’ laws or in support of the federal government. The federal government, in turn, often files briefs in support of state laws. For federal laws that have been challenged, I examine for States that have filed briefs of amicus curiae. For State laws that have been challenged, I examine for a brief of amicus curiae filed by the United States and briefs by other States that are filed in support. As Sternberg (2018) points out, these briefs are rarely ever real threats of non-compliance, but they indicate a certain degree of investment by the government in the legislature and desires to keep it as it is, especially since these briefs are public and extensive. Presence of such briefs is coded as 1; absence is coded as 0.
2 The used dictionaries and thesauruses are: Burton’s Legal Thesaurus; Collins English Thesaurus; Longman’s Dictionary of Contemporary English; Merriam Webster Dictionary of Law; Merriam Webster Thesaurus; Oxford Dictionary of English; Thesaurus.com (https://thesaurus.com); Visuwords visual dictionary
In Canada, amicus briefs refer only to invited or commissioned parties that were asked by the court to present their viewpoints on the matter. These parties are often experts in a certain field or otherwise ‘neutral’ to the case between the applicant and defendant. Canadian law does allow intervenors to present their viewpoint on the matter. Intervention is largely similar to the amicus curiae process in the US except it only refers to parties than intervene without explicit permission from the original litigants. Parties that wish to intervene can do so if they can show that they have an interest in the matter and wish to voice their opinion or additional facts. Similar to the United States amicus briefs, federal government may intervene in provincial law cases, and provinces may intervene for the sake of other provinces or the federal government. To measure non-compliance risk for SCC cases, I examine whether any Province or Federal intervention has been filed in defence of the legislation. Presence of government intervention is coded 1; absence is coded as 0.
The South African Constitutional Court does not allow for the government to intervene or file amicus curiae, although other parties may. Instead of a possibility to actively voice disagreement, the government has the possibility to actively voice agreement. The ZACC, being an exclusively constitutional court, often hears cases that were referred to them by the South African Supreme Court or other lower courts for confirmation of their declaration of a law being constitutionally invalid. Often, the government bodies related to the contested laws voice their agreement through a notice to abide. They still file merits on the case and are heard in court, but they proclaim that they will not contest the outcome. Absence of a notice to abide is coded as 1 (1 being the presence of a sufficient risk of non-compliance) and presence of such notice is coded as 0 (0 being insufficient risk of non-compliance).
It is important to be aware that the variation between these methods to assess risk of non-compliance may have an influence on the outcome of my research. Especially the contrast between voicing active disagreement versus voicing active agreement is something to concern. For a body of government that actively takes the time to file extensive amicus briefs in order to voice disagreement, that disagreement is clear. A body of government that voices disagreement by not doing anything at all lacks that image of determination in their disagreement. Although voicing active agreement is something different from voicing active disagreement, the notices to abide are relatively short documents compared to the amicus briefs and intervenors documents that indicate disagreements in SCOTUS and the SCC. They comprise of only a signed statement in which they proclaim to abide the ZACC’s order. Presence of a notice to abide occurs in more cases in the ZACC than absence of interveners or
amicus briefs is seen in SCOTUS and the SCC. Therefore, in light of their being a lack of any other (consistent) way to assess the risk of non-compliance for ZACC cases, the presence or absence of notices to abide is the most suitable way to do so. The SCOTUS amicus briefs and the SCC intervenors work in a mostly similar way, and although they have different names and different official functions the different briefs are similar in nature and content and have a similar possible influence on the final judgement.
Reflection on Validity and Reliability
Through these methods of case selection, data collection and analysis I strive to test Staton an Vanberg’s (2008) theoretic model as well as test the generalizability of Sternberg’s (2018) application of this model. By emulating Sternberg’s (2018) methods as closely as possible and utilizing the same programmes I increase the reliability of my research. Since I am using a computer programme (LIWC) these aspects of the analysis are replicable by others that wish to investigate vague language in judicial opinions. However, I must add that – as indicated by Gibson et. al. (1998) and Sternberg (2018) – assessment of public support for a court is not an exact science. Although theories can appropriate the expected response of the legislative to a court’s opinion as well as estimate the behaviour of courts and their individual judges, these institutions are comprised of people who will not always behave rationally or in line with theory. Each individual court case brings individual challenges and factors like public awareness of that specific case as well as the current political situation in a country may affect the way that courts produce their opinion. Each country and court has their own way of filing documents and briefs, which is also why the differences in measuring non-compliance risk are necessary. Additionally, the culture of the country in which the courts reside may affect the degree of vague language use. Local dialects or language may have found their way into the legal language or some cultures may have a tradition of using more vague wording in their language than others. In my analysis, I will have to keep these challenges and factors in mind and not take all results for face value. I will elaborate further on this in both the results section as well as the discussion section of my conclusion.
