Skip to content

Linguistics and conversation design

Wood Scrabble tiles spelling out the words choose your words

Most of us use chatbots every day, whether it’s helping us navigate government services or buy an item from our favourite online store. Sometimes they’ve helped us reach our end goal, other times not so much.

For a chatbot to be effective it needs to be able to understand and respond to human language in a natural way. This is where linguistics comes in.

Study of human language

Linguistics is the study of human language, and it can be used to help chatbots understand the nuances of human communication. For example, chatbots need to be able to understand the difference between literal and figurative language, as well as the different ways that people can ask the same thing. They also need to be able to understand the context of a conversation, to give appropriate responses. I’ve worked with chatbots where there were over 80 different utterances from users all wanting the same one piece of information. Due to users’ varied backgrounds they all have different ways of expressing the same need.

Natural language processing

There are several different ways that linguistics can be used to improve chatbot development. One way is to use natural language processing (NLP) techniques to train chatbots to understand human language. NLP is a field of computer science that deals with the interaction between computers and human (natural) languages. It can be used to extract meaning from text, and to generate text that is similar to human-written text.

Rule-based systems

Another way to use linguistics to improve chatbot development is to use rule-based systems. Rule-based systems are a type of artificial intelligence that uses a set of rules to determine how to respond to a given input. These rules can be based on linguistic knowledge, such as the rules of grammar and syntax.

Machine learning

Finally, linguistics can also be used to improve chatbot development by using machine learning techniques. Machine learning is a type of artificial intelligence that allows computers to learn without being explicitly programmed. This can be used to train chatbots to recognise patterns in human language, and to respond to these patterns in a natural way.

By using linguistics, chatbot developers can create chatbots that are more natural and engaging in their interactions with users. This can lead to improved customer satisfaction and increased sales.

Using linguistics to improve chatbot development

Following are examples on how you can use linguistics to improve your chatbot.

Understand the difference between literal and figurative language

Chatbots need to be able to understand the difference between literal and figurative language. For example, if a user says, “I’m feeling blue,” the chatbot should not respond by asking the user what colour they’re feeling. Instead, the chatbot should understand that the user is using figurative language to express sadness.

Understand the different ways that people can express the same thing

Chatbots need to be able to understand the different ways that people can express the same thing. For example, if a user says, “I need help,” the chatbot should be able to understand that this means the same thing as “Can you help me?”. The chatbot should also know the synonyms for help, such as assist, support, aid, a hand, guidance.

Understand the context of a conversation

Chatbots need to be able to understand the context of a conversation to give appropriate responses. For example, if a user asks, “What time is it?” the chatbot should be able to understand that this is a question about the current time, and not a question about the time it takes to get to a certain destination.

By using linguistics, chatbot developers can create chatbots that are more natural and engaging in their interactions with users. This can lead to improved customer satisfaction and increased sales.

Other techniques to help improve your chatbot

  • Use a variety of linguistic resources, such as dictionaries, thesauruses, grammar books.
  • Train your chatbot on a large corpus of text, so that it can learn the patterns of human language.
  • Use machine learning techniques to improve your chatbot’s ability to recognise patterns and respond to them in a natural way.
  • Test your chatbot with a variety of users, so that you can get feedback on its performance.

Many chatbot owners get excited about the development of their chatbot and what it’s going to look like on the website, often at the expense of content. You can have a great looking chatbot with the latest software but if your chatbot doesn’t know how to interact with your users then your chatbot won’t perform well. Conversation design should be included at the beginning of your chatbot project. A good conversation designer will be able to assess your user needs and create conversation flows that exceed user expectations.  

Helpful links