Unary Quantifiers in Regular Language Theory

• Unary quantifiers over regular languages are logical and algebraic tools that mark specific positions to analyze word structure and periodic patterns.
• They employ methods from first-order logic, modular arithmetic, and dynamic descriptive complexity to precisely characterize language properties and expressiveness.
• Research reveals a dichotomy in quantifier rank growth between aperiodic and cyclic languages, providing clear boundaries for decidable separation and membership problems.

Unary quantifiers over regular languages are studied in several closely related senses: as first-order quantification over positions in finite words, as generalized quantifiers such as existential, universal, and modular counting quantifiers applied to marked-word languages, as resources in dynamic descriptive complexity with unary auxiliary relations, and, in the unary-alphabet setting, as operators interpreted through ultimately periodic encodings of regular languages. Across these settings, the central questions concern expressive power, quantifier rank, recogniser transformations, algebraic characterizations, and decision complexity. The cited literature places these questions at the intersection of automata theory, syntactic monoids, concatenation hierarchies, codensity/profinite monads, and $\omega$-regular semantics (Gehrke et al., 2017, Bazarova et al., 29 Apr 2026).

1. Logical and automata-theoretic setting

First-order logic over words quantifies over positions and uses predicates such as $<$, $+1$, and unary letter predicates $P_a$. In the quantifier alternation hierarchy, a sentence is in $\Sigma_n$ if it is logically equivalent to a prenex sentence with at most $n$ quantifier blocks starting with $\exists$, and $\mathcal{B}\Sigma_n$ consists of finite Boolean combinations of $\Sigma_n$ sentences. In the dot-depth correspondence reported in the literature, dot-depth $n$ corresponds to $<$0, and dot-depth $<$1 corresponds to $<$2 (Place, 2017).

A second, more algebraic view treats unary quantification as an operation on languages with one marked position. In this setting, existential quantification asks whether there exists a marked position making the marked word belong to a base language, while modular quantification asks whether the number of marked positions satisfying the base language is congruent to a prescribed value modulo $<$3. The general construction described for these operators models quantification by an $<$4-valued transduction

$<$5

with $<$6 for existential quantification and $<$7 for modular quantifiers (Gehrke et al., 2017).

A third setting arises for unary alphabets. Over a one-letter alphabet, regular languages are ultimately periodic, and this arithmetic structure makes quantifier phenomena especially explicit. The literature defines, for a unary language $<$8, the bit sequence $<$9 with $+1$0 iff $+1$1, and the associated $+1$2-word $+1$3. This encoding is used to interpret unary quantifiers over regular languages as MSO-formulas on the corresponding $+1$4-word (Czerwiński et al., 2023).

2. Finite-horizon quantifier rank and the aperiodicity gap

The finite-horizon first-order rank profile of a language $+1$5 is the function

$+1$6

where $+1$7 means that $+1$8 and $+1$9 satisfy the same $P_a$0 sentences of quantifier rank at most $P_a$1. It measures the least quantifier rank needed by an $P_a$2 sentence to classify membership in $P_a$3 correctly on all words of length at most $P_a$4 (Bazarova et al., 29 Apr 2026).

For every language, regular or not, the general upper bound

$P_a$5

holds. The same work proves that $P_a$6 exactly when $P_a$7 is globally $P_a$8-definable, and in that case the supremum equals the minimum quantifier rank of a global $P_a$9 definition. For regular languages, the result becomes a sharp dichotomy: if the syntactic monoid of $\Sigma_n$0 is aperiodic, then $\Sigma_n$1; otherwise,

$\Sigma_n$2

The lower bound extracts a nontrivial cyclic component from the syntactic monoid and combines it with an Ehrenfeucht-Fraïssé power lemma for long repetitions of a fixed word (Bazarova et al., 29 Apr 2026).

This yields a precise limitation theorem for unary quantification in $\Sigma_n$3: regular languages admit no intermediate finite-horizon growth between bounded and logarithmic rank. In the unary-alphabet case, the dichotomy is stated especially explicitly: only finite or cofinite languages are $\Sigma_n$4-definable and therefore have constant profile; all others have logarithmic profile (Bazarova et al., 29 Apr 2026).

