It is common knowledge that “ums” and “ahs” are the sign of an ineffective speaker. But is common knowledge always correct? In this case, perhaps not, especially in the context of conversational, rather than formal, speech. In this month’s column we take a look at a topic of research in the field of psycholinguistics.


Leading edge research by psycholinguists suggests that words and sounds generally considered inappropriate to speech (um, ah, like, you know, etc.) are in fact useful and facilitate effective communication in many instances. It should be noted that all experts do not accept these findings.

These words and sounds are given the labels discourse markers and disfluencies. The formal study of these words and sounds was started by Frieda Goldman-Eisler in London in the 1950s, when she developed instruments that could measure and count pauses and the like in human speech. A leading researcher today in this area is Dr. Jean Fox Tree at the University of Southern California, Santa Cruz. For further information, try doing an Internet search on “disfluencies, “discourse markers”, or “Jean Fox Tree”.


Within a normal conversation, between 6-10% of the words used are disfluencies and discourse markers, but in some situations they can make up almost a fifth. They are found in all other languages. The French say something that sounds like euh, and Hebrew speakers say ehhh. Serbs and Croats say ovay, and the Turks say mmmmm. The Japanese say eto (eh-to) and ano (ah-no), the Spanish este, and Mandarin speakers neige (NEH-guh) and jiege (JEH-guh). In Dutch and German you can say uh, um, mmm. And so on.


When using speech to convey our thoughts, we face two main challenges.

First, if we get off track, make a mistake, or briefly forget where we are, it’s very difficult to review, revise, and retry in real time. If we were to do so in silence, the attention of our listeners would be lost, and the conversation would leave us behind.

Second, the thoughts we try to convey are often multi-level concepts that are very difficult to accurately explain in the linear way spoken language comes out of our mouths.

It appears as if disfluencies and discourse markers help to address these two main challenges. They fill in dead space while we correct ourselves, and signal the listener that we are doing so. They also allow us to indicate subtle relationships between concepts.


Ums & uhs

These are indicators for the listener that there will be either a short pause (um), or long pause (uh). Studies show that the short pause indicator um actually increases the attention levels of listeners. In other words it’s telling them, “Pay attention, just a brief pause here.”

The long pause indicator uh (or ah), generally creates a negative impression in listeners – that the speaker doesn’t know what they’re talking about, is unprepared, and is perhaps dishonest.

Restarts & repetitions

Interestingly enough, restarts at the beginning of a sentence (“I went to, um, we went to the store,”) and repeats (“I gave her a gift, a gift, for Christmas,”) have no effect on comprehension – they don’t improve or worsen listener comprehension. They appear to act as a way for speakers to correct themselves, or remember what they’re saying without interrupting the flow of their speech; in other words, they help speakers overcome unavoidable breaks in fluency without losing their listeners.

Thiy versus thuh

There are two ways to pronounce the word “the”: “thee” or “thuh”, indicated as “thiy” and “thuh” by linguists. We think we say them one way or the other all the time, but in fact we subconsciously switch back and forth depending on the circumstances. “Thiy” is followed by a pause or suspension in speech (um, ah, etc.) 81% of the time, “thuh” only 7% of the time. So it appears as if we use “thiy” as a way to indicate to listeners that there will be a brief pause, but we’ll continue on, much like “um”.

Discourse Markers

The most common discourse markers are “like”, “you know”, “well”, “I mean”, “oh”, “so”, and “and”. They appear to be used for a variety of purposes, some of them quite interesting. They can:

  • Show how two adjacent utterances are related
  • Highlight concepts to hold in mind for future reference
  • Show what is being said is not what is exactly meant
  • Help listeners recover from speech errors
  • Help listeners organize and evaluate the speaker’s talk
  • Help conversationalists monitor each other’s understanding
  • Help coordinate conversationalists taking turns speaking


“Oh” is a marker used to indicate a subtle break in logic or topic, including the commencement of sarcasm or other forms of humor. For example, “I met his brother, and oh, he wasn’t exactly a saint.” The “oh” here separates a factual phrase (“I met his brother”) from a feeble attempt at sarcasm (“he wasn’t exactly a saint.”)

