DRAFT, not yet for distribution. For inquiries, write to ron.levi@utoronto.ca.
the coding framework
Finding every statement that uses the term is mechanical. Deciding what each speaker is doing with it is not. This page sets out the twelve tags applied to all 2,858 statements, what each one means, and a real example of each drawn from the corpus.
A word count treats every mention alike. But a member condemning antisemitism, a member accusing an opponent of it, a member arguing that something has been wrongly called antisemitic, and a member relaying someone else's words are doing different things. Counting them together produces a number that measures attention without telling you anything about its content.
Coding separates them. It also makes it possible to ask the questions that motivate the project: on what grounds the claim is made, what is proposed in response, who is treated as competent to act, and whether any of that differs by party, by chamber, or across a century.
unsettled and awaits human review. Beyond that flagged subset, the full set
of codes is being checked by hand, statement by statement, before this site is made public. Three
faults have been found and corrected so far, including one that mis-attached three quarters of
the codes. Section 8 reports all of them.The primary tag, and the one everything else hangs on. Exactly one value per statement.
| Value | Meaning | Share |
|---|---|---|
| condemns | Objects to antisemitism without charging anyone in particular | 58% |
| names | Says a named person, party, institution or state IS antisemitic, or that a specific act of theirs was | 9% |
| blames | Says someone ignored, enabled, funded, or responded too weakly to antisemitism, without calling them antisemitic | 11% |
| disputes | Argues the label is wrong, overused, or applied unfairly | 9% |
| reports | Quotes, cites or reads it into the record, taking no position | 13% |
The tag records what a speaker does with the term. It is not a judgment about the speaker, and nothing on this site codes a member as antisemitic.
The line between condemns and the two charge tags, names and blames, carries the most weight. A general lament is a different political act from a specific charge, and the corpus holds all three tags in quantity.
These value names were themselves audited for loading, more than once. Accuses became attributes, since an accusation carries a faint suggestion of being unfounded that the tag has no business implying. Attributes itself later split into names and blames: it had been merging two different charges, that someone IS antisemitic and that someone failed to act against it, and collapsing them hid which one a speaker was actually making. Reputation replaced fame as a better rendering of Boltanski and Thévenot's renom, and the tag now called jurisdiction was once called expertise, which granted the very competence it was supposed to be recording a claim about.
Madam Speaker, with respect to racism, misogyny, anti-Semitism, Islamophobia, homophobia and transphobia, we are seeing a disturbing rise in all these toxic forms of hate in Canada as far-right extremists mobilize in our country. We see this manifested in despicable acts of hate and violence across the country. Canadians need to stand up and speak out against far-right extremism …
Madam Speaker, the member, while he was speaking, spoke about racism. Will he denounce the tweet from the member for Timmins—James Bay that was clearly anti-Semitic?
Mr. Speaker, did the Department of Canadian Heritage ask well-known, well-documented, self-declared anti-Semite Laith Marouf to apply for funding, yes or no?
Madam Speaker, it is the second time I have heard in the House of Commons that there is something anti-Semitic being referred to on this side of the House. Nothing could be farther from the truth, and I ask the member to withdraw the comment.
The definition for anti-Semitism that is part of Canada's anti-racism strategy is the one that has been given by the International Holocaust Remembrance Alliance.
When someone argues that a thing is wrong, they appeal to some standard of value. Luc Boltanski and Laurent Thévenot, in On Justification: Economies of Worth (1991), observed that public disputes in modern societies draw on a small number of such standards, which they call orders of worth. Each has its own idea of what makes a person or a thing worthy, and its own characteristic vocabulary and proofs.
Their point is not that people are insincere. It is that a claim has to be made in some register to be arguable at all, that the registers are limited in number, and that disputes often turn on which register should govern. When a member says antisemitism violates the equality of citizens, and another says it damages Canada's standing in the world, they are not disagreeing about the facts; they are appealing to different orders.