Results and Analysis
Each of the 153 court cases were individually run through LIWC and analysed per sentence-long segment. For each segment, LIWC compared the words and phrases to those on the extended wordlist of words and phrases that indicate vagueness. On the basis of this data, the percentage of sentences that are vague could be determined per court and presence of non-compliance risk. In this section, I will first visit the primary data on the judgements as found during the data collection and given by LIWC. Secondly, I will elaborate on the results in relation to three different levels of analysis: inter-judicial relationships, relationships between the judiciary and the executive and legislative, and relationships between the judiciary and the public. Appendix II contains a list of the collected data and the initial results for further reference.
Primary Results
The primary results as presented in Table 1 give some valuable insights. Firstly, I considered the amount of sentences that each court case consisted of. This was partly necessary in order to be able to calculate the percentage of sentences that contained vagueness, and partly
Table 1 – primary findings of analysis through LIWC
Average amount of sentences per judgement Average amount of words per sentence Average amount of vague sentences Average amount vague sentences in % of total
SCOTUS - sufficient risk of non-compliance 768.54 10.74 173.35 22.68
SCOTUS - insufficient risk of non-compliance 511.26 11.17 129.84 24.14
SCOTUS - total 639.90 10.96 151.60 23.18
SCC - sufficient risk of non-compliance 1298.96 11.11 250.74 19.21
SCC - insufficient risk of non-compliance 1057.50 11.71 231.25 20.76
SCC - total 1178.23 11.41 240.99 19.44
ZACC - sufficient risk of non-compliance 432.42 25.95 196.02 44.35
ZACC - insufficient risk of non-compliance 408.00 25.83 176.45 44.03
ZACC - total 420.21 25.89 186.24 44.30
Total - sufficient risk of non-compliance 704.41 18.35 199.55 32.76
Total - insufficient risk of non-compliance 542.12 15.98 156.85 30.18
in order to gain insight in other differences that the three courts might have in giving judgements and orders. There are large differences between the average length of the judgments. The SCC has the averagely longest judgements while the ZACC has the shortest. Although there is not yet enough data to draw hard conclusions on this, it is notable that most popular court would write the lengthiest judgements. Longer pieces of written text can be expected to be more difficult to comprehend by the reader (Bell and Perfetti, 1994; O’Connor et.al., 2010). For each of the three courts, judgments of cases in which there was a sufficient risk of non-compliance were longer than in those in which there was not a sufficient risk. Since this is the same regardless the amount of public support for the courts, it seems that there is no relationship between the popularity and the amount of sentences. The larger amount of sentences in cases with sufficient risk of non-compliance might have different causes, such as the need for extra explanation – making the judgement less vague although it is longer.
Secondly, I considered the average amount of words in each segment. Although the ZACC has the least amount of sentences, their sentences are incredibly long with an average of almost 26 words per segments compared to the SCC and SCOTUS, who fall around the same average of 11 words per segment. The reason for the ZACC’s longer sentences could be cultural: South Africans might be predisposed to using long sentences instead of short ones due to local linguistic developments and influences from other languages such as Afrikaans or Dutch. However, there is not enough research that supports the idea that South African English is predisposed to longer sentences in general. It could also be due to the nature of the ZACC’s judgments: the ZACC’s judgements comprise largely of constitutional interpretation, and these statutes and laws are lengthy pieces of text, that are often comprised of only one or two long sentences that contain many clauses. It is important to note that due to the longer segments, the chance that a vague word or phrase is present in a segments is notably higher. Sternberg (2018) encountered the same phenomenon with the cases of the French Constitutional Council, and like them I stress that the findings regarding the ZACC are therefore preliminary and should not be taken as-is.
The primary results support the hypotheses of this thesis. As Table 1 presents, the SCC and SCOTUS, which enjoy high and medium degrees of public support, use less vague language than ZACC, which has low public support. Additionally, the SCC and SCOTUS use less vague language in cases with sufficient risk of non-compliance, while the ZACC uses a higher degree in these cases. In the next part of this chapter, I will elaborate more on the degree of vagueness
and what the implication of these results are for the relationship between the judiciary and other actors.
