The core of the framework is a radical reframing of what fluency actually is. Rather than defining it simply as fast, accurate reading, Steklács proposes that fluency development changes how processing time is organized across connected text. In the earliest stages of reading, timing is dominated by word identification: children pause, hesitate, and stumble because decoding each word consumes attention and effort. But as word recognition consolidates through repeated exposure, timing becomes increasingly responsive to something else entirely, namely the demands of linguistic integration and the reader’s goals. A fluent reader may slow down deliberately at a complex sentence and then recover speed, and this selective slowing is not a lapse in fluency but potentially a hallmark of it.

This claim builds on well-established theoretical foundations. Automaticity theory, dating back to LaBerge and Samuels in 1974, holds that effortful word identification disrupts continuity and steals processing resources from comprehension. Synchronization theory emphasizes that reading depends on compatible timing among component processes. The Reading Systems Framework, developed by Charles Perfetti and colleagues, explains how lexical information is folded into an evolving mental representation of a text. Steklács draws on all of these but adds a new explanatory layer he calls connected-text coordination: a hypothesis about how successive word-identification events relate temporally to linguistic grouping and the ongoing updating of local meaning. Crucially, coordination is an explanation, not a new mental faculty, and the author is explicit that covariance among fluency scores cannot by itself establish that it exists.

To organize the developmental argument, the framework distinguishes four analytic categories that are often blurred together. Foundational resources, such as efficient word recognition, orthographic learning, lexical quality, phonological and morphological knowledge, and attention, enable fluent reading but are not themselves fluency. Core dimensions characterize fluent performance: accuracy, temporal continuity and pace regulation, and, in oral reading, prosodic organization. Observable manifestations are the modality-specific behaviors we can actually measure, including rate, errors, pauses, phrasing, acoustic prosody, fixations, regressions, and rereading. Finally, structural position denotes the role fluency plays in a statistical or theoretical model, whether outcome, direct contributor, mediator, or part of a reciprocal relation. A words-correct-per-minute score, the author notes, captures only a fraction of this architecture.

The framework proposes three overlapping developmental configurations. In the first, fluency is an emerging outcome of word-recognition development: recognition-related interruptions predominate, and continuity of processing is the prize. In the second, fluency acts as a mediator and coordinator of connected-text processing, with time being allocated selectively to unresolved linguistic relations. In the third, fluency participates in reciprocal, task-dependent relations with meaning construction, as the developing interpretation and reading purpose guide subsequent processing. These configurations are not rigid stages. Their number and ordering remain hypotheses, and their relative prominence may vary across readers, languages, texts, tasks, and even within a single reader confronting an unfamiliar passage.

The most consequential element of the paper is its primary hypothesis, a joint developmental prediction about what fluency measures actually tell us. As word identification consolidates under comparable reading conditions, Steklács predicts that the association between global reading rate and comprehension will weaken, while the association between demand-sensitive local timing and comprehension will remain stable or strengthen. In other words, how fast a child reads a whole passage will gradually become less informative about how well the child understands it, whereas where and how the child distributes time around integration demands will become more informative. This joint pattern is not entailed by faster word identification alone; it follows from the proposed change in which processes constrain comprehension. Longitudinal studies with preregistered multilevel models, counterbalanced texts, and independently specified processing regions would be needed to test it against alternative accounts.

Secondary hypotheses extend the framework to rereading and to modality. One predicts that when word-identification demands are comparable, readers will reread more selectively when resolving a contextual inconsistency serves their assigned goal, and that this selectivity will relate to understanding. Another predicts that local timing patterns responsive to integration demands will inform about comprehension in both oral and silent reading, even though the two modalities are demonstrably non-equivalent. Oral reading couples text processing with articulation, eye-voice coordination, and overt prosody, while silent reading permits flexible pacing and rereading. Evidence cited in the analysis shows age- and reader-group differences in relative oral and silent fluency, including a larger silent-reading advantage in adolescents than in children, supporting the idea of conditional overlap rather than interchangeability.

Prosody receives particularly nuanced treatment. A 2022 meta-analysis found an overall association between reading prosody and comprehension, but the framework argues that the interpretive meaning of prosody may shift with development. Early prosody may be constrained mostly by decoding and articulation, so flat or fragmented reading does not necessarily imply weak syntactic understanding. As identification improves, prosodic organization may become a more informative window onto linguistic grouping, and in advanced readers, phrasing and emphasis may reflect discourse, genre, and communicative intention. This implies that a single scoring rubric applied across ages could misclassify restrained or unconventional expression as disfluent. Longitudinal findings from Dutch schoolchildren, in which Grade 4 comprehension predicted Grade 5 prosody and Grade 5 prosody predicted Grade 6 comprehension, are compatible with a reciprocal relationship without proving a monotonic shift.

