The product team launches a new feature and calls it a ‘workspace’. Marketing introduces it as a ‘collaboration area’. Sales has been talking to customers about a ‘portal’ for months. The help centre uses ‘team account’, while the subsidiary in another country has decided to keep the name in English.

None of these decisions seems serious in isolation. Each may have a reasonable explanation. The problem arises when a customer tries to connect what they saw in a campaign with what they find in the product and the instructions they receive from support.

Is this a translation error? Not necessarily.

It is a more structural problem: several functions have authority to shape the company’s language, but none is accountable for the entire language system.

Most international companies know who is responsible for their finances, data, brand, cybersecurity or customer experience. Yet when asked who governs language and multilingual communication from end to end, the answer usually becomes a list: marketing, communications, product, legal, country managers, an agency, the content team—or ‘it depends on the project’.

When the answer is ‘it depends’, the company does not have an owner. It has participants.

If no one is responsible for the language system as a whole, inconsistency is not a mistake: it is the logical outcome of the organisational structure.

Corporate language and the languages you use: two challenges companies often confuse

It is worth distinguishing between the two concepts.

Language is the system a company uses to define and explain its reality: product names, concepts, promises, instructions, sales arguments, tone, regulatory statements and style conventions. It exists even when the company operates in a single language.

Operating in multiple languages adds another dimension: that system must work across markets, for people with different habits, expectations, cultural frames of reference and ways of searching. The organisation must decide not only what it wants to say, but how to preserve meaning and achieve the intended effect in each context.

International business research describes the multinational corporation as a multilingual community, not a monolingual organisation that occasionally translates documents. Luo and Shenkar argue that multinational language systems are linked to its strategy, structure and degree of internationalisation, and that aligning them improves communication, coordination and knowledge sharing [1].

This perspective changes the question. The challenge is not simply ‘Who translates?’, but:

Who decides how the company names, explains, adapts, approves, reuses and measures the language it needs to operate?

That is a management challenge.

A cross-Functional Capability Missing from the Organisational Chart

Language runs through almost every part of the company, but each function uses it for a different purpose.

All these functions legitimately own part of the meaning. None, on its own, understands or controls the entire journey.

That is why the problem often goes unnoticed. Marketing sees an approved campaign. Product sees a functioning interface. Legal sees compliant copy. Support sees a clear response. The subsidiary sees an adaptation that works locally. Every team can meet its objectives while, collectively, producing an inconsistent global experience.

Fragmentation does not necessarily stem from poor collaboration. It stems from a structure in which each function has authority to create language (content), but no responsibility for its effects beyond that function’s boundaries.

The Real Cost Does Not Appear in a Budget Line Called ‘Lack of Governance’

The absence of language governance rarely appears as a separate cost. It is spread across the organisation.

It surfaces when someone from product and someone from marketing debate the same term for the third time. When a local team recreates a campaign because the original does not work. When support uses different wording to explain what the user sees on screen. When legal changes a concept and no one updates the help centre. When a subsidiary keeps its own spreadsheet because it does not trust the corporate reference. Or when knowledge disappears as the only person who remembers why a decision was made leaves the company.

A 2025 Adobe survey of more than 400 content professionals identified fragmented tool use, content silos and the lack of a unified strategy among the leading challenges [2]. Although the report examines content management as a whole rather than LangOps specifically, it captures the environment in which language problems arise: too many repositories, too many parallel workflows and too little shared visibility.

Terminology management provides a clear example. RWS points out that it is not simply a matter of maintaining a glossary: it requires defining terms with subject-matter experts, approving them, documenting their context, controlling changes, maintaining multilingual equivalents and integrating decisions into the tools where content is created [3]. Without this, the same symptoms recur: repeated corrections, conflicting instructions, variable results and customer enquiries prompted by confusing terminology [3].

The real cost therefore takes several forms.

None of these problems is solved by commissioning a better translation. They are solved by governing how decisions are made, shared and maintained.

Having a Style Guide Does Not Mean You Have Governance

Many organisations would respond by producing a brand PDF, a glossary, a translation memory or a folder of reference materials.

These assets are necessary. But they do not constitute an operating model.

Data governance offers a particularly useful distinction: a framework defines principles, policies and controls; a governance model determines who decides, who maintains the assets, how rules are applied and how conflicts are resolved [4]. The same distinction applies to language.