3. Quantifier blocks without alternation

For first-order formulas without quantifier alternation, the regular case collapses to regular numerical predicates. The main theorem reported for this setting is

$\Sigma_n$5

with the analogous statement for $\Sigma_n$6. Equivalently, every regular language definable by a first-order formula with no quantifier alternation over arbitrary numerical predicates can also be defined by such formulas using only regular atomic formulas (Krebs et al., 2022).

A key device is the $\Sigma_n$7-ceiling of a language. For a regular language $\Sigma_n$8, the ceiling construction produces a best $\Sigma_n$9 approximation $n$0, and if $n$1 is regular then the associated predicates $n$2 are regular numerical predicates. For Boolean combinations, the paper uses a normal form in which every Boolean formula is equivalent to an iterated difference of monotone formulas, allowing repeated application of the ceiling construction while preserving regularity of predicates (Krebs et al., 2022).

The significance for unary quantifiers is that, at alternation depth $n$3, arbitrary numerical predicates do not add expressive power on regular languages. The collapse is specifically stated for formulas with only existential or only universal blocks, that is, for the most basic unary-quantifier regimes over words (Krebs et al., 2022).

4. Alternation hierarchies and separation

Beyond single quantifier blocks, unary quantifier alternation is organized by concatenation hierarchies. In the dot-depth hierarchy, the papers report Thomas’s correspondence: for each $n$4, dot-depth $n$5 corresponds to $n$6, while dot-depth $n$7 corresponds to $n$8. This identifies successive layers of regular languages definable with increasing numbers of quantifier blocks (Place, 2017).

The separation problem asks, for a class $n$9, whether given regular languages $\exists$0 there exists a language in $\exists$1 that includes $\exists$2 and is disjoint from $\exists$3. The cited result proves that separation is decidable for the level $\exists$4 of any concatenation hierarchy whose basis is finite, and, for the dot-depth hierarchy, decidable for level $\exists$5. In logical terms, this solves separation for $\exists$6, namely first-order sentences having at most three quantifier blocks starting with an existential one (Place, 2017).

These results delimit an effective boundary for regular-language analysis under bounded unary quantifier alternation. A plausible implication is that separation algorithms function as a structural probe of expressiveness: once a level admits separation, its logical content becomes substantially more analyzable. The papers also state a transfer to membership at the next level, reported as decidability for $\exists$7 in the corresponding setting (Place, 2017).

5. Recognisers, codensity monads, and quantified language operations

A general topo-algebraic account of unary quantifiers is given through Boolean spaces with internal monoids (BiMs). A BiM is a tuple $\exists$8 where $\exists$9 is a Boolean space, $\mathcal{B}\Sigma_n$0 is a monoid, $\mathcal{B}\Sigma_n$1 is dense, and $\mathcal{B}\Sigma_n$2 extend right and left monoid actions to $\mathcal{B}\Sigma_n$3. Languages are recognized as preimages of clopens under BiM morphisms (Gehrke et al., 2017).

The quantifier construction uses codensity monads and profinite monads. For the finite powerset monad, the codensity monad is the Vietoris monad. For the free $\mathcal{B}\Sigma_n$4-semimodule monad over a finite commutative semiring $\mathcal{B}\Sigma_n$5, the profinite construction yields a Boolean space of finitely additive $\mathcal{B}\Sigma_n$6-valued measures on the clopens of $\mathcal{B}\Sigma_n$7. Quantification is then recognized by lifting a base BiM $\mathcal{B}\Sigma_n$8 for the formula-with-variable to a new recogniser built from the free $\mathcal{B}\Sigma_n$9-semimodule on $\Sigma_n$0 and the corresponding measure space on $\Sigma_n$1 (Gehrke et al., 2017).