The framework also grapples with linguistic and orthographic variation. Children learning more consistent orthographies, such as Finnish or Hungarian, typically acquire accurate decoding earlier, yet fluency differences can remain salient after accuracy has largely stabilized. Cross-language evidence from Finnish, Hungarian, Dutch, Portuguese, and French shows shared predictors with varying relative influence, and in agglutinative Hungarian, a prior study of 4,134 children in Grades 2 through 4 found moderate positive associations between morphological awareness and reading comprehension across all grades. Orthographic depth, the analysis argues, changes which aspects of accuracy and timing discriminate among readers without necessarily implying a different underlying reading architecture. Transparency, in other words, should not be equated with processing simplicity.

Compensatory profiles add a further constraint on the theory. Studies of university students with dyslexia cited in the paper show weaker word- and text-reading fluency alongside good untimed comprehension, supported by listening comprehension, vocabulary, and background knowledge. Preserved comprehension, the framework insists, establishes neither efficient processing nor successful coordination, and additional time plus selective rereading may be the means by which understanding is sustained when identification remains effortful. For assessment and instruction, the practical upshot is that measures should match the developmental question: unstable recognition calls for attention to accuracy, integration demands call for timing and error profiles, and advanced reading may require silent measures of pacing and adaptation. The framework’s explanatory value, the author acknowledges, depends on its commitments distinguishing it from existing accounts, and its hypotheses remain open to revision if the predicted divergence between global rate and local timing fails to appear under adequately sensitive conditions.

Subject of Research: A developmental framework proposing that reading fluency reorganizes the temporal coordination of word recognition, integration, and comprehension during first-language reading development

Article Title: Beyond the bridge: an integrative developmental framework for first-language reading fluency

Article References: Steklács, J. (2026). Beyond the bridge: an integrative developmental framework for first-language reading fluency. Frontiers in Language Sciences, 5, Article 1955851. https://doi.org/10.3389/flang.2026.1955851

Image Credits: AI Generated

DOI: 10.3389/flang.2026.1955851

Keywords: reading fluency, reading comprehension, word recognition, reading development, prosody, oral reading, silent reading, automaticity, orthographic depth, morphological awareness, assessment, child development

The motivation for the study stems from a methodological bottleneck. As voice assistants, synthetic narrators, and conversational agents become embedded in daily life, researchers increasingly need experimental stimuli in which speaker identity and prosodic style can be manipulated independently. If a scientist wants to test how listeners react to a confident versus a doubtful statement, the comparison is confounded when different human speakers deliver the confident and doubtful versions, because listeners may respond to the person rather than the prosody. Voice cloning promises a solution: train a model on one speaker’s voice, then generate new sentences carrying a target emotional or attitudinal tone. But this only works if the cloning system genuinely transfers prosody rather than flattening it into a generic synthetic delivery.

To test this, the researchers recruited ten native Mandarin speakers who each produced thirty sentences, fifteen drawn from geography statements and fifteen from trivia statements, in three distinct intonations: confident, doubtful, and neutral. This design yielded a rich corpus of human recordings spanning different sentence contents and prosodic intentions. For each speaker and each prosody, the team trained separate AI voice clones using either the geography recordings or the trivia recordings as training material. Each clone was then asked to generate all thirty sentences, producing a total of 2,700 utterances for analysis: 900 human recordings, 900 AI-generated utterances from geography-trained clones, and 900 from trivia-trained clones. Crucially, the training and generation sets were crossed, so a clone trained only on confident geography sentences still had to produce doubtful trivia sentences, forcing the system to generalize prosodic style to entirely novel content.

The analytical strategy combined machine learning classification with traditional statistical modeling. The researchers extracted 88 acoustic features using the extended Geneva Minimalistic Acoustic Parameter Set, or eGeMAPS, a standardized toolkit widely used in voice research and affective computing. They then trained gradient-boosted tree classifiers, a machine learning approach implemented through the XGBoost framework, to distinguish confident from doubtful prosody within each source of speech. Within a single source, whether human or AI, the classifiers achieved high accuracy, averaging 0.85, indicating that the acoustic signatures of confidence and doubt are robust and detectable in both natural and synthetic voices. More importantly, when classifiers trained on human speech were tested on AI speech, and vice versa, accuracy remained substantially above chance, averaging 0.65. This cross-source generalization demonstrates that the prosodic cues of confidence and doubt are encoded in a shared acoustic currency that transcends the boundary between human and machine voices.

