Saturday, January 18, 2020

Meta-Linguistic Awareness and Syntactic Fieldwork

Meta-linguistic awareness is an important concept in linguistic fieldwork, but as far as I know it has never before been discussed in that connection. In this blog post, I will define the concept, give several examples and explain why it is important for fieldwork.


Meta-linguistic awareness is the conscious awareness of language as an object with its own properties that can be discussed and investigated.

From this perspective, language is no different from a car that can be taken apart and understood. People understand that a car is an object with parts, and although they may not understand the way the parts work, they know it can be taken apart. And that, in principle, they could study those parts and learn how to fix cars.

For example, one aspect of meta-linguistic awareness is the conscious awareness of what a word sounds like. Can the consultant tell you what other words sound similar (e.g, in pitch)? To answer that question, the consultant needs to go beyond simply using words, and be able to examine the properties of words. They need to think of words as objects that can be compared based on their properties, just like they might compare other kinds of objects. This requires a conceptual shift.

Why not just call it linguistic awareness, instead of meta-linguistic awareness?

Barring extreme circumstances, people have an excellent command of their language. They can produce an unlimited number of phrases and sentences, and they know what each of these sentences mean (as shown by the fact that they can understand them and act accordingly). Generally, these sentences are well-formed according to the grammar of their language (putting aside speech production errors and some other minor cases). They can express their thoughts and desires easily. They also understand the sea of language that they are submersed in without any difficulty at all.

But crucially, despite this kind of knowledge, they are not consciously aware of language as an object with its own properties, other than very simple and obvious properties, such as the fact they have a language and that it is different from other languages.

So we can draw a distinction between linguistic knowledge and meta-linguistic knowledge:

Linguistic Knowledge: The grammatical system that a person uses in speaking and understanding natural language, etc.

In Principles and Parameters syntax, this is referred to as competence or I-language.

Meta-Linguistic Knowledge: Conscious knowledge of language (the grammatical system) and its properties (e.g., properties of individual words, phrases, sentences, the whole system, etc.).

Although I refer to Principles and Parameters syntax above, the distinction between linguistic knowledge and meta-linguistic knowledge should hold for any syntactic framework (e.g., descriptive, formal, functional, typological).

When I use the term “meta-linguistic awareness”, I mean this as a prerequisite to having meta-linguistic knowledge. One first has to recognize the grammatical system as an object with its own properties, before one can have knowledge of it.

Across different cultures, people do not have high levels of awareness of their language as an object with its own properties. That is, they do not have high levels of meta-linguistic awareness. Rather, they use language as a tool, just as they use their eyes as tools. And just as people generally do not know what is going on inside their eye, and generally do not care (until something goes wrong), they also do not know what is going on with their language, and do not care.

In fact, for day-to-day language use, one does not really have to rely on any meta-linguistic knowledge at all. Just as for day-to-day kidney use one does not even have to know that the kidneys exist.

The ordinary, day-to-day use of language does not require meta-linguistic awareness.

For example, one aspect of meta-linguistic awareness would be to recognize that English has a stress system, so that one syllable in each word receives special prominence (e.g., MissiSSIppi, with the stressed syllable indicated in capital letters). In North America, most of us have finished high school, and have even gone to college. And so most of us have used dictionaries where stress is marked, and are aware (to some degree) of the fact that English words bear stress. But in other cultures, especially those where people are unable to read and write, they are definitely not consciously aware of similar facts, like the fact that they speak a tone language. And so they are not consciously aware of the fact that there are tonal minimal pairs. Or the fact that there are a fixed number of tonal contours (e.g., HH, MM, LL, HM, HL, LM for Kua). Or the existence of tonal sandhi, or the factors that trigger it. They are simply blissfully unaware of all of this rich detail in their language. They are even unaware that there is something called “tone”.

However, they might know that a different dialect with a slightly different tonal system “sounds different”. That is definitely a limited kind of meta-linguistic awareness, but they will generally not be able to tell you what constitutes the difference.

Formal Education

As alluded to in the aforementioned example of awareness of word stress, a typical North American educated person already has a high level of meta-linguistic awareness just by having successfully completed high school. So formal education (K-12, reading, writing and arithmetic) teaches students to look at their language as an object:

Formal education correlates meta-linguistic awareness.

Here are some of the things that a typical high school student will consciously know as part of their meta-linguistic knowledge:

English has regular spelling conventions where letters correspond to sounds. English also has irregular spelling conventions. For example “gh” can sound like “f” in “enough”.
Some letters are vowels and some are consonants.
[Although students may not know what the definition of a vowel or a consonant is, they can at least list the vowels as a-e-i-o-u.]
Words can be broken up into syllables, some of which bear stress.
Some words rhyme and some do not.
Some forms are prescriptively prohibited (e.g., splitting infinitives, stranding prepositions).
[Although they may be confused about the relation between linguistic competence and prescriptive rules.]
Words can be categorized as noun, verb, adjective, adverb and preposition.
[Although they might have a hard time giving definitions and identifying these categories in tricky cases.]

