Speaking Rank: Obligatory Honorifics and House Standing in Landsraad Galach, from Council Transcripts and Correspondence of 10100–10191 AG

1. Introduction

The Imperium orders its people through the faufreluches, the rigid class system that assigns every person and House a fixed station. In the Landsraad, the assembly of Great and Minor Houses, that order is performed as well as recorded: seating, the sequence of speakers and the forms of address all declare where each House stands. Landsraad Galach, as we use the term, is the formal council register of the Imperial lingua franca, in which sessions are conducted and minuted and Houses write to one another on matters of state.

One feature makes this register unusually informative about politics. Every second-person address carries an obligatory honorific affix, so a speaker who addresses another House must claim a position for it. The first author earlier described these forms from elicited judgments (Aldevash, 10236 AG), and an older survey treated court deference marking across the Imperium without quantitative data (Tessarin, 10172 AG). This paper offers, to our knowledge, the first corpus-based quantitative description of the system.

The nearest comparison is Chakobsa, whose archaic formal register, in Bantham and Threll's reconstruction, obliges speakers to mark how they know what they assert. Landsraad Galach imposes the same kind of obligation on social position: rank cannot be left unstated.

We ask whether honorific choices track the standing of Houses as the Landsraad recognised it, and whether a collective shift in the forms used toward a House appears before its standing is formally changed. We write in about 10240 AG. The corpus closes in 10191 AG, before the upheavals that reordered the Landsraad. Because the first author's home institution is on Ecaz, the subject of our case study, coding was kept blind to addressee identity.

2. Methods

The council material comprises all 189 formal sessions of 10100–10191 AG whose verbatim transcripts survive in the Landsraad Public Archives (Landsraad Council, 10100–10191 AG). The session calendar lists roughly 1,100 formal sittings in those years, so the transcripts are a surviving subset that favours sessions of recognised importance. We added 58 letter exchanges: 41 between Houses (22 from the Atreides Household Archive, 19 from the Harkonnen Administrative Records) and 17 petitions to the throne with replies, held in the Imperial Archives at Kaitain. The 247 texts total about 320,000 words, with spelling normalised to the Commission of Ecumenical Translators' conventions for Galach of record (Commission of Ecumenical Translators, 10088 AG).

Our unit of analysis is the address token, any second-person expression or pronominal affix directed at an identifiable addressee; we coded 18,640. Addressees fall into four categories: the Padishah Emperor in person, Imperial House officials speaking for the throne, the twelve Great Houses in the corpus, and Minor Houses taken together. Atreides, Harkonnen and Ecaz are analysed individually; the rank of Ecaz as a Great House is our own identification from the precedence rolls we reconstruct. The other nine, none represented heavily enough for separate estimates, are pooled.

Two trained Galach coders worked independently. Honorific affixes are morphological and identifiable in isolation, so addressee identity was stripped from the coding sheets and restored afterwards from session metadata. Agreement was κ = 0.91 for address forms, 0.88 for verbal agreement and 0.84 for adjectival intensification, and disagreements were resolved jointly.

Five of the seven address forms express graded deference and were coded on an ordinal scale from 1 (minimal respect) to 5 (supreme honour). The subordinate form, which declares a sworn or cadet relationship, and the formal-distance form, which declares deliberate reserve, do not belong on that scale; each was analysed as a binary outcome by logistic regression. Deference was modelled by ordinal logistic regression with addressee, speaker category and period (10100–10129, 10130–10159, 10160–10191 AG) as fixed effects and a random intercept for text. Addresses to the Emperor never varied and were excluded from the model.

For drift, ordinal models with bin as a linear term were fitted on four twelve-year bins spanning 10136–10183 AG, separately for Atreides, Harkonnen and Ecaz and once for the pooled nine as a group, since single pooled Houses are too sparse; the four trend tests were Holm-corrected. Lead time was the interval between the earliest bin whose deference was significantly below that of the initial bin (10136–10147 AG) and the Landsraad's formal change to that House's precedence. A House pair qualified for the asymmetry analysis with at least 30 tokens in each direction in some period, and was called asymmetric when the 95% confidence interval for the difference between its directional log-odds excluded zero. Conflict meant a kanly notice or an arbitration petition between the pair in the ten years after that period, up to 10191 AG; asymmetric and symmetric pairs were compared by Fisher's exact test. For the 14 diplomatic incidents recorded in the window, deference between the parties was compared across the five years before and after each incident, with a random intercept for incident. For insult by omission, we retrieved every minimal-respect token addressed to a House whose median received deference in that period was 3 or higher.

Nine consultants trained in Landsraad procedure judged 24 test sentences with the address affix removed, and 24 matched controls, as acceptable, marginal or ungrammatical.

3. Results

Table 1 gives the paradigm as we reconstruct it, cited on the neutral second-person stem hald-. Seven address forms agree with six verbal suffixes, because minimal and standard address share the plain indicative. The glosses hold wherever the text is legible; -ka, for example, is conditional and appears only with supreme-honour address.

None of the 18,640 tokens lacked an honorific affix. Of 216 consultant judgments on affixless sentences, 209 (96.8%) were ungrammatical, 7 marginal and none acceptable; controls were judged acceptable in 212 of 216 cases. Our claim of obligatoriness covers second-person address only.