Among the 88 features, one emerged as a consistently dominant contributor: spectral flux, a measure of how rapidly the frequency content of the speech signal changes over time. Spectral flux captures the crispness and dynamism of articulation, and confident speech, in both humans and their AI clones, exhibited higher spectral flux than doubtful speech. This makes intuitive sense. A confident speaker articulates with decisive energy, producing sharper spectral transitions, whereas a hesitant speaker tends toward softer, less defined acoustic edges. Notably, spectral flux showed no effect of speaker sex, which the authors interpret as evidence that it indexes prosodic style rather than stable anatomical differences between speakers. In other words, it is a marker of how something is said, not of who is saying it, which is precisely the kind of feature a stimulus designer would want to manipulate independently of speaker identity.

Linear mixed-effects models, a statistical framework that accounts for the nested structure of the data with multiple speakers, sentences, and prosodies, confirmed a coherent acoustic profile of vocal confidence. Across human and AI speakers alike, confident prosody was associated with higher spectral flux, a longer estimated vocal tract length, and lower fundamental frequency, the acoustic correlate of pitch, compared with doubtful prosody. The estimated vocal tract length, derived from formant frequencies, reflects how speakers shape their resonating cavities; a longer apparent vocal tract projects a larger, more authoritative body, echoing a well-documented literature on vocal size exaggeration in humans. That AI clones reproduced this constellation of cues, including the anatomical-sounding shift in apparent vocal tract size, suggests that modern cloning systems capture not only surface acoustics but the embodied gestalt of an attitudinal vocal stance.

The study also probed how prosodic information unfolds in time. Using time-resolved decoding of fundamental frequency, the researchers found that the distinction between confident and doubtful prosody was most reliably recovered in the late windows of each utterance, across human and AI sources alike. This late-emerging pattern aligns with the intuition that speakers often commit to an attitude as a sentence progresses, with final intonational contours sealing the impression of certainty or hesitation. However, a subtle asymmetry appeared: human speakers distributed their prosodic markers more broadly across the utterance, weaving cues of confidence and doubt throughout the signal, whereas the AI clones concentrated their prosodic differentiation more narrowly toward the end. This temporal signature may represent one of the remaining acoustic fingerprints distinguishing synthetic from natural expressive speech.

Indeed, the authors are careful to note that despite the impressive prosodic fidelity, AI-generated speech still forms a distinguishable acoustic population relative to human speech. The cross-source classification accuracy of 0.65, while well above chance, falls short of the 0.85 achieved within sources, indicating that machine and human voices are similar but not acoustically identical. This finding resonates with recent work showing that voice clones sound realistic but not yet hyperrealistic, and that human listeners and neural systems can, under some conditions, separate deepfake from genuine speaker identity. For the immediate purposes of psychological research, however, the residual human-machine gap may even be an advantage, since it allows researchers to verify that their synthetic stimuli behave acoustically as intended while retaining a measurable distinction from natural recordings.

The broader significance of the study lies in its methodological contribution. By demonstrating that voice cloning preserves prosodic style alongside speaker identity, Chen and Jiang provide empirical license for a new generation of controlled experiments on human-AI voice interaction. Researchers can now, with appropriate validation, generate stimulus sets in which the same voice delivers the same content with systematically varied confidence, doubt, or neutrality, isolating the causal impact of prosody on listener trust, memory, persuasion, and social perception. As AI voices become conversation partners, teachers, and companions, understanding how their vocal expressions of certainty are produced, perceived, and potentially distinguished from human ones becomes a scientific priority. This study shows that the tools for doing that science rigorously are already within reach, and that the line between human and machine vocal expression, while still detectable, is growing finer by the year.

Subject of Research: Whether AI voice cloning systems can reproduce human prosodic expressions of confidence and doubt

Article Title: Voice-cloning artificial-intelligence speakers can also mimic human-specific vocal expression

Article References: Chen, W., & Jiang, X. (2026). Voice-cloning artificial-intelligence speakers can also mimic human-specific vocal expression. Current Psychology, 45(17), Article 1490. https://doi.org/10.1007/s12144-026-09991-w

Image Credits: AI Generated

DOI: 10.1007/s12144-026-09991-w

Keywords: voice cloning, artificial intelligence, prosody, speech synthesis, vocal confidence, spectral flux, machine learning, human-AI interaction, acoustic analysis, eGeMAPS, fundamental frequency, Current Psychology