---
title: RFC: Down with Specification
author: uakci
---Neither the 2011 nor the 2020 edition of Ithkuil is lexically consistent.
For a clear statement of the intent of this proposal, please read the opening paragraph of the Proposal section.
[2038-01-19 Tue 03:14]
WIP Draft.
- @uakci
Specification→Format concept originally by @melopee; it is not implied that he endorses the RFC in whole, or even the author’s execution of his idea.
(In the following paragraphs, uppercased Stem refers to the tri- or tetravalent Category, and lowercase stem—to the linguistic concept.
In 2011 Ithkuil, the latest stable version of Ithkuil, all roots are expected to conjugate for three lexical Categories—Designation, Pattern, and Stem—yielding 18 possible stems. Designation is a well-defined bivalent Category which determines the presence or absence of permanence, institutionalization, or artificiality in a root concept (e.g., create vs. construct, wander vs. travel). The reference grammar explains Pattern and Stem with great vagueness; one may gather that these trivalent Categories’ main purpose is to subdivide the general, overarching semantics of the root in order to yield finer semantic distinctions—or to group related concepts, or their disparate facets, under a single consonantal affix.
The main point that the author wishes to make is that the above-presented system suffers from non-expansibility. Specifically, with each new root that the language creator (John Quijada or otherwise) wishes to add to the language, they must consider a dichotomy and two trichotomies to go with it. In order from least to most irrational:
-
While Designation is mostly predictable in its praxes, there exist examples which seem forced and/or humorous, even, in their absurdity: -DK- buttock → haunch, -Ph- wood → wood product, sound made by non-oral, non-vocal part of the body → phoneme (viz. -LŻ-), -Qh- spirit or soul → religious belief.
-
In most cases, the Pattern 1 meaning (the ‘Holistic Stem’) is rendered as as a sum of two parts, which are then used to populate Pattern 2 and 3—e.g., -L- vocal utterance → physical production of vocal utterance + content conveyed through vocal utterance. Another lexical arrangement common to the 2011 lexicon is to introduce a presence–absence distinction of a certain property within Patterns 2 and 3 (the so-called ‘Complementary Stems’)—e.g., -RR- cat → wild cat, domesticated cat. However, no Pattern pattern prevails throughout the lexicon, which engenders gratuitous vocabulary bloat.
-
Stem is the most irregular Category of all three. Sometimes, it is used in the same way as Pattern is (e.g., -SK- vs. -LTh-, -L- vs. -Th-). Quijada does not give any remote explanation of Stem apart from the statement that ‘[i]t is at the level of stem that Ithkuil roots become actual words with instantiated meaning’ (source).
-
Wherever the eighteen combinations being the Cartesian product of Designation, Pattern, and Stem are not enough room for housing a host of interrelated concepts, 2011 Ithkuil uses the SSD suffix. This suffix may be used on some—but not all, or even the majority of—roots to prescribe Stem-Specific Derivations (for example, the Stem-Specific Derivations for -TT- ‘salutation/greeting’ include shake hands upon leavetaking and verbal salutation only upon arrival, and the 31 (!) SSDs for -ŢW- ‘weather condition’ include such lexemes as smog and gentle breeze).
The above examples aim to show that while at first blush, the grid-like approach to vocabulary building and encoding of nuance may seem clever, Quijada treats the root as a ‘bag of words’ with little care for consistency, arriving at a Byzantine concept-organization which does not lend itself well to formalizing or reasoning about. This sorry state of affairs is only exacerbated by the SSD suffix, which breaks the neat structurization which follows from the use of three orthogonal Categories. While bags-of-words are not harmful in and of themselves, their deceitful appearance within 2011’s seemingly orthogonally-designed lexical schemata betrays the vapidity and inconsequentiality—ultimately, the redundancy—of those schemata.
There is also a trend highlighted in these examples: where Quijada’s wording is inexact, inspecific, or lacking in detail, the concept there described tends to suffer from inconsistency or arbitrariness. This will come into play in the following section:
2020 Ithkuil replaces 2011 Ithkuil’s 18-cell grid with a 24-cell grid. In particular, Pattern is replaced with Specification—a refreshingly regular 4-way distinction (or so it would seem). The possible values for Specification are:
- Basic (BSC) :: (core meaning of a Stem×Designation combination)
- Contential (CTE) :: ideal essence or function
- Constitutive (CSV) :: actual shape or realization
- Objective (OBJ) :: ‘[t]arget or "object" or product associated with the act/state/event’ (p. 13 of Morphology 0.12.3)
As before, Designation×Stem are oftentimes treated like bags-of-words: Informal Stems may not always correspond to Formal Stems. While most words do not have Formal Stems, all have all three Informal Stems, which leads to ridiculous forced-trilemmas in 2011 Ithkuil’s spirit, such as finger – thumb – toe or arabica coffee – robusta coffee – other coffee (roots -NK- and -ÇX-). In 2020 Ithkuil, it is Specification which causes the most problems.