| Order | What makes something worthy | How it sounds here | Share |
|---|---|---|---|
| civic | The general will, equality, collective interest | Rights, the Charter, all Canadians, citizenship | 72% |
| domestic | Tradition, personal bonds, standing in a lineage | Community, neighbours, generations, family memory | 15% |
| industrial | Efficiency, competence, measurable performance | Statistics, hate-crime data, expert reports, whether a programme works | 29% |
| inspired | Grace, conscience, singular insight | Moral witness, the singular horror, prophetic denunciation | 13% |
| reputation | Public esteem, recognition, reputation | Canada's standing, national shame, what the world will think | 7% |
| projective | Activity, connection, being engaged in projects | Coalitions, envoys, task forces, networks (and hatred spreading through them) | 12% |
| market | Price, competition, desire for scarce goods | Economic harm, boycotts and divestment, funding | 7% |
| green | The environment: sustainability, ecology, the physical world left to future generations | Essentially absent here, by design (see below) | 0% |
Six of these come from Boltanski and Thévenot. The projective order was added later by Boltanski and Ève Chiapello in The New Spirit of Capitalism (1999), where worth attaches to mobility, connection and participation in projects. The green order comes from Claudette Lafaye and Thévenot's 1993 study of environmental disputes.
Up to two orders are recorded per statement. Where two appear, the record is flagged a compromise, which Boltanski and Thévenot treat as an object of interest in its own right: the pairings that recur tell you which standards a society has learned to hold together.
Mr. Speaker, that is a subject I care very deeply about, and I would like to just make absolutely clear that acts of hatred, whether Islamophobic or anti-Semitic, have no place in this country. We must all work together in order to bring people together and ensure the safety of all Canadians.
Mr. Speaker, the Prime Minister 's actions speak otherwise. On one hand, he is giving these anti-Semitic religious extremists taxpayers' dollars to actively promote hatred using funds from the Canada summer jobs program. On the other, he has denied funding to faith groups that want to help those in need. Why is the Prime Minister allowing certain religious organizations to be funded to promote hatred toward Jews, but say …
… ommitted against a Markham mosque that could have turned deadly. Synagogues have also been a target of hateful acts, and the police report that hate crimes have increased a shocking 67%. Will the Liberals commit to protecting all Canadians by supporting the NDP's call for an emergency debate on recent Islamophobic and anti-Semitic attacks?
Honourable senators, Elexis Schloss first encountered anti-Semitism as a child growing up in Medicine Hat, Alberta. In an interview, years later, Elexis recalled how shocked she’d been when a childhood friend told her they couldn’t play together anymore because Elexis had killed Jesus. She said, “It wasn’t me, I don’t even know Jesus! It must have been my brother Lionel, he’s always i …
Naming a problem and proposing to act on it are two separate steps, and many statements take only the first. The tag records what, if anything, is offered as the response, grouped by mode. Answering antisemitism with criminal law, with education, with solidarity, or with closer ties to Israel are different political propositions.
| Mode | Values |
|---|---|
| coercive | criminal law · regulate platforms · immigration controls |
| institutional | adopt a definition or envoy · hold an inquiry · fund community groups · fund security |
| educational | education · commemoration |
| relational | solidarity and dialogue · leaders speaking out |
| international | closer ties with Israel · foreign policy |
A separate tag records the stance taken toward the response: whether the speaker proposes it, supports an existing measure, says existing action is inadequate, or opposes it. The third carries most of the partisan traffic: a member saying the government has failed to act is making a claim about remedy, not about antisemitism, and separating the two keeps that visible.
It was a great bill. Why didn't you follow up on some of the consequences that I had on hate speech, on anti-Semitism or on Holocaust denial? We have more anti-Semitism because of your Liberal government today than we did when I brought in the bill a year and a half ago.
Madam Speaker, with respect to racism, misogyny, anti-Semitism, Islamophobia, homophobia and transphobia, we are seeing a disturbing rise in all these toxic forms of hate in Canada as far-right extremists mobilize in our country. We see this manifested in despicable acts of hate and violence across the country. Canadians need to stand up and speak out against far-right extremism …
A claim about a problem often carries a claim about who has authority over it. This tag records which group, if any, a statement treats as the right one to act. It is a claim about jurisdiction, in Andrew Abbott's sense, and recording it does not concede that the group named is in fact competent. The tag was called expertise in an earlier draft, which granted exactly that: police, teachers, researchers, psychologists, courts, clergy, community organisations, technology companies, the civil service, Parliament itself, international bodies, or survivors.
Survivors are held apart from the rest on purpose. Their authority rests on having lived the thing rather than on training or office. Whether experiential authority gains or loses ground against credentialed expertise is a question worth being able to see, and folding survivors in with academics would hide it.
