2024 ends with a second Trump administration about to take office in the United States. Trump’s election campaign made immigration a central issue and promised stricter immigration controls. As voters went to the polls, the 2023 American Community Survey (ACS) data release revealed a significant increase in the number of non-English speakers in the US, reaching 71 million or 21% of the country’s population. 

In fact, studies show that the surge in the foreign-born population reached a rate not seen since 1910. All the while, the Department of Homeland Security (DHS) also increased deportations to the highest levels in over a decade. 

These are conflicting trends, but constitutionally speaking, language access rights are unlikely to change under Trump and will continue to be a strong demand driver for language services. However, implementation of AI translation by non-language organizations and translation-as-a-feature (TaaF) add challenges and uncertainty to an otherwise strong 2025 business outlook.

We asked readers what they thought would be the impact of Trump’s re-election on demand for US language services, and most (62.5%) believe it will be net negative. Less than a quarter of respondents (20.8%) adopt a neutral position on the matter, while a small percentage (16.7%) see it as net positive.

AI Speech Translation Fever

There were constant mentions of AI speech translation in 2024 in the academic and AI commercial areas. For instance, in the last quarter, Apple researchers published a paper on speech-to-text translation (S2TT) highlighting limitations in the way S2TT-capable systems process and incorporate nuance in translation.

Then, in November, Microsoft announced the preliminary launch of a new mobile AI speech translation app for Azure subscribers. The Microsoft Translator Pro, as it is called, has similar features to those in other standalone mobile translation apps, like Samsung’s Live Translate and Google Translate.

And it is not just translation. In November, OpenAI’s Whisper was in the news due to hallucinations in healthcare transcription. The AP reported that more than 30,000 clinicians and 40 healthcare systems have already begun using a tool from Nabla based on Whisper.

We asked readers how often they use AI speech-to-text tools, and most respondents (40.7%) said they never do. Close to a quarter (24.1%) rarely use it, and the rest use it often (14.8%), sometimes (11.1%), and very often (9.3%).

December to Remember?

Throughout 2024, there was no shortage of M&As and funding rounds in tech and services in the language industry, leaving an impression both of financial recovery (compared to what for many was a slow 2023) and optimism for the year ahead.

Slator reported on over 30 of these transactions and advised on some, with a few Super Agency and Leader language services providers (LSPs) completing more than one. Among others, TransPerfect acquired at least four companies and Translate.One acquired a minimum of three companies. Teleperformance acquired ZP in November, and Stoquart, LXT, and ArKadia all completed acquisitions in December.

Signs of a healthy industry notwithstanding, we asked readers about feelers for their business at year’s end, and for a little over a third (31.7%) it was quite busy. For two equal-size groups (20.5% each) business activity was normal or somewhat slow. For the rest of respondents, it was either very slow (18.2%) or the busiest ever (9.1%). 

Slow-Changing Pricing Models

Despite the rise of machine translation post-editing (MTPE) and increasingly automated translation workflows, the Association of Language Companies (ALC) 2024 survey revealed that 87% of language service providers still rely on the traditional per-word model. 

Alternative pricing methods based on edit distance or actual post-editing time have been discussed at industry events, but the subscription pricing model, with service levels tied to different degrees of editing (typically human) and automation, is only now starting to inch ahead.

The move towards subscription revenue in some corners has not gone unnoticed by investors and shareholders at companies like Australia’s AI Media, which has seen its shares soar by 250% since adopting a similar approach in April. 

We asked readers how they charge or pay for translation, and consistent with the ALC survey results, most (68.0%) are still charging or paying by the word. A small cohort (15.0%) has begun going by edit distance, while two identical groups are using a per hour, or another metric (7.0% each) and the rest (3.0%) are using the subscription model.