Specification is intimately related with the syntactic function of the formative that the root is embedded in—the verbal sense is linked to the nominal. In a majority of instances, Specification follows the scheme (example root is Stem 1 of -JḐ- ‘predicament’):
| Nominal | Verbal | |
|---|---|---|
| Basic | state of facing a predicament | to face a predicament |
| Contential | state of not knowing how to escape a predicament | to be a state of not knowing how to escape a predicament |
| Constitutive | state of causing a predicament | to cause a predicament |
| Objective | the predicament-circumstance itself | to be the predicament-circumstance |
We can extract several law-like observations from this table:
-
There are two feasible possibilities for a Specification’s nominal and verbal readings:
- ‘state of X-ing (nominal); to X (verbal)’
- ‘X (nominal); to be X (verbal)’
-
Furthermore, the latter option is characteristic of Contential and Objective—Basic and Constitutive are action-like.
Nevertheless, as one analyses the lexicon, one will find that these observations do not hold. Further remarks:
-
The Objective Specification—the most loosely defined out of the four—can assume any of the following roles: subject (-KŠ-), object (-MŠ-), goal (-SŘ-), tool (-ẒḐ-), particular instance (-LZ-), cause (-MR-), essence or manner (-ḐḐ-), related action (-MŘ-, -N-), the one impacted by the presence of a being or an action (-ŠW-), and other single-issue examples, like -G- or -ŠL-.
-
Sometimes, the Contential sense of a root is the same as or similar to the Objective; sometimes, it is the Basic sense which is paired with the Objective. In the former case, the Basic Specification may embrace an awkward general-like meaning, as it is in the case of -LKS- a state of there being food available for consumption; in some of the latter cases, it is the Contential the takes on such an empty meaning, as in -LLY- the state of there being singing to be heard. In other words, there are numerous samples in the lexicon where a root would benefit from one of its Specifications not being specified.
-
Among misfitting lexical items, one may give -CJ- (Objective is missing for no reason), Stem 3 of -JJ- (Objective is ‘same as CTE’)…
-
The four Specifications closely resemble Aristotle’s Four Causes; despite this, many Contential and Constitutive senses miss the Aristotelian ‘letter vs. intent’ distinction, and, as demonstrated above in (1.), the Objective Specification, supposedly equivalent to the final cause, is in fact treated as a catch-all utility conjugation whose meaning is fully dependent on that of the particular root.
In conclusion, 2020 Ithkuil’s Specification does not introduce any systematicity to the system. While it is the case that most phenomena can be portrayed in terms of the BSC–CTE–CSV–OBJ framework, a large, unignorable part of the lexicon fits in it poorly, which breeds exceptions and irregularities reminiscent of 2011 Ithkuil’s.
(Note: affixes are called suffixes in 2011 Ithkuil. There is no difference between the two terms, up to the version of the language involved.)
Affixes and affix degrees have little or nothing to do with the corresponding roots. 2020 Ithkuil’s Affixes document defines six affix gradient patterns (see p. 3 therein); however, they only conceal the fact that there is no feasible way to derive the degrees from the consonantal affix values. The degree assignments are predominantly pragmatics-driven; some of them, like -ct Sequential Participation, the grammatically salient affixes (like -ň COO, -n TPF, -t DCD), having no semantic impact, do not correspond to any root in any way, nor under any mapping (from degrees to their meanings). This is unacceptable: an ideal Ithkuilic language would ideally restrict its irregular syntactic affixes to a minimum, and leave the rest of the heavy-lifting semantic work to the underlying roots. It is not a light job to fill all nine degrees with sensible values; thus, some of the affixes present in the lexicon contain awkward ninth values, while others resemble bags-of-words (whose implicitness is what the author means to avoid at all costs).
In addition, it is a great drawback that only those roots which John Quijada has considered to render as affixes can be used as affixes, with all other roots only appliable via incorporation or periphrasis.
In the previous section, we saw that any grid-like approach to semantic refinement eventually leads to over-analysis and false dichotomies. This Request For Comments, with 2020 Ithkuil’s morphology, Categories, and Case values at the outset, intends to:
- replace Specification with a Case-Accessor-like mechanism, hereafter called Format (in analogy with 2020 Incorporation Format);
- introduce a series of Cases which help cover the semantic space of 2020’s Contential, Constitutive, Objective;
- revamp Degree into a specialized set of adjectival/adverbial relations (resembling 2011 Ithkuil’s Format), dictated by pragmatics.
