Dear friends,

I am writing to ask for ideas on how to measure morphological "sensitivity" amongst 10 year old children (monolingual Arabic speakers). Arabic is a Semetic lanaguge with a highly "dense" morphology and the ability of reading and spelling, I suspect, will depend on the ability of Arabic speakers to "unpack" single words into their constitutent morphemes.

So far, I could think of 4 different tests to measure such sensitivity but I am not sure if these measure only morphological sensitivity or if any other skills (such as phonoligical or syntactic) components are also included.

I would be very grateful for your faddback and insight on how to go about measuring such ability and/or skill.

Very best, Gad

Thank you, Gad, for such a thought provoking question.

I find it difficult to move from teaching the morphology of a foreign language to people who have some grasp of English morphology to understanding how one might measure such understanding as a key to coping with dyslexia.

It might help if youy outlined the four tests that you had already thought about - it would certainly give me somewhere to start.

- Garzo.

Thanks Garzo, Of course. Here are the four tests I could think of. Please, feel free to give me your feedback in case I am confusing somthing:

1. Morphological Segmentation Test:
A list of words (25 in my test) that have two morphemes such as /maktab/ "desk in Arabic", where participants are asked to insert a slanted line where they think the word can be devided into two units (morphemes). In the above example, if the slanted line is inserted after the initial /m/, the word becomes /katab/ which is "to write: an infinitive" or /kutub/ "books" "short vowels are added here only to make it clearer for forum readers. Short vowels are not written as part of the Arabic script, but are diacritical marks which are largely absent in the every day written Arabic.    

2. Morphological Discrimination Test:
25 pairs of words. Some of the pairs are derivationally related words (cow - cows) while other are spelling related only (ear - earth). Participants are required to only put a tick next to pairs which are derivationally related.

3. Sentence Completion Test:
10 sentences. Each sentence goes like:
The man who plays is a ..............
Participants are asked to provide a word which is derived from the verb "play" to complete the sentence (player), and so on.

4. Sentence Completion Test using Pseudowords:
Same as above, but contextual clues are taken out by using pseduwords as verbs.

Gad

Dear Gad!

Test 2 seems the most straightforward and useful to me. I’m sure it also distinguishes between different forms of dyslexia. Test 3 might come across as rather contrived to the ten-year olds - I could see them finding it far too easy to go off task - "The man who plays is a footballer/a musician/silly". Test 1 is a classic look at Arabic morphology - it is the kind of approach I would use to teach adults Arabic as a foreign language - the morphology may seem quite tenuous to a monolingual ten-year old. Test 4 might work best if combined with test 3, then it would show how a different degree of sensitivity. Whereas a student confronted with a list of pseudowords may simply find them impenetrable.

- Garzo.

Excellent, thanks Garzo. Yes, you are right, I think combining tests 3 & 4  (10 real words and 10 pseudowords) together might be a better idea than presenting them seperately.

The test you suspect is too easy for a 10 year old does actually sounds easy in English, but may not be so in Arabic for two reasons:

First, all tests are presented after participants are given two examples and after making sure they understand what is required from them. In this case, they are explicitly told that the word they need to come up with must be derived from the verb. The two examples will illustrate this I think.

Second, in Arabic, the pattern "Fa’il" (Active Participle) is used to indicate the person who does the verb and it is usually systematic. So, someone who knows is "Aa’rif", somone who plays is "La’ib", someone who drinks is "Sha’rib" and so on. So, it is in fact a valid point. However, I think you are right, perhaps more than 2 examples are needed here to make it clear for a 10 year old that only derived words are acceptable.

Cheers, Gad  

Hallo Gad,

I am just wondering if you need not test participants’ ability to distinguish morphological and semantic associations.  Here is an idea I think not too detached from your intended study.

Which is the Odd-Man-Out Test

Each question item consists of four or five words that share the same root and one that is semantically equivalent but from an irrelevant root.  For each item participants are asked to mark out the word "not related to other words" or "most distant from the rest" (here, careful wording by an Arabic expert is necessary).  If an item consists of following items, the expected odd man out is "madrasa," because only this is derived from the root d-r-s.
[list][*]maktub[*]kaatiba(t)[*]kutub[*]madrasa[*][s]katabta[/s][/list]

In terms of semantics (in its broadest sense), "maktub" may be chosen as the irrelevant, whereas "madrasa" is the odd man out in terms of morphology.  In this test design, choosing "maktub" or failing to choose "madrasa" is to be construed as the respondent’s cognitive preference of semantic association over morphological one.  If "madrasa" is chosen, the respondent can be regarded as capable of morphological judgment.

If your thesis is "the ability of [Arabic] reading and spelling, I suspect, will depend on the ability of Arabic speakers to "unpack" single words into their constitutent morphemes," this test would serve as the null-hypothesis, that is, there may be some dyslexic Arabic speaker who associates the spelt form of a word (signifiant) and its meaning (signifiee) by photogenic memory but unable to decipher the "visual image" (the sequence of letters) into smaller phonetic constituents.

Thank you and good luck,
Flam

Perhaps I have misunderstood something, but it is very possible that I would single out katabta as being the only verb in a tense-and-person form, the others being nouns or participles.

Thank you so much Flaminus. This is really helpful. It’s a great idea and I have already made a test of it based on your explanations. Things are looking promising.

Big thank you for all those who have taken the time to respond and I welcome any more feedback on this regard.

My next question is, how can one measure reading rate "fluency"?

Gad

Thanks Anders, yes you are right. I have taken the idea and I have come up with the following groups. Perhaps I would give only two groups here for space; but it should give you the idea. Please let me know if this looks Ok.

Group one:
1. Safar
2. Musafir
3. Rihlah
4. Mosafreen

Group two:
1. Bana
2. Binaa
3. Mabna
4. Bayt

Rihla and Bayt are sematincally Associated while the rest are morphologically associated in the above two examples. Participants are required to put a circle around the odd one out (semantically associated)

Gad  

I am puzzled by mosafreen. Do you mean what I would write musaafiriin (travellers, oblique case)?

If, then, the purpose is clearly presented to be picking the word which is (semantically but) not morphologically related to the others, I think it should work, with the following caveats:

There might be a risk that somebody in the first set would pick the only plural, musaafiriin, if that is the word.
If my interpretation is correct, there is only one verb, banaa (w/alif maksuurah) in the second set, and it seems that the odd one is the only one not ending in alif maksuurah, so even a non-Arab might get it right.

For the umptieth time, it is a pity that we don’t have Unicode…

What I can’t judge is, if it would work for a person who isn’t perfectly familiar with MSA/al-lu[U]gh[/U]atu-l-fusHa. I know absolutely nothing of regional languages.

Hallo Gad!

I am glad that you liked my ideas.  Please keep on posting concerning the progress of your research.

About the next question "reading rate fluency," I would recommend comprehension test based on reading out a text.  My suggestion includes the following items to measure.

1. time the participant needed to read out the text
2. instances of error the participant made while reading out
3. comprehension test of the text (why who did what to whom in what fashion)
4. control test of text comprehension on participants who are instructed to read it silently.

I think this is it, for now, at any rate.
Flaminius