Vagueness and Public Support
Table 1 shows that a court with higher public support will write judgments that are averagely less vague than courts with lower public support. The SCC, which was evaluated as a court that enjoys high degrees of public support writes judgments that consist for 19.98% of vague sentences. The SCOTUS, which was evaluated as a court with medium levels of public support comes relatively close to the SCC – SCOTUS judgments contain 23.41% vague sentences. The ZACC, which was evaluated as a court with low degrees of public support, is the vaguest of the three, with its judgments containing 44.19% sentences that are vague. These results support the existing theories that courts use vaguer language if they are unpopular but more specific language if they are popular with the public in their jurisdictions.
The use of different language has an effect on the public and how it perceives the judgments of the court and the court itself. Very specific and straightforward judgments as well as shorter judgments or judgments with less legalese and other difficult phrases are easier to understand for a broader public. This increases the access the public has to the legal system and their understanding of the processes, their rights and obligations. Vaguer and/or longer or more difficult judgments will have the opposite effect on the public. Whether or not the public understands the courts can be beneficial to the court. As explained in the theoretical framework of this thesis, the public has a huge role in policy-making in that they directly influence the executive, legislative and sometimes also judicial branches through elections. In the United States, Supreme Court Justices are appointed by the President and the Senate, which are both elected offices. In Canada, the SCC judges are also appointed by the prime minister, and in South Africa, the ZACC judges are appointed by the president. Higher understanding of the language in a court’s judgments may also lead to more or less support of actual orders in the judgments. If the public understands what exactly a court is doing and what the viewpoints of the individual judges are, they might choose to vote for a political party that affiliates with the public’s stances on the judiciary. Through the election of executive and legislative officials, the composition of the court may be influenced and the outcome of future cases may be more or less conservative/progressive or left-wing/right-wing. A court that uses vaguer language may shield themselves from criticism by the public, but also shield themselves from support. Vice versa, a more specific judgment may entice public outrage when dealing with controversial topics which could hurt the court’s public support.
The result is a need for a delicate balance between specificity and vagueness in order to keep or improve the court’s degree of public support. An already unpopular court might hurt itself even further if it is too specific in yet another controversial case, but may also miss opportunities if they are too vague and therefore also shield the public from decisions that are made in line with the current public sentiment. If not careful, a downwards spiral may occur in which a court with lack of public support writes increasingly vague decisions which leads to less public support which leads to vaguer decisions – and so on. The other way around, the courts with high degrees of public support can take the risk of being controversial and straightforward instead of vague and therefore use the higher visibility to their advantage.
Vagueness and the Legislative and Executive Branches
The courts are influenced by public support and may use it to protect their standing with the public. However, the interaction between the courts and the public is not a self-contained process. The judiciary, the public (including companies and other non-governmental institutions) and the executive and legislative branches of the government continuously interact on many levels. Through the aforementioned interaction between the courts and the public, the courts are also in a position to influence the other governmental branches, as well as in a position to be damaged by said branches. In this part of the chapter, the results are analysed against the relationship between the judiciary, the legislative and executive, and the risk of non-compliance by the latter.
The relationship between the judiciary and the other branches is an important factor for the public support a court may receive. Additionally, the way in which the other branches interact with the courts is an important factor when it comes to their degree of public support. All three interact and are heavily reliant on one another. If a court makes a decision, the legislative and executive branches often have some sort of way to interpret that decision and can choose to (partly) not comply with a court’s order. However, by not complying to orders by a court that has high public standing, these two branches might expect some degree of public discontent or even outrage. Vague language in judgments enlarges the leeway the legislative and executive have to do so, since the public then has less awareness of the judgment, or these two branches could argue that they ‘technically’ adhered to the order in some way. The executive and legislative have to be careful not to lose their public standing with their constituents or be voted out in the next election. Similarly, a court which orders are not complied with may lose some legitimacy in the eyes of the public. Again, a delicate balance between vagueness and specificity is needed for the judiciary, while the other branches have to
be careful as well. As explained in the theoretical framework of this thesis, courts with high degrees of public support may take the risk of non-compliance: their public standing lessens this risk as the public will likely take it out on the elected officials instead. Courts with low degrees of public support will try to increase their public standing and try to avoid (visible) non-compliance. This is one of the main factors for the increased use of vague language in unpopular courts as opposed to the decreased use of vague language in popular ones. If there is a large enough risk of non-compliance, this effect will be even more visible. In the Figure 2 as well as Table 2, a representation of the amount of vagueness can be found in regards to the risk of non-compliance.
