A theoretical model of dynamical grammatical gender shifting based on set-valued set function

Imagine you are walking through a giant, chaotic library where books (words) are constantly being rewritten, re-categorized, and given new covers. Sometimes, a book about a "lion" gets a new cover that says "lioness." Other times, a book about "butter" (which usually has no gender) suddenly gets a masculine or feminine label depending on the language.

This paper is like a detective story trying to solve why these changes happen and how the library organizes them. The author, Mohamed El Idrissi, proposes a new "rulebook" for this library using math.

Here is the breakdown of his theory using simple analogies:

1. The Problem: The "Gender" Mystery

In many languages (like French, Spanish, or Riffian, a language spoken in Morocco), nouns have a "gender" (Masculine or Feminine).

The Puzzle: Sometimes, you can change a word's meaning slightly, and its gender flips.
- Example: In French, le gland (the acorn, masculine) comes from la glande (the gland, feminine).
- Example: In Riffian, you can take a word for "hand" (masculine) and turn it into a word for "bouquet" (feminine).
The Old Theory: Linguists used to think this happened because there was a tiny, invisible "switch" (a suffix) added to the word that forced the gender to change.
The Flaw: Sometimes, there is no switch. The word looks exactly the same, but the gender changes. The old rules couldn't explain this "magic."

2. The Solution: The "Template" System

The author suggests that words aren't just random strings of letters. Instead, every word is a mold (a template) filled with clay (the meaning).

The Clay (The Item): This is the core meaning of the word (e.g., "hand," "butter," "lion").
The Mold (The Template): This is a pre-made container that holds grammatical rules like "Singular," "Plural," "Masculine," or "Feminine."

The Analogy:
Imagine a bakery.

The Clay is the dough.
The Mold is the cookie cutter.
Usually, if you want a "Star" cookie, you use a star cutter. If you want a "Heart," you use a heart cutter.
The Author's Discovery: Sometimes, you take the same dough (the word "lion") and suddenly, instead of using the "Male" cutter, you switch to the "Female" cutter. The dough didn't change, but the mold changed.

3. The "Item-Template Pairing" (The New Machine)

The author invents a new machine called the TBMC Model (Template-Based and Modular Cognitive model). Think of this machine as a smart vending machine for words.

How it works:
1. You put a word (the item) into the machine.
2. The machine looks at the word's history and its meaning.
3. It doesn't just look for a "suffix" (like an -s or -tion). Instead, it checks a mathematical rule to see which "Mold" (Template) fits best right now.
4. If the rules say "Change the gender," the machine swaps the mold from "Masculine" to "Feminine" instantly.

4. The Math Part: The "Symmetric Difference"

This is the most complex part, but here is the simple version:
The author uses a specific type of math called Set Theory (dealing with groups of things). He treats "Masculine" and "Feminine" as two different groups.

The Magic Formula: He uses a concept called "Symmetric Difference." Imagine you have two lists of features.
- List A (Old Word): Has "Masculine."
- List B (New Word): Has "Feminine."
- The math calculates exactly what needs to be swapped to get from A to B.
- If the math says "Swap M for F," the machine does it. If the math says "Do nothing," the gender stays the same.

This explains why sometimes the gender flips (the math demands a swap) and sometimes it stays the same (the math demands no change), even if the word looks identical.

5. Why Riffian?

The author uses Riffian (a Berber language) as his main example because it is a "laboratory" for this. In Riffian, you can see these gender shifts happening very clearly, often without any extra letters added to the word. It's like watching a chameleon change colors in real-time, proving that the "mold" is what's changing, not the "dough."

The Big Takeaway

The paper argues that grammar isn't just about adding little tags to words. It's about a dynamic, shifting system where words are constantly being matched to different "molds" based on complex, non-linear rules.

Old View: Words change gender because we add a specific "feminine" sticker to them.
New View: Words change gender because the entire "container" (the template) they sit in gets swapped out by a hidden, mathematical rule system.

In a nutshell: The author built a mathematical model that acts like a chameleon translator. It predicts exactly when a word will keep its gender and when it will flip, not by looking for stickers, but by calculating the perfect "mold" for the word's current meaning. This helps us understand how human brains organize language in a way that is both chaotic and perfectly logical.

Here is a detailed technical summary of the paper "A theoretical model of dynamical grammatical gender shifting based on set-valued set function" by Mohamed El Idrissi.

1. Problem Statement

The paper addresses the complexity of grammatical gender shifting in synthetic languages (e.g., French, Riffian/Berber), where nouns change gender markers during word formation (derivation or conversion) without obvious derivational affixes.

The Gap: Existing linguistic theories (semantics-driven or faithfulness-presumption models) struggle to explain why some derived nouns shift gender (e.g., French le gland [M] $\to$ la glande [F]) while others do not, or why shifts occur without overt morphological markers.
The Challenge: Current models often treat gender as a static feature or rely on "derivational markers" that may not exist. There is a lack of a computational mechanism that dynamically maps lexical items to morphological templates, accounting for non-linear variations and intra-lexical shifts (noun-to-noun) where the radical remains unchanged.
Specific Context: The study focuses heavily on Riffian (a Berber language) due to its rich morphological variation and asymmetrical gender encoding, but aims to provide a cross-linguistic framework.

2. Methodology: The TBMC Model

The author proposes a new computational framework called the Template-Based and Modular Cognitive (TBMC) model. This model treats language generation as a mathematical process involving set-valued set functions.

