Linguistic patterns on many levels.

Photo of hill ranges fainter with distance.
Patterns on many levels

Linguistic research as pattern recognition.

The human brain is amazing at recognizing patterns. There are certain patterns that it is especially good at recognizing and faces are one of the visual patterns that we excel at recognizing. We’re so good at seeing faces we even see them where none exist, look at the ‘Man in the moon’ for an example of that.

Linguistic research of a previously unresearched language is all about pattern recognition at many different levels. From the sounds the language uses to the meanings it associated with them through to phrases that convey meanings apart from the meaning of the words used.  This short article will look at the different layers of patterns present in every spoken language.

Phonetic patterns

Often the first task for linguists is to document a language by recording spoken examples and transcribing them with the International Phonetic Alphabet. Once there is sufficient data linguists can look for patterns in the sounds. I.e. at the phonetic level. No language uses all the sounds that we are capable of uttering, there are too many for one, and some are too similar to be easily distinguished. The basic question that the linguist is asking at this level is which sounds do these speakers use and which are not used.

Phonemic patterns

The next question is which of the sounds that are used carry meaning? That is to say that usually there are sounds in every language which are suffciently similar so that the speakers don’t use them to differentiate between words.  In Mandarin, for example, no distinction is made between r and l so English speakers may hear ‘lice’ instead of ‘rice’. In English the p in pin and the p in spin are usually pronounced slightly differently from each other. There’s a little puff of air (called aspiration) after the p in pin that doesn’t occur usually after the p in spin. Pin and spin are pronounced as [pʰɪn] and [spɪn]. However if I were to say [spʰɪn] it would be recognised as ‘spin’ because the aspiration alone never marks a different meaning in English.

Linguists use a method called minimal pairs to help them discover which sounds carry meaning and which do not. If I find in English the two words [pɪn] and [pɪt] I can see that they differ only in one sound [t] and [n] so we can immediately say that both of those sounds carry meaning in English.  If [pʰɪn] and [pɪn] had different meanings then we would be say that [pʰ] and [p] both carry meaning. It’s much harder to arrive at the point where we can say that aspiration doesn’t carry meaning in English because we need a very large list of example words to prove that there are no two distinct words that only differ in whether or not there is aspiration.

Morphological patterns

At the level of the word we begin to enter the more commonly recognised patterns. The ones that we learn in school. This is where units of meaning can be brought together to create a new word. The study of the affixes is included at this level. If we take a few verbs see how they are used in different contexts we can easily see some patterns. The exceptions to the general patterns tend to occur on the most common words and that can make it harder to spot the general pattern.

verb present tense past tense third person
walk walking walked walks
run running ran runs
swim swimming swam swims
search searching searched searches

When patterns at this level are well documented by the linguist in FieldWorks, then the program can begin to analyse new words automatically using a choice of parsers.


At the word level there are lots of patterns, prefixes and suffixes that add or change the meaning of a word.
un- and -ing are examples. A simple word like ‘do’ can take many affixes such as -ing, un- or re-.  We string these together to form: ‘doing’, ‘undo’ and ‘redo’. We can add two affixes and create:  ‘undoing’ and ‘redoing’ too. Since the meaning is carried in the affix we understand the meaning of a word such as ‘re-undoing’, even if we’ve never seen it before.

Syntactical patterns

This is the thing that we might commonly think of as grammar. When we ask is this sentence grammatically correct, we’re wondering whether the words in the sentence are in the correct order. English syntax, in common with around 42% of the worlds languages uses the Subject-Verb-Object order. For example: I will run home. Yoda however, might say: Home run I will, presumably his first language uses the much rarer Object-Verb-Subject order.  There are often large numbers of patterns at this level because there are many ways to string together a sentence. Often different word orders may be understood but carry a different meaning.  Something I came across recently was the idea that there are many kinds of adjectives, and in English we can have more than one at a time to describe something. So we can say

‘Three curious little green men came from Mars’.

We’re saying something different and somewhat ambiguous if we say

‘Green little curious men came from Mars.’ or

‘Pointy green curious five men came from Mars.’

It’s because adjectives in English can be strung together but the pattern has them in this order:

  1. Quantity or number
  2. Quality or opinion
  3. Size
  4. Age
  5. Shape
  6. Color
  7. Proper adjective (often nationality, other place of origin, or material)
  8. Purpose or qualifier
Patterns when words don’t mean what they say

Idioms, hyperbole, metaphor, proverbs and irony are all types of phrases which carry different meanings than the words themselves. ‘A stitch in time saves nine.’ might refer to sewing stitches, but more often simply means that nipping a problem in the bud prevents it from getting worse and saves work. Of course that’s not talking about pruning either…

Irony is a great example of words not meaning what they say.  There’s the story of the linguistics professor telling the class that there are many languages where double negative is a positive. They continue to say that they don’t know of any langauge where a double positive is used as a negative. A voice pipes up from the back of the class : ‘Aye right!’                                        We need the ironic intonation to get the meaning…

Discourse patterns

The largest level of patterns in language occur at the level of a discourse.

Consider this construction. Once upon a time…. they all lived happily ever after. This pattern is easily recognizable though it is usually split many pages apart. It indicates that the bit in the middle is a fairy-tale or a children’s story. Even though there are no words in that phrase that mention fairy tales or children’s stories, meaning is conveyed by the pattern just as is the case for words and sentences.

There are many other patterns in discourse that communicate to us something more or other than the meaning of the words. I heard the story from a translation team that had completed a large amount of work and were doing final comprehension checks with speakers of the language. They read a translated passage of the Bible to those that have never heard it before. Then they asked many questions about the story to verify that it had communicated clearly. All the questions were answered correctly showing that the story had been well understood.  Then came the question ‘Was there anything unusual about the story?’. “Yes, we found it very odd. Why were there so many people called ‘Jesus’?”  Came the reply.

Further investigation revealed that a discourse pattern in this language is that proper names are only used to introduce a character the first time they are mentioned.  Every time Jesus was named the passage was introducing someone else, also called Jesus!  There followed much work to edit the passages before publication.

SayMore and other software can help with transcription and FieldWorks Language Explorer (FLEx) assists with the management and analysis of the data.

Very many thanks to Colin MacDonald for permission to use his photo!