In addition to these topics, North American children have lots of hands-on experience working with language in the classroom. When they are younger they have workbook and classroom activities designed to help them learn to read and write. In high school, students spend time learning about writing essays, and learning how to take standardized tests which have questions concerning language. Students also regularly study foreign languages in middle school and high school. All of these activities have the result of increasing the awareness of language as an object with its own properties that can be studied. Meta-linguistic awareness is not all-or-nothing, but rather is incremental. Gaining meta-linguistic awareness takes practice and years of work.

I am not claiming that all formal education on language is well-devised or useful. In introductory linguistics courses we spend a lot of time undoing misconceptions. But the fact that students have been taught to look at language as an object of study is important.


In working with consultants who speak the languages that I investigate (very highly endangered Khoisan languages), the issue of meta-linguistic awareness becomes important. Almost all of the speakers are elderly (60 and over) and are unable to read or write in any language. The practical consequence is that they cannot perform all the tasks that I would like them to do, or they require extensive training to be able to perform some tasks. Some useful tasks they are never able to perform at all.

A clear example of meta-linguistic awareness is given by the tone matching task. In this task, the consultant is played a word, and then they are asked “What is it the same as?”. In response, they give one of a small number of words whose tonal contours (e.g., HH, MM, LL, HM, HL, LM) are known. In this way, the consultant identifies the exact tonal contour of a new word. Anybody who has experience in working with tone languages will recognize that tone matching would be a valuable skill for a consultant to have. If the consultant has such a skill, it entails that the linguist will be able to transcribe tone 100% accurately.

The tone matching task requires a very high level of meta-linguistic awareness. The consultant must know that there are a small number of tonal patterns. The consultant must also know that any new word will have its own tonal pattern, and as a consequence will fall into one of those patterns. And they must be able to pronounce two words, and make a judgment about whether they are the same or not in their tonal pattern. All this requires explicit awareness of language as an object.

The knowledge needed to do the tone matching task certainly requires linguistic competence. Only a native speaker will be able to do it. But in addition, it requires conscious awareness of how the grammatical system is structured. In over 23 years of work on Khoisan languages (starting in 1996), I have only had one consultant who is able to do this task. Not surprisingly, this consultant is also my only consultant to have basic literacy in Setswana. They learned how to read and write Setswana while working in the mines in SA, but they do not have any knowledge of English reading or writing.

Other apparently simpler tasks also require high levels of meta-linguistic awareness. Take for example the task of giving a grammaticality judgment. For a given sentence S, the subject must say whether or not they find it grammatically acceptable. But to do this they need to be consciously aware that sentences are objects that can be classified as grammatical or not. Also, they need to have the surprising and counter-intuitive knowledge that one can discuss ungrammatical sentences, which are not sentences that they would ever actually use or hear.

In my experience, I have found many people who are unable to give grammaticality judgments. Usually in this case they will assent to any sentence presented to them, in the hopes of pleasing the linguist.

See the following blog post for further discussion of grammaticality judgments:

An important point about grammaticality judgments for this post is that even if a consultant is unable to give meaningful grammaticality judgments, they still are competent in their language. The sentences they produce when speaking are all perfectly grammatical, and they can do other tasks (such as the translation task) that clearly reveal their linguistic competence.

Lack of ability to do a task does not always indicate lack of linguistic competence.

Another useful skill for linguistics consultants is to be able to repeat back a line of text. From experience, I know that this can be quite challenging for consultants. Some older speakers cannot do it at all, and younger speakers (in my case between 60-70) need a lot of practice to become good at it. I believe this is also an issue of meta-linguistic awareness. In no other context have they ever had to repeat a sentence back exactly. At most, they will have to clarify what they have said, but that does not require them to repeat the exact same sentence. Rather repeating a sentence back exactly requires recognizing a particular sentence (not just its meaning) as an object.

A conflating factor in this case is that, as pointed out to me by Veena Dwivedi, "Older adults are better at gist memory for language vs. word for word memory." Still I would predict that controlling for age there would still be an effect of meta-linguistic awareness.

Yet another useful skill is for the consultant to be able to come up with an example sentence. In doing lexical elicitation, it is often useful to know what kinds of context a word can occur in (especially for verb). You can ask the consultant: “Please give an example of a sentence containing the word.” In my experience, consultants have no idea at all what this question means, since they are usually unfamiliar with the word “sentence” (in Setswana in my case). And even with some examples to illustrate what is wanted they find the task difficult. Only after many attempts do they become comfortable with the task.

Consultants can be trained in tasks requiring meta-linguistic awareness.