It can indicate that a segment is out of place and pertains to a previous concept. For example, “I love apples, Macintosh, golden delicious, gala. They’re so crisp and sweet. Oh, and granny smith.” The “oh” obviously indicates that granny smith belongs to the previous list of apples, and not to properties of the apples, “crisp and sweet”.

So, and

“So” and “and” can act as little markers that divide up long sentences and thoughts into more easily understood chunks. They are shown to actually improve comprehension in listeners. Because the speaker understands what he/she is saying, and knows what the entire thought is before it’s finished, he/she can subconsciously insert little markers along the way at logical points to indicate the starts and ends of sub-thoughts. You could think of them as spoken commas or hyphens.

I mean

“I mean” is an interesting discourse marker. It appears to forewarn listeners of an upcoming adjustment, and signals them to pay attention. If you were to say, “She was a bad mother,” listeners would interpret that as a factual statement without supporting evidence, and expect no more. But if you were to say, “I mean, she was a bad mother,” then the listeners would really prick their ears up because they would be expecting some juicy supporting facts.

You know

“You know” is similar to “I mean”, but on inspection is quite different. It invites listener inference. So the speaker is more or less saying, “This is my opinion, but I’d like to hear your opinion.”


The use of “like” as a discourse marker is extremely prevalent in our culture. It appears to be used for a variety of the purposes listed in the bullets above. Dr. Fox Tree says she will tackle the study of “like” once she receives tenure, as it will likely turn into her life work! Perhaps you can think about the ways in which you use the word “like”.

Personal thoughts

I have pondered this topic more, and come up with a few little theories of my own. I suspect that these disfluencies and discourse markers are also used as social mechanisms.

I think people can utilize them to indicate deference or humility. A total lack of disfluencies and discourse markers indicates extreme confidence (which is more or less what we’re aiming for by eliminating them from formal technical presentations), therefore the opposite, using them in abundance, indicates a lack of confidence, or humility. Think of the Matlock TV series and how he would win the juries over to his side, or get a key witness to let down his guard through the use of “Gosh”, “oh”, repeats, etc. He was creating an impression in the listener that he (Matlock) was humble, was subservient, and therefore was not a threat.

I think they can also be used as tribal markers. Teenagers like to create an almost tribe-like peer group, and membership in that tribe can be indicated by the overuse of one or more of these disfluencies and discourse markers, or in fact the creation of an entirely new one. A few years ago young people who considered themselves intellectuals identified this self-image by liberally using “no doubt” as a discourse marker. Similarly, around the same time members of the Paris Hilton tribe were saying, “That’s hot,” also for no apparent reason.

I suspect that general social trends cause the actual words being used as disfluencies and discourse markers to change over time. Most social trends (clothing, behavior, etc.) are considered to be class distinction mechanisms. In other words, what society thinks of as fashionable and higher class, is imitated by lower classes in an attempt to gain entry. The higher class therefore needs to continually change in an attempt to differentiate itself from the “others”. It would make sense that these same mechanisms are at play with disfluencies and discourse markers. So if the Paris Hilton social set was something a statistically significant proportion of society aspired to, we’d eventually all be using “That’s hot” instead of “like”!


Although there are many linguists who disagree with these theories, the concepts seem to have an element of truth to them. In an analytical sense they allow us to view the use of “um”, “ah”, and other such pervasive words in an understandable way.

A critical point to make, assuming you buy into these theories, is that disfluencies and discourse markers are mechanisms that seem to have evolved to help conversational speech, and do not necessarily apply to formal monologues. This means that if your presentation to management (for example) requires you to appear confident, knowledgeable and honest, then disfluencies and discourse markers should not be prominent. However, if your speech involves some informality, or interaction with your audience, then perhaps a certain level of disfluency and discourse marker use would be appropriate.


Share This Column