The explicit semantics matches familiar quantifiers. For existential quantification, the construction uses the Vietoris/powerset case and tests whether any marked variant lands in a designated clopen. For modular quantifiers, with $\Sigma_n$2, the recogniser computes the number of marked variants landing in a clopen modulo $\Sigma_n$3. The main theorem stated in this setting says that if a language $\Sigma_n$4 is recognised by a BiM morphism $\Sigma_n$5, then the quantified language $\Sigma_n$6 is recognized by the derived BiM. The corresponding Reutenauer-type theorem states that the Boolean subalgebra closed under quotients generated by the languages recognized by all length-preserving BiM morphisms into the derived structure is generated as a Boolean algebra by the base languages and their unary quantifications (Gehrke et al., 2017).

6. Unary auxiliary relations in dynamic descriptive complexity

Unary quantifiers also arise in dynamic descriptive complexity through update formulas using only unary auxiliary relations. The starting point reported in the literature is Hesse’s result that all regular languages can be dynamically maintained in first-order logic using only unary auxiliary relations. The refinement proved subsequently is that all regular languages can be maintained by update formulas in the fragment $\Sigma_n$7 with only unary auxiliary relations: $\Sigma_n$8 This uses Green’s relations on the syntactic monoid and unary relations encoding crucial properties of the word (Barloy et al., 26 Jan 2026).

At lower logical strength, the maintainable classes become algebraically sharp. With quantifier-free formulas and unary auxiliaries,

$\Sigma_n$9

With positive existential formulas and unary auxiliaries,

$n$0

The paper further states that the algebraic characterization for general existential $n$1 updates with unary auxiliaries remains open (Barloy et al., 26 Jan 2026).

These results make the role of unary quantification highly explicit. With one quantifier alternation, unary auxiliary data suffices for all regular languages; with quantifier-free or positive-existential updates, the maintainable class contracts to well-defined monoid-theoretic varieties. This aligns the dynamic setting with the older algebraic study of regular languages while preserving the emphasis on unary logical resources (Barloy et al., 26 Jan 2026).

7. Unary alphabets, $n$2-words, and quantifier-like decompositions

Over a unary alphabet, regular languages are studied through NFAs, UFAs, DFAs, Chrobak normal form, and ultimately periodic semantics. The survey paper on unary automata emphasizes that over a unary alphabet any NFA or UFA can be represented as a stem followed by a set of disjoint cycles, and for UFAs this normal form can be computed in polynomial time without increasing the number of states. In this setting, inclusion and equivalence for unary $n$3-state NFAs can be decided in time $n$4, improving the previous upper bound $n$5, with an ETH lower bound $n$6. For formulas involving only union, intersection, complement, and star, evaluation can be done in quasipolynomial time and POLYLOGSPACE; if concatenation is involved, evaluating such formulas or converting to a UFA needs exponential-type time. The same work defines $n$7 and states that the mapping $n$8 allows interpretation of unary quantifiers over regular languages as MSO-formulas on the corresponding $n$9-word (Czerwiński et al., 2023).

The $<$00-regular decision problem attached to this encoding is also explicit. Given a regular language $<$01 and a fixed $<$02-regular language $<$03 recognised by a Büchi automaton, the problem asks whether $<$04. If $<$05 is given by a DFA, this can be done in polynomial time. For UFA input, the upper bound reported is $<$06, with ETH lower bound $<$07; for NFA input, the upper bound is $<$08, with ETH lower bound $<$09 (Czerwiński et al., 2023).

Quantum automata over unary alphabets provide a parallel operational perspective. The construction of Latvian quantum finite state automata for unary regular languages decomposes any unary regular language into a finite part $<$10 and an ultimately periodic part $<$11, then combines two LQFAs recognizing these components with isolated cut point $<$12. The framework explicitly uses languages of the form $<$13 and $<$14, and the paper states that the general framework echoes the Boolean closure property of regular and group languages under unary quantifiers such as length and position modulo period. The resulting LQFA for $<$15 has a number of basis states exponential in the size of the minimal DFA for $<$16, and the isolation around the cut point also depends exponentially on the size of the finite prefix (Mereghetti et al., 2023).

Taken together, the unary-alphabet results show that unary quantification over regular languages is not only a logical notion but also an operational one. In this regime, thresholds, congruence classes, periodic tails, and ultimately periodic $<$17-words become the concrete carriers of quantifier semantics (Czerwiński et al., 2023, Mereghetti et al., 2023).