Research Should Begin With Evidence, Not Assumption

The PFA Research Programme is intended to examine the characteristics of domain names through evidence, research and structured investigation.

The framework draws upon knowledge from several fields relevant to phonetic, linguistic, technological and human-computer interaction.

These sources help inform the questions that PFA asks, but do not automatically validate the PFA Framework itself.

That distinction is fundamental.

A published study may provide evidence that a particular linguistic phenomenon exists. A technical paper may demonstrate a limitation within speech recognition. An experiment may identify a measurable difference between two groups of domain names.

And while each may contribute to our understanding, none should be treated as proof of a broader proposition without examining what the evidence actually demonstrates.

The purpose of this section is therefore to establish the evidence base from which the PFA Research Programme develops, while maintaining a clear distinction between evidence, interpretation and validation.

What Does The PFA Evidence Base Include?

The PFA Research Programme may consider evidence from a range of relevant disciplines and sources.

Academic Research

Peer-reviewed research in areas such as:

  • Phonetics.
  • Phonology.
  • Psycholinguistics.
  • Language processing.
  • Lexical processing.
  • Naming and brand-name research.
  • Pronunciation.
  • Speech perception.
  • Related linguistic disciplines.

Such research may help establish what is known about how linguistic and phonetic characteristics are perceived, processed or represented.

Speech and Language Technology Research

Research concerning:

  • Automatic speech recognition.
  • Speech-to-text systems.
  • Conversational interfaces.
  • Voice interaction.
  • Named-entity recognition.
  • Pronunciation modelling.
  • Contextual speech recognition.
  • Language models.
  • Emerging agentic or conversational technologies.

This area is particularly relevant to the changing environments in which domain names may increasingly be encountered through voice and conversational interfaces.

Experimental Research

Controlled experiments and structured testing may provide evidence that is more directly relevant to individual PFA propositions.

Where appropriate, the Research Programme may conduct or examine experiments involving human participants, speech systems, domain-name datasets or other measurable variables.

Empirical Data

Datasets and measurable observations may provide another form of evidence.

These may include domain-name characteristics, pronunciation results, recognition results, linguistic classifications or other structured observations relevant to a particular research question.

Industry and Market Research

Industry research may provide useful context concerning naming practices, domain usage, technology adoption and changing digital environments.

Such material can help identify questions worthy of further investigation.

It should not automatically be treated as equivalent to peer-reviewed academic research or controlled empirical evidence.

Independent Contributions

Researchers, linguists, academics, practitioners and other knowledgeable contributors may submit relevant material through the PFA Research Contact process.

Submitted material may include, but is by no means limited in scope to:

  • Published research.
  • Datasets.
  • Experimental findings.
  • Technical documentation.
  • Comparative and analytical observations.

Additional, supportive and substantive evidence seen, considered, or interpreted to be of relevance to the programme will always be considered and evaluated.

Not All Evidence Serves The Same Purpose

Evidence should not be reduced to a simple hierarchy in which one category is automatically considered valuable and another automatically dismissed.

Different research designs answer different questions.

A controlled experiment may be appropriate for one question while an observational study, linguistic analysis or technical investigation may be more appropriate for another.

The quality of the methodology, the relevance of the evidence, the limitations of the study and the question being examined all matter.

Evidence should therefore be evaluated in context rather than according to source type alone.

A well-designed study may provide useful evidence even when it does not directly test PFA.

Conversely, a poorly designed study should not be treated as strong evidence simply because its subject appears closely related to PFA.

How Sources Are Interpreted

The PFA Research Programme seeks to distinguish between what a source actually demonstrates and what may reasonably be inferred from it.

For each relevant source, consideration may be given to:

What Was Studied?

What question, phenomenon, population, technology or dataset did the research actually examine?

How Was It Studied?

What methodology was used?

Was the work experimental, observational, computational, theoretical or another form of investigation?

What Was Found?

What were the actual findings?

What Are the Limitations?

What conditions, assumptions, sample characteristics, methodological limitations or uncertainties affect interpretation?

How Is It Relevant to PFA?

Does the source directly address a PFA proposition, provide supporting context, challenge an assumption, or simply inform the broader subject?

This final distinction is particularly important.

A source may be highly relevant to the subject of phonetic fluidity without directly validating a PFA assessment principle.

Supporting And Challenging Evidence

The PFA Research Programme does not seek evidence that supports PFA alone.

