Unexplored Area in Natural Language Processing: Cultural Differences and Interpersonal Communication

There is potential for Natural Language Processing in People Analytics to provide transnational corporations with a tool to improve interpersonal communication across it multicultural teams as well as “soften the blow” of cultural biases and misunderstandings, uncovering what’s otherwise “lost in translation”.

Reference Article: https://www.shrm.org/resourcesandtools/hr-topics/technology/pages/the-new-analytics-of-workplace-culture.aspx 

This article studies the effects of tools using language-based culture measures can raise ethical questions. I agree that, while algorithms make estimates, it is eventually humans’ responsibility to make informed judgments using them and management must keep metadata anonymous and should verify that algorithmic decision-making is not subject to bias and negative consequences. However, I believe there are still a large number of unexplored areas that natural language processing.

Source: https://www.analyticsinhr.com/blog/natural-language-processing-revolutionize-human-resources/

As we can see, there are many applications of Natural Language Processing for People Analytics. However, social information can vary greatly between language and cultures and research using text analysis methods should be expanded to capture those subtle cultural differences that would uncover social relations.

 For example, Dr. Keith Chen, a behavioral scientist, had discovered that languages vary in how much of a grammatical distinction they require speakers to make between the present and the future (M. Keith Chen, American Economic Review, 2013). His hypothesis was that languages that grammatically associate the future and the present, foster future-oriented behavior. For example, English requires speakers to talk about the future as different from the present (“it will rain tomorrow”), while German allows speakers to talk about the future as though it is the present (“Morgen regnet es” or “it rains tomorrow”). This example from German language is something that he refers to “futureless language” in an article about the way grammatical distinctions in languages affects economic behavior (The ‘Futureless zone’: Can language affect economic behavior? | DW | 24.06.2013)

So the way our language makes us think about time – affects our propensity to behave in the real world and across time. With English, a “futured language” that means every time you discuss a future event, you’re being made to cleave that from the present and treat it as if it’s something completely different, according to Dr. Chen. Suppose these difference makes you subtly dissociate the future from the present every time you speak. It then, makes the future feel something more distant and more different from the present, that’s going to make it harder to save, for example. If, on the other hand, you speak a futureless language, the present and the future, you speak about them identically. If that subtly nudges you to feel about them identically, that’s going to make it easier to making saving.

In a similar fashion, if a German colleague, whose language is “futureless” commits to a project and she is frustrated with the motivation of a colleague from a “futured” culture that affects interpersonal relationships, leading to conflicts – the kind of ‘fires’ that could be put out at the early stages and lead to smooth intercultural teams’ collaboration. This is the type of issue that could potentially be resolved by the use of Natural Language Processing in bridging a gap in language difference, lowering misunderstanding and “translating” the subtle cultural meanings, without human interference.

There is potential for Natural Language Processing in People Analytics to provide transnational corporations with a tool to improve interpersonal communication across it multicultural teams as well as “soften the blow” of cultural biases and misunderstandings, uncovering what’s otherwise “lost in translation”.

It is also, an interesting topic to explore in workforce motivation strategies across a diverse range of cultures, including the language data for bilingual or multilingual employees. Employees would not have to rely solely on corporate intercultural training and fragmented work experiences alone, and instead, utilize factful ‘off-the-shelf’ tools as both People Analytics and Natural Language Processing becomes more widespread.

Previous:

Course correcting your leadership journey with people analytics

Next:

A world with deep people analytics capabilities allow us to experiment with radical ideas like the “Talent to Value” method

Student comments on Unexplored Area in Natural Language Processing: Cultural Differences and Interpersonal Communication

  1. Aliya, thank you for covering this article. It gives a detailed review of how a cultural background’s effect is easy to misinterpret. At the same time, it offers research based ways to take advantage of the cultural/language differences to foster innovation in an organization. A meta point that is not specifically mentioned in that article but one that must be named is that true diversity is important in organizations as well as in developing the algorithms that these organizations use in their people analytics. Without such diversity in building the tools, the biases and errors that often exist in homogenous teams will easily migrate to the models and algorithms they produce.

    1. Tyuchi, thank you very much for your excellent observation and comment. I absolutely agree that having algorithms “fed” with that ‘diversified’ data would significantly improve its ability to withstand the biases and errors within homogenous teams. Just like the decision to diversify teams, the motivation to focus on expanding its scope remains within the leaders’ decision-making realm. While it is easy to justify team diversity for “productive conflict” effect as well as to a greater range of talent and “world-views”, it might be harder to consistently motivate them to include these data into analytical dimension. For example, racial bias awareness is such a zeitgeist of the moment but will it always be…

      Another concern I have is with data becoming more convoluted, cleaning it up from the analytics standpoint would be hassle-some and therefore, priorities might be drawn away from exploring cultural differences. Leaving it only to a few large transnationals to explore. Again, thank you for engaging in this discussion with me and your excellent point!

  2. Thanks for bringing in this topic, Aliya. I think it connects very well to the discussion we had about the video critique platform, Quantified AI. I think it is a great idea to be inclusive of different aspects such as a language’s grammar and a region’s dialect. However, I find the execution of this idea to be overwhelmingly challenging as these minor details can/do change with time. I don’t believe that one company will crack the code on this. Instead, it will be an uncoordinated group effort from a plethora of sources that can make this happen and keep the information updated.

    1. Shekeyla, thank you very much for your note and the tie to the QC case discussion. The reading for it and that discussion is what inspired me to explore this topic further. By the way, I appreciated your comment to Quantified Communications and opening up about being concerned of the ‘biased comparison’ of its AI algorithm. Thank you for not being afraid of being vulnerable. You surfaced an important concern.

      I absolutely agree with your prediction there with potentially a range of groups from various fields working on it. It might be a good thing with competition driving the innovation. As long as there is demand for cracking the ‘culture code’ and creating sustainable business models around it to continue the updates.

  3. Excellent point, Aliya! Thank you for bringing this point up. It’s a constant struggle in my multicultural home communications. For instance, the way my husband and I describe dates in the future (this Saturday or next Saturday) mean very different things and even after 21 years of being together those epistemological models are difficult to overcome. The way different cultures address others or how straight-forward they are can confuse the message, and having tools created with a truly multicultural ethos would help so much.

    1. Ana, thank you so much for sharing this personal life example. Glad you support the development of tools for resolving it as much as I do, having lived in multicultural environments since a very young age. Tolerance and understanding for fostering strong inter-ethnical social connections comes from awareness about every bit of subtlety of these differences. The developments in AI make me cautiously optimistic about the tools for this to be there for us to use in our daily lives.

      In fact, you just made me realize there is definitely an interesting use case, making these tools helpful not just organizationally but also, having affordable B2C solutions developed. It will take some time and research, technology, resources (including, data resources!) for it to happen but the future is near.

Leave a comment