ICJ Additional Opinions Corpus by Gender (1985–2024)

The ICJ.M&W.24 corpus was created in the fall of 2025 for the purpose of linguistic analysis of male and female language in additional opinions of judges of the International Court of Justice. 

Judges of the International Court of Justice may attach individual or joint opinions to the Court's judgment. Unlike the text of the judgment, each opinion can be attributed to a specific judge or group of judges. 

In compiling the corpus, I sought to create a corpus that 1. would enable a comparison between male and female language, 2. would enable a comparison between the language of majority decisions and the language of individual opinions attached by judges, 3. would take into account the existence of joint opinions, which may be authored by both men and women, 4. includes all court decisions issued by the end of 2024 to which at least one female judge and at least one male judge has attached an individual opinion.

The texts forming the corpus were downloaded individually by hand from the website of the International Court of Justice in .pdf format. Information about the type of text was included in their names (D for Court decisions, M for male opinions, W for female opinions, X for opinions prepared jointly by men and women). Furthermore, information about the type of attached opinion (declaration, separate opinion, dissent) and the surname of the author or co-authors of the opinion was added to their names. 

The ICJ.M&W.24 corpus consists of 665 texts divided into 4 subcorpora. The division of texts into subcorpora was based on information about the gender of the author of the text. 

The Court's Voice subcorpus (Decisions_zip) includes ICJ judgments, ICJ orders, and ICJ advisory opinions. These are decisions adopted on behalf of the Court. It contains only those decisions to which at least two individual opinions have been attached, at least one of which was written by a woman and at least one by a man. The subcorpus does not include decisions to which no opinion was attached or to which only opinions written by male judges were attached. The subcorpus consists of 80 texts and 1,513,242 tokens. 

The Female Voice subcorpus (Women_zip) contains individual opinions of female MSD judges on decisions included in the Court's Voice subcorpus, including ad hoc opinions of female judges. It is a complete set of opinions written by female judges. It does not contain any joint opinions, as in the court's practice to date, no joint opinion or joint declaration has been attached that was authored exclusively by female judges. The subcorpus consists of 115 texts and 511,837 tokens.

The Male Voice subcorpus (Men_zip) consists of individual and joint opinions of MSD judges, including ad hoc judges, on decisions included in the Court's Voice subcorpus. Joint opinions are included in this corpus if all their authors are men. The subcorpus consists of 449 texts and 2,784,097 tokens.

The Mixed Voice subcorpus includes joint opinions on decisions included in the Court's Voice subcorpus, provided that their authors are both men and women. It is the smallest subcorpus, with 21 texts and 92,807 tokens.

Regarding the structure of the corpus, it can be further noted that 427 texts (Court decisions and appended opinions) relate to Court judgments, 187 to orders and 51 to advisory opinions. The proportions of attached opinions are as follows: 206 declarations, 245 separate opinions, 135 dissenting opinions (including 6 opinions that the judges described as partially dissenting). 27 opinions and declarations were designated as joint. 

The corpus covers the period from 1985, when a woman first participated in the decision-making of the MSD – ad hoc judge Suzanne Bastid – to the end of 2024.