The primary data already saw a difference in vagueness between the courts with lower, medium and high public support.. Figure 2, a boxplot graph visualises the primary data, including not only the averages but also the outliers. The boxplot presents the results in the following way: everything between the T-shaped brackets represents 99.3% of the data for each court. The line within the box is the median amount of vagueness, while the cross is the average amount of vagueness. The box itself represents the interquartile range: the percentages of vagueness that occur the most, around the median, with less regard for the minimum and maximum outliers. As Figure 2 shows, the SCOTUS and SCC results have quite some overlap, while the ZACC does not. Their range of degree of vagueness is largely similar. The primary results and this boxplot confirm the hypotheses, but the reality is a bit less crude than this graph
makes it seem. Although the ZACC does indeed write with a high degree of vagueness in cases with a sufficient risk of compliance, they also do so in cases with insufficient risk of non-compliance. If South African courts are more vague by nature, regardless of their degree of public support, then this does not confirm the hypotheses of this thesis. What matters more is the difference between sufficient and insufficient non-compliance risk. Table 3 shows not only the percentage of vague sentences for the courts, but also the difference in vagueness between cases that have sufficient and insufficient risk of non-compliance. This difference is shown in percentage points as well as in percental increase or decrease.
Table 2: vagueness in the courts % of sentences that contain vagueness – insufficient risk of non-compliance % of sentences that contain vagueness – sufficient risk of non-compliance Difference between sufficient and insufficient risk of non-compliance – percentage points Difference between sufficient and insufficient risk of non-compliance – % SCOTUS 24.14189 22.68412 -1.457771 -6.03834 SCC 20.75823 19.20643 -1.551807 -7.47562 ZACC 44.03411 44.35276 0.318649 0.723641 all courts 29.64475 28.74777 -0.896976 -3.02575
The third and fourth column in Table 2 shows that the SCC, on average, has the largest difference between cases with insufficient versus cases with sufficient risk of non-compliance. Compared to insufficient risk cases, sufficient cases are 1.55 percentage points or around 7.5% less vague. This confirms Hypothesis 1 which posed that courts with high public support would write increasingly specific judgements.
SCOTUS also writes increasingly specific judgements in cases with sufficient risk. For SCOTUS, this is 1.45 percentage points or around 6% less vague than in cases with an insufficient risk of non-compliance. Since I assessed SCOTUS as a court that enjoyed medium support, there were little expectations as to whether SCOTUS would have more in common with courts that have high support or courts with low support. These findings reveal that SCOTUS falls within the scope of Hypothesis 1 and has more specific judgements in cases with a sufficient risk of non-compliance. SCOTUS has, still, less public support than the SCC, and this is visible in the fact that the difference in specificity between sufficient and insufficient risk of non-compliance cases is higher for the SCC than for SCOTUS.
It has already been established that the ZACC writes the cases with the highest degree of vague language of the three courts. Since ZACC is a court that enjoys low public support, the expectations were that the judgments would be increasingly vague in cases with sufficient risk of non-compliance. The results confirm this expectation. The difference between cases with sufficient and insufficient risk of non-compliance is less significant than for SCOTUS and the SCC, but still relevant. On average, the ZACC writes around 0.32 percentage points vaguer rulings if the risk of non-compliance is sufficient, which is about 0.72% vaguer compared to when it does not fear non-compliance. This is a very small effect compared to the effects that can be seen in the SCC and SCOTUS. Although these results confirm the expectation in the hypotheses, the effect in the results is so small that it cannot be taken at face-value. At the end of this chapter, I discuss the statistical significance of the results and whether they confirm the hypotheses.
Intra-Judicial Vagueness
Another interesting factor that was revealed in the data collection process was the amount of contention within the courts concerning the outcome of the cases. All three courts offer their judgements and orders through the majority judgement. This majority is sometimes unanimous, but a 50% +1 simple majority is enough in the case of these courts to be accepted as the final order. Judges may file alternative concurring judgements, meaning they agree with the final outcome but disagree on the reasoning. They may also write and file dissenting judgements, in which they propose their alternative reasoning and outcome. During the data collection process, I noted the amount of judges ‘for’ (majority and concurring) and ‘against’ (dissenting) the majority judgement outcome. If a judgement was not made unanimously, this means that within the court there was disagreement over the ruling. This may have an effect on the risk of non-compliance, as government officials and administrators could use this contention in their argument against compliance. Simultaneously, there seems to be a relationship between the risk of non-compliance, the amount of vagueness and whether the court was unanimous or not. For the SCC and SCOTUS, the percentage of unanimous rulings is lower in cases with sufficient risk of non-compliance. The SCC and SCOTUS are also more specific in their rulings of cases with sufficient risk of non-compliance.