Evidence that challenges an existing assumption is equally relevant.

Sources may therefore be considered according to their relationship to a particular research proposition.

Supporting Evidence

Evidence that is consistent with a proposition being examined by PFA.

Challenging Evidence

Evidence that contradicts, weakens or places limitations upon a proposition.

Contextual Evidence

Evidence that provides useful background or establishes a relevant phenomenon without directly testing a PFA proposition.

Experimental Evidence

Evidence produced through a structured experiment or test relevant to a PFA research question.

Unresolved Evidence

Evidence that raises a meaningful question but does not currently provide sufficient grounds for a reliable conclusion.

These classifications may well describe the relationship between evidence and a research proposition. However, they may not necessarily represent a recognised and universal ranking of scientific quality.

The Importance Of Primary Sources

Where possible, the PFA Research Programme should identify and reference the original research rather than relying solely upon summaries or secondary commentary.

Primary sources provide access to the methodology, data, findings and limitations on which a conclusion is based.

Secondary sources may still be useful for context, discovery and synthesis, but their interpretation should not automatically replace examination of the underlying research.

Where a claim depends upon a particular study, the relevant source should be identifiable.

Evidence, Interpretation And Conclusion

The Research Programme distinguishes three related but different stages.

Evidence

What has been observed, measured, reported or established by a source?

Interpretation

What might that evidence mean in the context of the research question?

Conclusion

What can reasonably be stated after considering the evidence and its limitations?

These stages should not be collapsed into one.

For example, research may clearly demonstrate that certain pronunciation characteristics influence human processing.

That is evidence.

However, interpreting that finding as potentially relevant to domain-name communication is a further, and therefore different step.

Concluding that a specific PFA Score predicts a commercial outcome would require additional evidence to qualify such a claim.

The strength of a conclusion should therefore reflect the strength and relevance of the evidence supporting it.

Understanding The Difference Between Correlation And Causation

The PFA Research Programme will exercise particular caution when interpreting relationships between variables.

A relationship between two observed characteristics does not necessarily establish that one causes the other.

For example, if a particular domain characteristic is associated with an observed outcome, several explanations may remain possible:

  • Other variables may be responsible.
  • The relationship may be coincidental.
  • The sample may not be representative.
  • The observed relationship may exist only within a particular environment.
  • The relationship may be genuine but substantially more complicated than initially assumed.

For this/ these reason(s), observational evidence may identify important relationships without establishing a causal mechanism.

The PFA Research Programme recognises where stronger causal claims are considered, the supporting evidence should be correspondingly stronger.

Reproducibility And Replicability

Confidence in research findings can increase when results can be reproduced or replicated under appropriate conditions.

The distinction is useful because:

Reproducibility concerns obtaining consistent computational results using the same data, methods and analytical conditions.

Replicability concerns obtaining consistent results across studies addressing the same question using newly collected data or independently conducted work.

The PFA Research Programme recognises that neither should be treated as a simplistic pass-or-fail concept.

Research findings can vary for legitimate reasons, and differences between studies may reveal limitations, contextual effects or new questions.

Nevertheless, independent consistency can provide greater confidence than an isolated result.

Where PFA-related research becomes sufficiently developed, reproducibility and independent replication should therefore form part of the broader evidence assessment.

Uncertainty Matters

Research findings are rarely free from uncertainty and sources of uncertainty may include:

  • Sample size.
  • Sample composition.
  • Language.
  • Accent.
  • Geographic variation.
  • Experimental conditions.
  • Technology.
  • Measurement methods.
  • Data quality.
  • Model assumptions.
  • Limitations in the research design.

The presence of uncertainty does not make research unusable.

Uncertainty means that conclusions should be expressed at a level of confidence appropriate to the evidence available.

A research programme does not become credible, or more so, when uncertainty is hidden rather than openly acknowledged.

Cross-Linguistic Evidence

PFA concerns characteristics that may interact with language.

A pronunciation that appears relatively straightforward in one language may not behave identically in another.

Differences may arise from:

  • Phonotactic rules.
  • Pronunciation conventions.
  • Orthography.
  • Accents.
  • Native-language effects.
  • Unfamiliar sound combinations.
  • Differences in speech-recognition systems.

Consequently, evidence obtained in one linguistic environment should not automatically be generalised to all languages.

Cross-linguistic evidence may therefore become increasingly important as the PFA Research Programme develops.