An objection to this line of thought would be the following: Consultants do in fact have meta-linguistic awareness of language and its components, but do not have any practice with the relevant tasks. So the issue is lack of practice with the tasks, not meta-linguistic awareness. My response is that it is difficult to separate the issues of lack of practice and lack of meta-linguistic awareness. Consultants gain meta-linguistic awareness only by engaging in the kind of activities that require them to recognize language as an object of study. In day-to-day life, they do not encounter such activities, since they use language as a tool for speaking, understanding and thinking. And such uses do not require meta-linguistic knowledge of language, just exactly like use of the kidneys requires no knowledge of them.

Furthermore, my prediction is that consultants with some meta-linguistic awareness will learn unfamiliar tasks more easily than subjects with very little meta-linguistic awareness (as with my example of the tone matching task).


What concrete consequences should this post have for syntactic fieldwork?

For one thing, it helps the linguist understand what is going on and why the consultants have the problems that they do with the task demands. More specifically, it helps the linguist factor variation in task performance into those factors related to linguistic competence (the grammatical system) and to other non-competence related factors (in this case meta-linguistic awareness). Just because a consultant cannot give consistent grammaticality judgments, that does not mean that they have a different grammar. It might also be that they are having problems with the task in part due to lack of meta-linguistic awareness.

Furthermore, this post should help the linguist to think about the kinds of skills that they require that consultants have, and the kind of training needed to acquire those skills. Even very basic skills, which the linguist might consider to be easy and obvious, might require extensive training.

On basic skills for syntactic fieldwork, see the following posts:

It should be clear that the notion of meta-linguistic awareness, and its consequences for syntactic fieldwork are not confined to work in a particular syntactic framework. In any framework (descriptive, formal, functional, typological, etc.), if one asks a consultant for information about their language, one will run up against the same set of issues.


  1. This comment has been removed by the author.

  2. Veena Dwivedi:

    What an interesting post. I have some comments on what you say, and I summarize these below. First, I agree with your basic idea, which says that metalinguistic awareness is a skill that has to be acquired. This is in contrast to ‘acquiring’ language, which happens without instruction and for which the brain is ready at birth.

    It is for this reason that EEG and psycholinguistic studies try to uncover what happens in language comprehension in real-time—presumably in order to capture how the unconscious rules of grammar might be used in this process. Although it is the case that several studies have acceptability judgment tasks after participants have read sentences, the authors of those studies would have to acknowledge that the addition of that task taps into a different underlying mechanism in the brain, consistent with your thesis.

    Next, I’d like to point out that it is best not to conflate rules of reading with language. Reading is another acquired skill. It allows for language to be viewed as an object—depending on the writing system, it will separate out words and sentences. However, I would look at building reading/writing skills as an example of an athlete cross-training at the gym. Many times, in order to achieve an athletic goal, other muscles must be developed (too bad I’m not an athlete, otherwise I’d have specific examples, but you get the idea). Thus, building on reading/writing allows for this other goal, building metalinguistic awareness to grow. But again, these are independent cognitive and neural systems.

    Literacy comes into play at two levels when it comes to metalinguistic awareness. That is, there is individual literacy, and there are cultures that value literacy, in the sense that they have developed writing systems. You can have an illiterate individual in a culture that does not have a developed writing system, and you can have an illiterate individual in a culture that does. Having done fieldwork on Hindi, it was always my experience that even individuals who might not have been very good at thinking about language, nevertheless always understood the notion of acceptability. This was due to the fact that all my consultants were educated and familiar with grammatical rules of Hindi and occasionally, Sanskrit. However, even illiterate individuals in India would know of stories from the Hindu epics (Ramayan and Mahabharat) and these tend to be recited by local Brahmins. In other words, even if they are illiterate, they would be exposed to language as art, as form. Something else to think about, for sure.

    Next, even when literacy is controlled for (both re: individual and culture) there are individual differences regarding metalinguistic awareness for language. Certainly, in my experience, linguistic undergrads in my lab have the ability to think about language, as do students who are not ling majors but are multi-lingual. This is in contrast to my bright Neuroscience students, who have a harder time thinking about language. Of course, one could say that this is the result of training in classes; however, I would have to add that, there is of course self-selection going on in the sample: individuals interested in language tend to apply to ling programs (and at Brocku, it’s an applied program, so most classes are not even LING classes).

    Finally, I see you have incorporated my comment regarding aging and the brain: we know that older adults in North America, are better at memory for the gist of language, not word for word, or syntactic structure. Thus, the effect you describe for your 60+ years old consultants could be a factor of their age/brain development vs. metalinguistic awareness.

    Ultimately, I completely agree with what you are saying: we want to separate out the underlying algorithms associated with language, as these are developed unconsciously during development without any instruction. Metalinguistic awareness, on the other hand, is a skill that has to be developed and this is best done or at least is easier when the culture and the individual is literate.


Note: Only a member of this blog may post a comment.