Do you click, tap, swipe … or type? Touch screens have been around for some time, but text input - and, therefore, the humble keyboard - remains at the core of how we work and interact with each other.
Tablets and smartphones have virtual keyboards and new messaging services such as SnapChat and WeChat rely on the written word, so many of us spend much of our day typing.
Productivity may still be about words, but there's now no reason why they must be typed out letter by letter on either a keyboard or a touch screen.
New developments in speech recognition and natural language processing are already revolutionising how business and commerce works behind the scenes, and it won't be long before we're all talking instead of typing.
For anyone who has ever suffered aching arms and shoulders from typing too much, the return of the art of dictation should be welcome relief.
Although there are small cloud-based speech-to-text features built into Android and Apple smartphones, tablets and desktop computers, these are slow, limited to single sentences, and don't allow any corrections or control of other functions.
The best off-the-shelf speech-to-text software around with a far wider remit is Nuance's Dragon Dictate. It's now in its fourth version, and much improved since I tried it two years ago.
As well as recognising more words more accurately, it's now far easier to go back and both edit and make the inevitable corrections without having to touch the keyboard.
The accompanying free Dragon Recorder app for smartphones isn't new, but where once it was used only as a remote microphone, it's now possible to use it to record a voice memo and have it transcribed into text by the desktop software.
For students, mobile workers and anyone conducting interviews, this is excellent news, though Dragon Dictate 4 is at its best when it's transcribing a voice it knows. A 10-minute training session - consisting of reading and getting used to speaking punctuation - at installation is crucial to its success.
Could software like this change the way we work? "How we talk differs from how we type," says Jonathan Whitmore, UK, Ireland and Middle East regional sales manager of Nuance Communications, which makes Dragon Dictate 4.
"A dictation device allows users to capture their thoughts naturally wherever they are, before having them transcribed accurately."
Whitmore thinks that this process better meets the needs of today's digital culture and digital businesses than typing.
"Given that society today has a culture of sharing - whether via email or social media platforms - dictation and speech recognition play a role in the effective capture and near-instantaneous sharing of an event, activity or even a document."
It could also make a big difference to any industry that charges by the hour or by tasks completed, although anything written using Dragon Dictate 4 still needs to be proofread thoroughly.
That's because, while errors are fewer than before, errant words can appear at any time. In my test, "toothache" was transcribed as "238".
Interestingly, it doesn't take long to get used to talking at a computer instead of typing, and by the end of a day spent with Dragon Dictate I had started dictating SMS messages on my phone. I was also beginning to wish I could tell the TV, the kettle and the lights to switch on or off.
"The most natural form of communication is talking, and so investment and development in this area will continue to grow," says Whitmore.
"It offers a common means of interfacing with multiple devices, from phones to televisions. A voice is unique to an individual so it is a secure way of identifying a person, and it is much easier to talk to a phone than trying to type messages."
But while speech is very natural, creating a responsive website or app that can understand and give intelligent responses is much more complex. Work is under way on understanding semantics, linguistics, the context of conversations, the way people search for information, and the relationship between data.
Some of that research is already bearing fruit as sentiment analysis, which is helping computers understand the meaning of phrases.
"When analysing recorded voice calls and emails, chat or social media, for some businesses, understanding the attitude or mood of the person speaking or writing might be a determining factor in how they evaluate or make a judgment, and then, consequently, how they behave," says Simon Richards, CEO of Fonetic USA, a linguistics company whose voice-recognition technology is now used on trading floors.
It works by analysing strings of speech in recorded conversations and, in the financial sector, alerts compliance teams to possibly fraudulent conversations on trading floors. Fonetic knows 79 languages and has a financial industry-specific lexicon that's taken five years to build.
As soon as a trader starts having an unusual conversation, or one that features unknown language, the alert goes out.
"Fraudulent behaviour is likely to be coded language or could be an act that is related to market abuse, such as insider dealing," Richards says. "The analysis tools can detect when something makes sense and when it is odd, and determine patterns and trends and send alerts to surveillance or compliance officers to take a look at something that is out of the ordinary."
The Fonetic solution uses direct phrase recognition, a major next step for the accuracy of natural language processing, and while it's been designed for data-rich places like banks and call centres, there's no reason why it couldn't be used in all kinds of devices.
"The technology analyses voice calls that have been recorded and can analyse any data source, email, chat, instant messaging and social media," Richards says. "If it's captured or recorded, we can analyse it."
Aside from worries about snooping, natural language processing technology like this is already making communication smart, too. Smart messaging tools for teams, such as Slack, HipChat, Convo, Peak and ChatGrape, are using the technology to kill off the bane of office workers' lives: internal email, the scheduling of meetings and hunting for files to attach.
Based on its finding that an office worker spends 1,500 hours per year searching for documents and looking up files, ChatGrape is smart software that auto-completes messages, attaching relevant files without any search required.
It's cloud-based, searchable and encrypted - a bit like email - but it will also soon feature Speech Act Detection on an internal phone system, which will have natural language detection at its core.
"Speech Act Detection tracks ongoing conversations and semi-automatically triggers workflows based on what is being discussed," says Leo Fasbender, COO and co-founder of the Vienna, Austria-based ChatGrape.
"This could be something along the lines of 'Hey Jamie, let's meet up tomorrow at noon to discuss the front-end issue', which would trigger a calendar event for both our calendars, check whether we have time, then set up the meeting," Fasbender says. "It would also correctly reference the issue we're discussing, thus eliminating communication-based misunderstandings."
The holy grail of natural language processing is conversational interfaces where speech isn't just a new way to control old technology, but becomes the central construct of how products and services are designed.
Natural language processing is about customising everything, everywhere, and for everyone. Let the conversation commence.