Technology Changes The Evidence Environment

The technological systems through which domain names are encountered are not static.

Speech-recognition systems, conversational interfaces, AI assistants and agentic systems continue to develop.

A finding observed in one technological environment may therefore not remain equally applicable in another.

This creates an additional reason and need for continued research.

PFA should not assume that a characteristic has a permanent technological effect simply because it produces a particular result under one system or at one point in time.

Where technology changes materially, relevant assumptions may need to be re-examined.

What Evidence Does Not Establish

The existence of relevant research does not automatically establish that:

  • PFA predicts domain value.
  • A particular PFA Tier translates into a financial valuation.
  • A PFA Score predicts commercial performance.
  • A higher score guarantees better recognition.
  • A lower score means a domain has no or is of low economic value.
  • A domain will necessarily be easier for an AI system to resolve.
  • A particular phonetic characteristic causes increased traffic.
  • PFA predicts conversion or revenue.
  • The current methodology represents a final or universally accepted model.

Such hypothesis either in part or whole would require additional evidence in order to support any individual claim.

The PFA Research Programme therefore seeks to avoid extending a source beyond the conclusion that its evidence can reasonably support.

Building The PFA Evidence Library

As the PFA Research Programme develops, relevant sources may be organised into a structured evidence library.

Individual sources may eventually include information such as:

  • Title.
  • Author or organisation.
  • Publication date.
  • Publication or source.
  • Research area.
  • Methodology.
  • Relevant finding.
  • Limitations.
  • Relationship to PFA.
  • Evidence classification.
  • Associated PFA proposition or research question.

This structure is intended to make the evidence base easier to examine and understand as well as provide a foundation for future research papers and validation.

Evidence And The PFA Methodology

The relationship between evidence and methodology should remain explicit.

The PFA Methodology describes the current PFA Framework used to assess phonetic and linguistic characteristics.

The Evidence & Sources section of the PFA Research Programme records the research and evidence relevant to the concepts underlying that framework.

Where new evidence materially affects a methodological assumption, the implications should be considered through the appropriate research and methodological processes.

This may ultimately result in:

  • No change.
  • Additional research requirements.
  • A qualification.
  • A revised interpretation.
  • A methodological amendment.
  • The removal of an unsupported assumption.

The existence of an evidence base therefore does not mean that the methodology is permanently fixed.

Evidence Does Not Replace Validation

Evidence & Sources and PFA Validation serve two differing and independent purposes.

Evidence & Sources asks:

"What relevant evidence exists, and what does that evidence actually tell us?"

PFA Validation asks:

"Does the PFA Framework itself withstand structured examination and testing?"

It is essential to understand and differentiate between the two.

  • The first provides an evidence base.
  • The second examines the framework.

Keeping these functions separate helps prevent a common methodological error: treating research about a related subject as proof that a particular model or scoring system is valid.

An Open Evidence Base

The PFA Research Programme is intended to remain open to evidence that supports, challenges or complicates the current PFA Framework.

It is therefore implied that:

  • A source does not need to support PFA to be relevant.
  • A contradictory finding may identify a limitation.
  • An unexpected result may reveal a new research question.
  • A failed experiment may demonstrate that an assumption requires reconsideration.
  • A finding that cannot currently be replicated may require further investigation.

Such outcomes are part of the research process, which is why the objective of the PFA Research Programme is not to accumulate evidence in favour of a predetermined conclusion.

The objective is to develop a clearer understanding of what the available evidence can actually support.

The Principle

Follow the evidence beyond the assumption.

PFA begins with questions about the characteristics of domain names in an increasingly spoken, conversational and machine-mediated digital environment.

The PFA Research Programme exists to investigate those questions, and :

  • Where evidence supports an assumption, it should be documented.
  • Where evidence challenges it, that should also be documented.
  • Where evidence is insufficient, the uncertainty should remain visible.
  • Where the evidence requires the framework to change, the PFA Framework should be capable of changing.

From Evidence To Validation

Evidence provides the foundation for further examination.

The next stage is therefore not to assume that the evidence validates PFA, but to ask whether the framework itself withstands testing.

That process is addressed separately through PFA Validation.

Evidence & Sources establishes what is known, what has been observed and what research is relevant.

PFA Validation examines what happens when the PFA Framework is subjected to structured testing.

Together, these form an important part of the continuing development of the PFA Research Programme.