Table 3 presents the percentage of vague sentences for each court in relation to the degree of agreement amongst the judges. For this, I have separated the individual court cases on both risk of non-compliance and amount of contest. Unanimous rulings and orders ‘per curiam’ or ‘by the court’; rulings with a large majority support in which only one or two judges
dissented; and cases in which only a small majority of judges supported the ruling. The results are presented as following: for each court, the cases with sufficient risk of non-compliance are followed by the cases with insufficient risk as well as a total. From left to right the type of majority is represented. The results are average percentage of vague sentences within court cases that fall within that category.
Table 3: Amount of Vagueness in relation to agreement between judges
Unanimous Large Majority Small majority SCOTUS - sufficient risk of non-compliance 25.646 22.334 22.319 SCOTUS - insufficient risk of non-compliance 17.040 27.001 28.362
SCOTUS - total 21.343 24.667 25.340
SCC - sufficient risk of non-compliance 19.625 17.839 n/a SCC - insufficient risk of non-compliance 22.819 14.575 n/a
SCC - total 21.222 16.207 n/a
ZACC - sufficient risk of non-compliance 44.371 44.891 40.722 ZACC - insufficient risk of non-compliance 44.307 43.306 n/a
ZACC - total 44.339 44.099 n/a
Total - sufficient risk of non-compliance 29.880 28.355 n/a Total - insufficient risk of non-compliance 28.056 28.294 n/a
Total 28.968 28.324 n/a
The first thing that can be noticed is that over-all, unanimous courts utilize more vague wording in their judgements than courts with a large or small majority. It is possible that their increased specificity increases contention within the court. Another judge might not agree with a highly specific interpretation, explanation or ruling, as they themselves might interpret something in a different way. Since only a simple majority is needed, some diplomacy might be beneficial within the court. In diplomacy, vague language and ambiguity are key in order to reach agreements between nations. Especially when it comes to sensitive topics such as religion, human rights or peace agreements language can be the determining factor (Scott, 2001; Pehar, 2001). By utilizing vague language, a ‘swing’ vote may be convinced to vote in favour or against a judgment. In regards to SCOTUS, an interesting occurrence can be seen in Table 3: in case of sufficient risk of non-compliance, the degree of vagueness decreases as the majority decreases as well, while in cases with insufficient risk, the opposite is true. Another possibility is that the nature of constitutional law causes more contention, as the topics – such as
incarceration, gun laws, and fundamental rights – are topics that people have strong opinions on generally. Also noteworthy in the results is that compared to the ZACC and SCC, the SCOTUS is barely ever in agreement amongst its judges. While only around 21% of SCOTUS decisions are made unanimously, 37.5% are made with a 5 against 4 vote. Although the data is too limited to draw hard conclusions on the possible relationship between court contention, non-compliance risk and vagueness, these findings might be researched further.
One factor that would be worth exploring is the way in which the judges are appointed and how long their tenure may be. All three courts’ justices are nominated in different ways and appointed by the executive and legislative branches of the country. In the United States, Justices are appointed for life and have no obligated retiring age, while the ZACC allows judges to serve for a maximum of 12-15 years (depending on age and previous work in major courts) and until the age of 75. The SCC judges are also appointed for ‘life’ but mandatorily retire at 75 as well. These individual judges are not directly elected, but their appointers are in elected offices. This makes the appointment procedure a political one. Decisions made by these courts can influence politics and the elected officials. Judges may or may not feel obligated to rule a certain way because of expectation by their affiliated parties or elected officials, even if the courts are meant to be ‘independent’ in theory. Additionally, the long tenure means that these judges have to work together for years on end (decades in the case of some current SCOTUS judges) which arguably asks for some degree of diplomacy in order to work together.