A guide may describe the brand’s voice, but it does not determine what happens when the legal department needs an exception. A glossary may contain an approved term, but it does not establish who can change it or how that update reaches product, support and the markets. A policy may state that consistency matters, but it does not allocate budget, authority or response times.

A document contains rules. Governance turns those rules into decisions, responsibilities and operational practice.

That is why a company can have dozens of language assets and still not know who governs its language.

Three Models: Central Control, Local Autonomy or Federated Accountability

There is no single structure for every organisation. Localisation governance models are commonly grouped into three broad approaches: centralised, decentralised and hybrid [5]. Data governance uses an almost identical classification, describing as federated a model that combines central standards with execution by individual domains [4].

For most international companies, a federated model provides the best starting point. It does not require a central office to approve every sentence, but neither does it allow each department to create its own version of the company.

The logic is simple: centralise what must be shared, and distribute what requires specialist or local knowledge.

Brand identity, strategic definitions, product names, risk criteria, the architecture of language assets and the process for recording decisions should, for example, be shared. Technical, legal and functional decisions should remain close to the relevant domain. Search intent, cultural conventions, acceptability, local tone and customer expectations should remain close to the market.

Governance does not mean taking authority away from those who understand the product or the country. It means preventing that knowledge from remaining isolated.

Your Company does not Need the Language Police

Appointing someone ‘responsible for language’ may seem like the answer. It can also create a new problem if the role is interpreted as reviewing and approving everything.

A central function that decides every term, campaign and exception will quickly become a bottleneck. Nor can it replace product expertise, legal accountability or the experience of local teams.

A mature approach will always separate four responsibilities:

1. System Ownership

One person or role must be accountable for the company’s end-to-end language capability. They do not need to write or review everything; they must ensure that principles, owners, shared assets, metrics and mechanisms for resolving conflicts are in place.

Depending on the company’s structure, this accountability may sit within operations, global marketing, customer experience or content, or it may belong to a dedicated LangOps function. The job title is secondary. What matters is that the authority is explicit and cross-functional.

2. Ownership of Meaning

Experts in each domain must decide what concepts mean. Product defines a feature; Legal, an obligation; Security, a warning; and Brand, a corporate promise.

A language specialist can identify ambiguity and shape the wording, but should not invent commercial policy, technical functionality or legal interpretation.

3. Market Ownership

Local market owners must decide how that meaning works in their context. They contribute insights into search behaviour, buying expectations, conventions, cultural sensitivities and the competitive landscape.

Their role is not to ‘sign off on the local-language version’, but to provide market evidence and record when an adaptation needs to depart from the global standard.

4. Operational Stewardship

A LangOps function turns these decisions into a reusable system. It maintains terminology, rules, context, approval workflows, versions, exceptions, integrations and measurement. Its job is not to replace the experts, but to ensure that their decisions reach every point where language is created.

This model centralises accountability and distributes knowledge.

A RACI Matrix for Language: Deciding Who Has The Final Say

A RACI matrix distinguishes who performs a task (Responsible), who is answerable for the outcome and holds final authority (Accountable), who must be consulted (Consulted) and who must be kept informed (Informed). A key recommendation is to assign just one accountable person to each decision to avoid conflicting leadership [6].

Applied to LangOps, the matrix should not assign responsibilities for ‘translation’ in general. It should assign them to specific decisions.

The precise matrix will vary from company to company. The value of the exercise lies in revealing three problems: decisions with no owner, decisions with multiple owners, and people who should contribute but only discover the outcome after publication.

The Company’s Shared Language Is An Asset Too

Governance is not only about preventing errors. It also helps build a shared way of thinking and working.

Aichhorn and Puck use the concept of company-speak to describe the vocabulary, terminology, expressions and communication rules shared within an organisation. Their qualitative research in European corporations suggests that this common language can support communication, shared understanding and corporate identity [7].

This does not mean filling the company with jargon. It means that strategic terms such as ‘active customer’, ‘critical incident’, ‘premium service’ or ‘launch’ should retain a recognisable meaning as they move from leadership to product, from product to sales, and from sales to each market.

