This note reads aggregated counts of deferral phrasing — "kentou shimasu" (we will consider), "kentou shite mairimasu" (we will keep this under consideration), and related variants — across Japan's local assemblies, using the machikarte corpus (~126 million assembly speech records). Naming individual assemblies or councillors in evaluative ranking terms is outside the article's purpose. The scope of observation is held at municipality-aggregate and prefecture-aggregate granularity; no top-or-bottom ranking by municipality name is published here.
What we are observing
The aggregation covers 870 municipalities whose assemblies have at least 500 recorded speeches, for a combined approximately 18.97 million responses. Speeches that include deferral expressions — "kentou shimasu" (will consider), "kentou shite mairimasu" (will keep under consideration), "kentou suru" (consider), "kentou chuu" (under consideration), and similar — are mechanically extracted, and the count is divided by the total number of speeches in that municipality to derive the "deferral phrasing ratio".
The weighted national mean is 3.58%; the simple mean is 3.72%; the median is 3.01%. The 25th percentile is 1.66% and the 75th percentile is 4.87%; the inter-municipality distribution spans from a minimum of 0% to a maximum of 21.07%.
The deferral phrasing ratio is not, on its own, an indicator of assembly quality. The same word "kentou" can occur in contexts that immediately commit to a policy direction, in routine usage that simply marks a stage in policy formation, and as conventional response phrasing. What this article tries to read is the regional and structural unevenness of that language use.
Background and context
What "kentou" does in assembly responses
"Kentou shimasu" in administrative responses serves several functions in the policy process. One is a procedural reply that suspends an immediate yes/no to a councillor's proposal while securing internal review time. Another, in some cases, is a softened expression of de facto non-adoption. Comparative analysis across municipalities of how reservation phrasing affects transparency in policy formation has remained scarce.
In the study of national-Diet responses, gradient adverbs such as "kenni kentou" (vigorous consideration) and "maemuki ni kentou" (forward-looking consideration) have been noted to carry implicit gradations that, in practice, correlate with downstream budgeting and institutionalisation. Systematic equivalents for local assemblies have been thin, partly due to the absence of a cross-cutting search base. Aggregating across the country on the machikarte base offers a starting point.
Prefectural distribution
At the prefectural level (limited to prefectures with at least six municipalities in the sample), the deferral phrasing ratio also varies clearly. The highest prefecture is 10.46%; the lowest is 0.96% — roughly an eleven-fold range.
Variability tends to be larger in prefectures with few sampled municipalities, and the endpoints in those prefectures may be statistically unstable. Even when restricted to prefectures with 20 or more sampled municipalities, prefecture-level medians still spread across a range from under 2% to above 5%. The article refrains from naming the highest and lowest prefectures individually and reads the spread as a structural observation.
Reading the structure
The width of the distribution is itself the structural observation
The inter-municipality distribution spans nearly 21 percentage points, and the gap between the weighted mean (3.58%) and the median (3.01%) indicates a right-skewed distribution with a long upper tail. Most municipalities cluster near the median, with some sitting at noticeably higher ratios.
What that spread suggests is that the choice of response language is not uniformly distributed across the country, and may carry systematic regional and procedural differences. The fact that, within the same national legal framework, the linguistic features of assembly responses are not uniform is itself worth treating as a structural object of study.
Multiple interpretive routes run in parallel
A high or low deferral phrasing ratio does not, on its own, map directly to assembly quality. At least the following interpretive routes run in parallel:
- Stage differences in policy formation: Assemblies handling more topics still at the consideration stage will tend to show higher shares of reservation phrasing
- Differences in assembly procedural culture: Assemblies where avoiding immediate commitment has become an established response style versus those that prefer explicit answers
- Differences in agenda portfolio: Different ratios between topics amenable to immediate decision and topics that inherently require longer review
- Differences in coverage: Differences in the time range and committee scope of recorded minutes across municipalities
Separating which of these is operating, and in what proportion, for which municipalities, requires layering topic classification, temporal normalisation, and text-classification models for contextual analysis. This article stops at presenting the distribution from surface aggregation as a starting point.
Room for cross-lab citation
Structural skew in reservation phrasing belongs to the larger theme of "making the debate process visible", alongside the hollowing of debate and the structuring of silence. Forthcoming articles in this lab — contextual classification of "kentou" expressions by topic, relationships with policy propagation patterns, and regional differences in how citizens appear in the discourse — will cross-reference the prefectural distribution presented here.
Caveats — what is not yet covered
- No semantic classification: Contextual classification of "kentou" expressions (forward-looking / hesitant / procedural / euphemistic) is not applied in this version. The aggregation stays at surface keyword level
- No agenda-portfolio normalisation: The aggregation is not normalised for agenda composition, so differences in agenda mix may show up in the ratio
- Coverage differences: The time range and committee scope of recorded minutes vary by municipality. Long-term and short-term recorded municipalities are mixed in the sample; temporal normalisation is on the roadmap
- No individual ranking: The article stays at the structural level and does not publish municipality-level or councillor-level top-or-bottom rankings. Drilling down to a specific municipality is provided through the municipality karte pages on machikarte itself, where readers can navigate the data themselves
These limits will be lifted incrementally. The next version (v2) is planned to add contextual classification of "kentou" expressions (forward-looking, hesitant, procedural, euphemistic) and agenda normalisation.
Verifiability
The query specification (spec_version v1-tier1-500threshold), prefectural aggregates, and municipality-level raw data are scheduled for publication on machikarte's methodology pages. The BigQuery aggregation queries will also be placed in the machikarte GitHub repository so that third parties can reproduce the results independently.
A corrections contact is provided separately to receive notices about aggregation errors and notation drift. The lab's editorial procedure — including the three-tier rule on councillor-level data publication — is documented in the lab overview note (hypothesis-overview).
References
machikarte — Nationwide Local Assembly Speech Search Platform (Beta) — Institute for Social Vision Design (ISVD). ISVD
machikarte (GitHub) — schema, aggregation queries, licenses (MIT + CC BY 4.0) — Institute for Social Vision Design (ISVD). GitHub
Verification of Role Classification Methods for Diet Records Using BERT-Based Classifiers (in Japanese) — Miyaki, Y. and Uchida, Y.. Journal of Japan Society for Fuzzy Theory and Intelligent Informatics, Vol. 37, No. 1, pp. 530-534