All 767 supreme-honour tokens were addressed to the Emperor in person, and 759 (99.0%) carried the conditional; the remaining eight fall in passages the clerk marked as garbled. Praise of the Emperor likewise used a deferentially marked superlative in all 211 attestations.

Addressee strongly predicted deference (Table 2). Against the nine pooled Great Houses, Imperial officials received much higher deference and Minor Houses much lower. Atreides and Harkonnen stood modestly above the pooled Houses, while Ecaz stood below them across the window. Of 1,146 subordinate tokens, 1,081 (94.3%) came from cadet-branch members or sworn vassals of the addressee's House, and the formal-distance form was far more likely in arbitration sessions than elsewhere (OR 3.9, 95% CI 3.4–4.5).

Two of the four series showed drift that survived correction. Ecaz declined, and Atreides rose (β = 0.19 per bin, 95% CI 0.07–0.31), in line with the growing Landsraad standing of that House over these decades. Neither Harkonnen (β = 0.04, 95% CI −0.08 to 0.16) nor the pooled nine (β = −0.06, 95% CI −0.13 to 0.01) showed a significant trend.

In the Ecaz series, the share of tokens at elevated or higher fell from 129 of 212 (61%) in 10136–10147 AG to 131 of 238 (55%) in 10148–10159, 110 of 251 (44%) in 10160–10171 and 71 of 197 (36%) in 10172–10183. The negative trend coefficient (β = −0.34 per bin, 95% CI −0.46 to −0.22; OR 0.71) means lower deference in each successive bin. The second bin did not differ significantly from the first (OR 0.80, 0.56–1.14); the third did (OR 0.50, 0.35–0.71). By our reconstruction from the Landsraad precedence rolls, which draws on no later chronicle tradition, Ecaz was moved down the order of seating in 10184 AG. On our definition the decline was detectable 13 to 24 years earlier.

Of the 66 possible pairings among the twelve Great Houses, 23 met the data threshold and 8 were asymmetric. Conflict followed in 6 of the 8 asymmetric pairs and in 7 of the 15 symmetric pairs (Fisher's exact p = .38). Among the asymmetric pairs, Harkonnen speakers addressed the Atreides through 10190 AG with more deference than they received (OR 1.6, 95% CI 1.2–2.1). Given the long enmity between the Houses, we read this as the elaborate courtesy with which formal hostility was conducted under the conventions of kanly, and not as a concession of rank. Across the 14 recorded incidents, deference between the parties was lower afterwards than before (pooled OR 0.71, 95% CI 0.55–0.92, p = .009).

We found 31 minimal-respect tokens addressed to highly ranked Houses, and 23 of them (74%) drew a recorded objection or demand for retraction within the same exchange. Of 200 randomly sampled standard-respect tokens to the same addressees, 3 drew objection. The objected cases include Harkonnen minimal address to the Atreides in the 10191 AG sessions on the transfer of the Arrakis fief, a sharp break with the over-honouring of earlier decades.

4. Discussion

Because every address must carry a rank, Landsraad Galach turns each act of address into a declaration about the faufreluches. Table 2 shows that these declarations track recognised standing, separating the throne from the Great Houses and the Great Houses from the Minor, as Reyes-Okafor and Aldevash's account of Great House governance would predict. The deference paid to Imperial officials fits the Houses' stake in the throne's legitimacy through their ratifying role in succession, as Kesteven and Reyes-Okafor reconstruct it.

Honorific drift is the more consequential finding. In the Ecaz series, collective downgrading became measurable more than a decade before the Landsraad acknowledged a loss of standing. If the pattern generalises, council transcripts read for their honorifics could serve as political intelligence, an application Mentat analysts have proposed for other features of the record (Ossandre, 10233 AG). One case supports a hypothesis and does not establish a law. With only four series tested, we also cannot say how often drift runs its course without formal consequence.

Houses may also read the forms others use and adjust their own, so that downgrading reinforces itself; our data cannot separate such feedback from tracking of a decline with other causes.

Insult by omission is where obligatory grammar bites hardest. No address exists without an honorific, so a speaker cannot plead that one was forgotten, and minimal address to a highly ranked House was read as deliberate: three quarters of such uses were formally contested. The objection records bear out the procedural weight legal commentary assigns to formal address (Hollin-Varr, 10229 AG). Asymmetry between pairs, by contrast, did not distinguish those that went on to conflict; with 23 qualifying pairs the result is exploratory and cannot show that asymmetry is harmless.

5. Limitations

The corpus is confined to the formal register and to surviving transcripts, which favour important and possibly contested sessions. The correspondence comes from three archives, so the Atreides and Harkonnen chanceries are better represented than other Houses.

We did not model individual speakers, and a few prolific envoys could shape a House's received profile. The Ecaz trend rests on 898 tokens, and its dating depends on our reading of the precedence rolls. Although coding was blind to addressee, the choice of Ecaz as the case study was not independent of the first author's institutional position.

Causal direction is undetermined. Nine procedurally trained consultants form a small panel, and their training may have sharpened their judgments. The commercial Galach of CHOAM, where shareholding may cut across feudal rank, lies outside this corpus, and whether it shares the paradigm remains unknown.