‘Linguistic-fying’ Your Bot to Chat Responsively with Multilingual Audiences

Chatbots – love them or hate them… they’re here to stay.

With today’s techy-influence customers wanting fast and prompt answers, businesses have to respond fast too, or lose a potential deal to a competitor.

Hoteliers, tourism and hospitality establishments are introducing online Chatbots to help increase the level of efficiency for interacting with customers. Call it your virtual messenger, customer assistant or a front desk avatar.

With the use of AI invading every facet of the business world, all of us will have to deal with bots one time or another.

Sometimes they can be spot on, while other times our experience can range from the usual “I’m sorry…”, to something that’s entirely not what we’re looking for or even an utterly incompetent response.

Chatbots that do not respond appropriately do get people annoyed and irritable. After several failed attempts, it will escalate the enquirer to feeling frustrated, perhaps hostile too, for wasting their time.

Understandably, Chatbots are created to be intuitive when responding to queries. But, unlike FAQs that are pretty static and users have to scroll all over, bots can help spark curiosity and even amuse people with their funny suggestions or laughable tips.

The objective of a bot system is to arrive at a definitive answer or an answer that will somewhat concur with the user’s intended query. Over time, these bots are trained to become smarter and to respond to dialogic phrases more predictively.

But the end goal of a bot is not to become human-like.

Rather, it is to be an effective robotic interlocutor equipped to imitate human conversations in whatever language it has been programmed to handle.

Implementing a Chatbot should not just stop at having a virtual messenger that only communicates in one’s own language. This is especially so for businesses that are serving global markets, particularly audiences of different linguistic and cultural backgrounds.

Having a monolingual Chatbot would surely be deprived of the opportunity to interact with the millions around the world who are not familiar in that particular language.

It’s even more so for hoteliers, tourism and hospitality service providers that want to increase their online engagement with major source outbound travellers’ markets, such as China, Southeast and Eastern Asia, the Middle East region, etc.

The business would be marginalised in those markets if it does not facilitate the first-level encounter with potential customers in their preferred language.

The Era of Polyglottal Chatbots

Several months ago, IPPWORLD assisted a leading Integrated Resort to revitalise their Chatbot to converse in additional languages besides English.

As a start, their online bot is now able to respond instinctively to queries directly from communities living in China and Indonesia… in their respective local languages.

Since the rollout of their virtual multilingual ‘customer-care assistant’, the client observed encouraging activities recorded by the bot as it keeps track on the interactions with enquirers, who are each using their respective everyday vernacular and informal slangs.

The benefits of having implemented a polyglottic Chatbot utility is limited only by one’s creativity and the professional support in Linguistic Transcreation.

Undoubtedly, a sensible and functionable Chatbot offers one of the best entry points for businesses to reach out and interact with global audiences… all in the comfort of their native language.

Developing A Conversational Multilingual Chatbot

Without delving into the technicalities or capabilities of AI, the most cost-effective approach for creating a Chatbot with multilingual capabilities is by engaging the support of a Language Service Provider (LSP) that’s familiar with the client’s line-of-business.

The appointed LSP must have proven experiences in offering transcreation (translations infused with creative editing), or commonly referred to as creative translations.

Pre-localisation workflows involving the bot’s databases include the following steps:

• Understanding the client’s business, branding strategy and key-selling points.
• Identifying demographics, key profiles and cultural-mindset of their target audiences.
• Learn the parameters of the source language scripted questions and their preset answers.
• To further consider on the various permutations that the targeted (human) audience would likely phrase or word their question to obtain a specific answer.

So, how can we turn an existing Chatbot created to converse, say in English, to also communicate in other languages, for example Chinese?

Firstly, if we simply consider translating keywords taken from the database of English texts into Chinese language equivalents, it’s not going to work.

We need to consider that there are variations between written and dialogic connotations.

As an example, a simple question such as “How much?” in Chinese can be conveyed in several different ways, such as:

• “多少钱？” (“How much money?”)
• “费用是多少？” (“What’s the cost?”)
• “你会收多少钱？” (“How much do you charge?”)
• “价格是多少？ (What’s the price?)
• “我要付多少钱？(How much do I need to pay?)
• “多贵？” (How much is the charge?)
• “怎么收费？” (How is the charge?)
• “价钱怎么算？” (How is the price calculated?)
• “收费标准是什么？” (What’s the standard for charges?)

In this case, the system Developer must prepare the bot with common but nuanced ways that a Chinese national might ask.

Hence, it’s not advisable to try and populate the bot with Chinese translated meanings derived from bilingual dictionaries.

As a further example, phrasal questions such as “How much per room per night?” can be asked in the Chinese language in a variety of ways:

• “每晚多少钱？” (“How much per night?”)
• “客房每晚的费用多少？” (“How much is the cost per guest room per night?”)
• “一个晚上的房费多少？” (“What are the room charges per night?”)
• “一晚多少钱？” (How much for a night?)
• “住一晚多少钱？” (How much to stay for one night?)
• “住一晚要付多少钱？” (How much do I need to pay to stay for one night?)
• “房费怎么算？” (How are the room charges calculated?)
• “一晚房费是多少?” (What’s the room charges for one night?)
• “每晚房费是多少？” (What are the room charges for every night?)
• “房费多贵？” (How expensive are the room charges?)
• “客房/住宿怎么收费？” (How is the room/accommodation charged?)
• “房费标准是什么？” (What’s the standard for the room charges?)

Unfortunately, where some multilingual Chatbots have failed, it’s likely because it wasn’t ‘bred or built’ to anticipate the possible ways a person with a particular linguistic or cultural background might ask (input a query)… even for the same query.

Thus, Chatbots need to be continuously fed with the right or most appropriate keywords and key phrases typically used in everyday conversational lingo, across all age groups as well as economic social stratum.

The concerted efforts will progressively improve on its capabilities and provide respective target users the CX which is so vital in business development.