Honourable senators, Elexis Schloss first encountered anti-Semitism as a child growing up in Medicine Hat, Alberta. In an interview, years later, Elexis recalled how shocked she’d been when a childhood friend told her they couldn’t play together anymore because Elexis had killed Jesus. She said, “It wasn’t me, I don’t even know Jesus! It must have been my brother Lionel, he’s always i …
Mr. Speaker, today the people of Ottawa stand united against the twin evils of racism and anti-Semitism. Over the last several days, a criminal has targeted peaceful Jews, Muslims, and Christians from the black community with hateful vandalism. We hope the police catch and the courts punish the perpetrators to the fullest extent of the law …
A statement naming antisemitism is rarely only about antisemitism: it comes up inside a debate on immigration, a funding announcement, a school's curriculum, a bill in the House. This tag records the real-world domain the statement is substantively about: history, violence, funding, Israel, education, law, community, immigration, or media. A tenth value, accountability, is a bounded residual, used only when a statement is mainly an exchange between parties about responsibility or conduct and is not primarily about any of the other nine. It is coded by a language model from a short passage around the word, not by the same historical multi-pass pipeline the tags above use; see methods and the subjects analysis page for how it was checked.
…olocaust but it is so much more. It is a monument to those who fought in World War II, for the many who went over and were killed in that war, the families of the victims of that war. It honours the men and women in uniform today. I believe it honours the survivors of the Holocaust, and others who also face racism and anti-Semitism…
Mr. Speaker, expressions of intolerance have no place in Canadian society. Racist and anti-Semitic activities destroy the fabric of our society, destroy and undermine our fundamental values of respect, of equality and security. Canada and Canadians will not tolerate expressions of hate. Canada's strong anti-hate legislation, the Criminal Code and the Canadian Human Rights Act, makes it a crime to incite hate and to…
| Tag | Records |
|---|---|
| where it happened | Whether the antisemitism described is in Canada, abroad, or both |
| what prompted it | Legislation, a specific incident, commemoration, a foreign event, a report |
| named alongside | Whether another form of hatred is named in the same breath, and which |
| quoted | Whether the speaker is quoting someone else rather than speaking in their own voice |
| to whom attributed | When a charge is made, against whom |
| Israel and Zionism | Whether the statement conflates anti-Zionism with antisemitism, distinguishes them, alleges anti-Israel bias, or leaves the boundary alone |
The last is the most contested question in this material. The tag records the position a speaker takes; it does not adjudicate between positions. Note too that the corpus is deliberately not searched on Zionism: searching for anti-Zionism as though it were antisemitism would settle the question in the sampling frame, before any coding began.
Every statement was coded twice under different settings and the passes compared. Agreement is reported per tag rather than as one figure, because the tags differ a great deal in how much interpretation they ask for.
| Tag | Agreement | Read as |
|---|---|---|
| quoted | 96% | solid |
| where it happened | 93% | solid |
| what the speaker does | 92% | solid, and since re-coded (below) |
| what it's about | 92% | solid; a separate, later coding pass (section 6) |
| named alongside | 90% | solid |
| what prompted it | 88% | usable |
| stance toward response | 85% | usable |
| response · jurisdiction | 74% | interpretive |
| orders of worth | 70% | interpretive, treat as exploratory |
Records where the passes disagreed on the primary tag, or where the coder flagged low
confidence, carry an unsettled marker and can be isolated with the query
review:yes.
Reporting agreement is not the same as being right. Three faults were found after the first run, two of them by a reader looking at a single record. All three are described here because a page claiming to say how far the codes can be trusted has no business hiding them.
A related fault. Where a coding produced two tiers, a confident one and an uncertain one, both were being displayed identically. Reading the uncertain tier showed it was doing real work: much of it was criticism of Israeli policy being read as hostility to Jews, which is the false positive an adversarial check is built to catch and had only partly caught. When a model is unsure on this subject, it leans toward treating criticism of Israel as antisemitism, a bias worth reporting in its own right for anyone coding this material automatically. Where a coder's own grading distinguished confident from uncertain judgments, that distinction has to survive into the output. Collapsing a hedge into an assertion is a coding error even when every individual judgment is defensible.
What agreement between passes is worth. Two language models concurring means neither refuted the other. It is weaker evidence than it sounds, since they share training and can share a blind spot, and no human has read any of this coding. The value of a second adversarial pass is what it removes, not what it leaves standing.
unsettled queue.