Statistical Significance of the Results
The total sample size of the data was 153 court cases, which were all of the cases in the past 10 years in which these three courts have deemed legislation to be invalid for constitutional reasons. The data set could not have been bigger unless the scope of the research would have been enlarged to encompass a longer period of time. However, this would have been unadvisable since – as stated before – public support for institutions does shift throughout the years. This would have had an impact on the results and would have called for a longitudinal as well as comparative study. The likelihood that the connection between public support, vague language and non-compliance risk is found because of chance as opposed to an existing relationship is small. In Table 4, the standard deviations, standard margin of errors as well as the variant coefficient can be found to support this. The standard deviation (SD) indicates to what degree the results are spread out. A larger number means a wider spread of results in the data set. This number is small, but that does not necessarily mean anything. The standard error (SE) is more indicative of the representativeness of the results. There is a 95% chance that the
population mean is within +/- 1.40 (=2*0.698) of the average amount of vagueness for the SCOTUS (23.18%).
The statistical significance is determined by the likelihood that a null hypothesis of this thesis is true: that there is no relationship between public support, non-compliance risk and the use of vague language. This is calculated by determining the amount of results that fall within an acceptable range within the mean. Usually, that range is set at 0.05 or 0.10. For this thesis, I have calculated the Z-score and P-value for the results of the cases that had a sufficient risk of non-compliance against the results for the cases that had an insufficient risk of non-compliance. As seen in Table 4, the results in total, for SCOTUS and for SCC are statistically significant at p < .05. The results for ZACC are not statistically significant enough to draw conclusions. Therefore, the first hypothesis is confirmed, and the second is not, as there sufficient probability that these findings are a result of chance instead of the relationship between the variables.
Table 4: SD, SE, Z-score and P-value of the Results.
Mean % of vague sentences - insufficient non-compliance risk Mean % of vague sentences - sufficient non-compliance risk standard deviation standard error of mean Z score p-value total 30.17953 32.75524 12.44958 1.006489 2.256913 .012007 SCOTUS 24.14189 22.68412 5.227159 0.698508 -1.69638 .044943 SCC 20.75823 19.20643 4.38004 0.842939 -1.69912 .04466 ZACC 44.03411 44.35276 5.833334 0.697217 0.419586 .337425 Policy Implications
The results of this empirical research confirm the hypotheses as posed in the theoretical framework. As expected and hypothesised, the judges of these courts craft vaguer or more specific rulings based on the risk of non-compliance for a given constitutional law case. Judges are active policy-seekers. Although it is not clear from the data whether they utilize vague
language as a deliberate or inherent tool in writing judgments, the data does confirm that is indeed a tool. In regards to the relationship between the judiciary and the public, they may actively use vagueness to their advantage. Individual judges, courts and broader judicial systems may benefit if they (start to) use vague or specific language deliberately as a tool to strengthen their position in the government and influence the direction that legislation and policy takes. This way, the judiciary may achieve ground-breaking results that would not be possible or accepted without vaguer language. These other policymakers from the legislative and executive branches have to be aware of the ever increasingly politicized nature of the judiciary and the implications that non-compliance may have on their public support. The usefulness of vague language as a tool is seemingly most useful within the judiciary. Through diplomacy, argument, and ambiguity, agreements may be reached and judgment may be passed as the majority. Especially in contemporary concerns and cases regarding for example racism and homophobia, vague language can mean that difference between a majority vote for or against. A suggestion would be for the judiciary to be aware of and utilize the way in which language can work for them, and for the government as a whole – including the two other branches – to be aware of the way in which language can be used for and against them.
Conclusion
Vagueness is and forever will be a vague term without clear boundaries. It is also irrefutably intertwined with the law. Constitutional law judgments are largely comprised of interpretation, explanation and argumentation of laws and statutes, and even specific words. The presence of vague wording can have large implications for a government, society, and individuals, especially in constitutional review cases. These cases assess whether legislation can stand against the primary laws of a state, which often concern fundamental and human rights. In contested cases, vagueness may be used to postpone or refuse compliance by the government. In all constitutional court cases, there is always some risk that the governmental body responsible for invalidated legislature will not comply. The courts, naturally, would prefer them not to. Although vagueness can lead to an increased risk of non-compliance, courts do still make use of vague language. This thesis set out to test and confirm the theorised model by Staton and Vanberg (2008) and Sternberg’s (2018) study of said model’s empirical implications. In this model and study, it is argued that courts with high public support will write increasingly specific rulings when faced with risk of non-compliance, while courts with low public support will write increasingly vague rulings.