Core Components

The model is defined as a six-tuple $\mathbb{M} = \langle W, K, T, f, g, h \rangle$ :

$W$ (Items): The set of lexical items (radicals/roots).
$T$ (Templates): The set of grammatical templates (sets of features like Gender, Number, Countability).
$K$ (Retrospective Determinants): The set of factors determining template shifts (e.g., the type of word formation process).
$f$ (Backward Recursive Function): Maps an item $w \in W$ to a subset of $K$ (identifying the formation process and base properties).
$g$ (Gradient Function): Maps elements of $K$ to the output template in $T$ .
$h$ (Morphosyntactic Transfer Function): The composition $h = g \circ f$ . It is a set-valued set function ( $h: \mathcal{P}(M) \to \mathcal{P}(M)$ ) that predicts the nonlinear dynamic mapping of an item to its final morphological template.

Mathematical Formalism

Set-Valued Set Functions: Unlike traditional functions that map a single input to a single output, this model uses set theory to handle the complexity of linguistic features.
Symmetric Difference ( $\Delta$ ): The core operation for determining shifts. The relationship between a base template ( $t_i$ ) and a resultant template ( $t_j$ ) is governed by a "gradient condition" involving the symmetric difference:
$t_i \Delta t_j = p$
Where $p$ is a set of changing features (e.g., gender flip).
Gradient Conditions: Two specific conditions are defined to determine the outcome of word formation:
1. Gradient Condition 1 (Non-Stationary State): Used for Conversion (e.g., noun-to-noun). If the base is masculine, the derived item shifts to feminine (and vice versa). The symmetric difference $p$ contains the gender flip set $\{-F, +M, +F, -M\}$ .
2. Gradient Condition 2 (Stationary State): Used for Semantic Widening. The symmetric difference $p$ is the empty set ( $\emptyset$ ), meaning no template shift occurs; the item retains the base template.

Data and Analysis

Data Source: Riffian (Ayt Waryaghel and Ayt Said varieties) and comparative data from French, Italian, and German.
Method: The author employs dimension reduction to filter out high-frequency, irregular, or historically complex words to identify "initial templates" for specific cognitive sets (Countable, Uncountable, Noun of Action).
Process Chains: The model traces 11 distinct derivational chains (e.g., Verb $\to$ Noun of Action $\to$ Countable/Uncountable) to demonstrate how specific formation processes trigger specific template shifts.

3. Key Contributions

Formalization of Item-Template Pairing: The paper introduces a distinct computational module for pairing items with templates, separate from morpheme pairing. It argues that gender is not a derivational marker but an integral part of the template assigned during formation.
Mathematical Proof of Intra-Lexical Conversion: The author provides a formal proof (Theorems 1 & 2) that intra-lexical conversion (noun-to-noun) is mathematically valid within this framework, challenging the view that such shifts require derivational affixes.
Unified Framework for Shifts: The model unifies the explanation for both gender shifts (non-stationary) and non-gender shifts (stationary) under a single mechanism governed by gradient conditions, rather than treating them as separate phenomena.
Rejection of "Faithfulness Presumption": The paper argues against theories that assume a strict form-to-form or meaning-to-form correspondence, proposing instead that templates are dynamic and determined by the cognitive set and formation process.
Application to Asymmetrical Encoding: The model successfully explains languages with partial asymmetrical gender encoding (like Riffian and Spanish), where gender markers may appear only at the beginning or end of a word, by treating them as separate affixes within a unified template.

4. Results

Predictive Accuracy: The TBMC model successfully predicts the gender and number markings for derived nouns in Riffian and French. For example, it correctly models the shift from a-gezːaɾ (butcher, M) to ða-gezːaɾθ (butchery, F) via the symmetric difference operation.
Identification of Initial Templates: The study identified specific "unmarked" initial templates for different cognitive sets:
- Countable Nouns: Feminine + Singulative.
- Uncountable Nouns: Feminine + Collective.
- Nouns of Action: Masculine + Singulative.
Mechanism of Shift: The results confirm that gender shifts are not random but are deterministically governed by the type of word formation (Conversion vs. Semantic Widening).
- Conversion triggers a template shift (Gradient Condition 1).
- Semantic Widening preserves the template (Gradient Condition 2).
Cross-Linguistic Validity: The model was applied to synthetic languages with and without definite articles (e.g., French vs. Riffian), demonstrating that the underlying structural patterns of template shifting are universal, despite surface-level differences.

5. Significance

Theoretical Impact: The paper bridges the gap between cognitive linguistics and formal mathematics. It moves linguistic theory away from static descriptive rules toward a dynamic, computational model of word formation.
Computational Linguistics & AI: By utilizing set-valued set functions, the model offers a new approach for Natural Language Processing (NLP) that handles non-linear mapping without requiring massive datasets or deep learning optimization. It is particularly valuable for low-resource languages where data is scarce.
Paradigm Shift in Morphology: It challenges the traditional distinction between inflection and derivation regarding gender, positing that gender is a property of the template assigned to an item, not a feature carried by a specific affix.
Future Applications: The framework suggests potential applications in modeling other non-linear dynamical phenomena in linguistics and potentially in other fields requiring precise modeling of set-based patterns (e.g., control systems, economics).

In summary, El Idrissi presents a rigorous mathematical model that explains grammatical gender shifting as a deterministic process of template shifting driven by set-valued functions, offering a unified solution to a long-standing problem in morphological theory.