The following is a possible implementation of the goals outlined above:
-
Remove Specification. Move Stem and Function to Slot IV (where Designation, Version, Relation reside currently). With Slot VI empty, introduce the new Category, Format, identical to Case, there. Format is functionally identical to a Case Accessor and scopes over Designation, Version, and Stem, but is scoped over by the Ca complex. (One might favour exchanging Slot VI with Slot IV in such a setting, but this issue is highly dependent on the frequency of these slots’ exhibiting of non-default values relative to one another.)
-
Extend the Vowel Sequence with a ninth column (this should be a feasible job, as there exist 81 V’V forms, and only 36 are currently taken).
-
Augment 2020 Ithkuil’s Cases from
THM INS ABS STM AFF EFF ERG DAT IND POS PRP GEN ATT PDC ITP OGN IDP PAR APL PUR TRA DFR CRS TSP CMM CMP CSD FUN TFM CLA RSL CSM CON AVR CVS SIT LOC ATD ALL ABL ORI IRL INV — NAV CNR ASS PER PRO PCV PCR ELP — PLM REF ASI ESS COR CPS COM UTL — RLT ACT DSC TRM SEL CFM DEP PRD — VOC to (changes in bold):
CTE INS ABS STM THM AFF ERG DAT IND FUN MAT MTH MTV XPL — SNT EFF MPL POS PRP GEN ATT PDC ITP OGN IDP PAR APL PUR TRA DFR CRS TSP CMM CMP CSD TFM CLA RSL CSM CON AVR CVS — SIT LOC ATD ALL ABL ORI IRL INV — NAV CNR ASS PER PRO PCV PCR ELP — PLM REF ASI ESS COR CPS COM UTL — RLT ACT DSC TRM SEL CFM DEP PRD — VOC -
Case series 2–8 are shifted by one row (with series 8 taking the new Vowel Sequence column).
-
The Transrelative Cases are rearranged slightly. Importantly, the Effectuative Case is moved to the new second row.
-
A new Case is added: the Contentive, which refers to the state or action or entity that is the semantic content of the root. As a Format, Contentive is a no-operation; as a Case, one may instantiate it to equate the predicate with the noun marked by that Case—for example,
angry-(verbal) you-AFF this-CTE, meaning: ‘this is [an instance of] you being angry’. While 2020 Ithkuil’s Thematic Case may refer either to the content of the action or to the action itself (that which one knows vs. the state of one knowing something), depending on whether the formative is verbal or nominal (and sometimes, the only choice available is the latter—compare the table in the 2020 Ithkuil Criticism section). In contrast to this, this proposal’s Thematic always refers to the theme of the state/action (that which one knows), while Contentive serves as a way to specify or restate this state/action itself (the state of one knowing something). -
Clause-level Cases refer to the case roles of the concept extracted by Format. If Format is the default Contentive, this is equivalent to the case roles of the bare root.
-
The Functive Case is moved to the newly-created second row, and the remaining cases shifted, creating a gap to imitate the following rows. The remaining seven new cases are as follows:
- Material: the medium through, surface on, or system via which the action/state is performed/realized/implemented (e.g., air, canvas, telephony)
- Methodic: the physical movement or change-in-state through which the state/action is expressed (e.g., she greeted me with a handshake)
- Motivational: the motive behind the actors’ actions which has led to the commencement of the state/action (e.g., I hit him out of spite).
- Explicative: reason (rational explanation, rationale/-ization, the (not necessarily directly causal) event which the speaker ascribes the action to)
- (empty slot for future expansion)
- Essential: the ideal/intended cause-and-purpose or profound meaning of a state/action (e.g., He gulps the soylent for nutrition)
- Implicative: a/the minimal set of circumstances which predicates the inevitable/invariable occurrence of a state/action
Methodic and Essential are opposites of one another (just like 2020 Contentive and Constitutional). (TODO: how about implementing inverse Case accessors, too? This would help shave off a significant amount of coantonymic Cases. Observe that V’V forms are not restricted to monophthongal Vs, and instances of )
-
-
Affix degrees are replaced with Case accessors (either the second row of Cases or all Cases); save for a handful of grammatically prevalent operator-like affixes (such as COO and TPF, which it should be noted may at any time be recast into Categories of their own), all consonantal affix values name the same semantic content as that of the corresponding root values.
-
Designation is removed. Stem is now an explicitly arbitrary lexical Category which culls a single, specific meaning out of a pool of max. 8. (Or, to match Ithkuilic tradition, 6, like 2020 Ithkuil’s Stem × Designation). The general, overarching meaning of the root is expressed by Zero Stem, which shall be the default (implied, omissible) value for this Category (like 2020’s Stem 0). In other words, related meanings should only be grouped by necessity (that is, when they are truly closely related), and the general (Zero Stem) sense shall be the implicit, omissible default.
| Vote | Name |
|---|---|
| +1 | @uakci |
| +1 | @porpoiseless |
| +1 | @toimine |
Draft.
@porpoiseless, @melopee.