The connection between language and operations is not abstract. An experiment involving local managers at multinational subsidiaries in Myanmar found that reducing language barriers increased interaction with foreign managers and enabled faster supervision of a simulated task without increasing errors [8]. The study examines language proficiency and knowledge transfer, not content governance. Nevertheless, it illustrates an important reality: language affects how instructions, knowledge and the ability to act flow through an organisation.

When a company governs its language, it not only communicates more consistently. It also coordinates more effectively.

Five Questions To Establish Whether Real Governance Exists

A company can make an initial assessment without buying technology or creating a new department. It simply needs to answer these questions honestly.

If the answer to three or more questions is no, the company probably carries out a great deal of language-related work, but does not yet have a governed language capability.

Five Actions To Get Started Without Adding Bureaucracy

The first step should not be drafting a fifty-page policy. It should be observing how decisions are made today.

1. Choose a critical journey. This could be a feature launch, customer onboarding or issue resolution. Follow the content from creation to use in two markets and record who changes it.
2. Identify repeated decisions. If the same term, tone or message is debated across several teams, the problem is not one of wording: a reusable decision is missing.
3. Assign a single point of final accountability. Every critical decision needs one accountable person. They may consult many others, but the organisation must know who resolves the disagreement.
4. Create a living decision log. Each entry should include meaning, context, affected languages or markets, owner, date, examples, exceptions and the reason for any change. A word list without context is not enough.
5. Measure friction before and after. Approval time, number of rounds, repeated corrections, incidents and recreated assets can show whether governance reduces work and improves the experience.

These actions do not require centralising every piece of content. They require making visible the decisions that are currently being lost.

How LangImpact Can Help

LangScore makes it possible to observe the consequences of fragmentation before redesigning the organisation. It can analyse consistency across channels, markets and content types; identify divergences in terminology, voice and suitability; and establish a baseline to prioritise the problems with the greatest impact.

LangFlow addresses the operational cause. It maps who creates, reviews, approves and publishes; defines decision rights; organises language assets; embeds rules into the workflow; and designs mechanisms that turn corrections and exceptions into reusable learning.

The distinction from a traditional translation service is essential. The objective is not to produce more words in other languages. It is to help the company make better language decisions, make them once, and apply them consistently across the organisation.

The Question Is Not Who Writes: It Is Who Is Accountable

Language is already a business function, even if the organisational chart does not yet recognise it. It affects brand, product, sales, compliance, support, internal knowledge and the ability to operate in new markets.

The organisation can continue treating it as a collection of projects distributed across departments and suppliers. In that case, teams will continue solving their own part of the problem, while the company continues paying the full cost of fragmentation.

Or it can assign accountability, distribute knowledge and turn language decisions into infrastructure.

The question is not who writes, translates or reviews each text. It is who is accountable for how language and multilingual communication work across the company.

If no one can answer that question, that is precisely where the problem begins.

References

[1] Luo, Y. and Shenkar, O. (2006). The multinational corporation as a multilingual community: Language and organization in a global context. Journal of International Business Studies. https://link.springer.com/article/10.1057/palgrave.jibs.8400197

[2] Adobe (2025). State of Long-form Content Management in the Age of AI. https://business.adobe.com/resources/reports/2025-survey-state-of-long-form-content-management.html

[3] RWS (2026). Terminology management in localization: the foundation of accuracy, consistency and brand trust. https://www.rws.com/blog/terminology-management-in-localization/

[4] Snowflake. Data Governance Models: Comparing Centralized, Decentralized and Federated Approaches. https://www.snowflake.com/en/data-governance/models/

[5] Translated. Governance Models for Global Localization: Organizational Structure. https://translated.com/resources/governance-models-for-global-localization-organizational-structure

[6] Atlassian. RACI Chart: What is it and how to use it? https://www.atlassian.com/work-management/project-management/raci-chart

[7] Aichhorn, N. and Puck, J. (2017). Bridging the language gap in multinational companies: Language strategies and the notion of company-speak. Journal of World Business. https://www.sciencedirect.com/science/article/abs/pii/S1090951617300287

[8] Guillouet, L., Khandelwal, A., Macchiavello, R. and Teachout, M. (2022). Language barriers in multinational companies: Evidence from Myanmar. VoxDev / NBER. https://voxdev.org/topic/firms/language-barriers-multinational-companies-evidence-myanmar