III.
a. To unwritten Languages.
After the explanations contained in the first and second parts, there is little more to be said on this point.
The missionary who attempts to write down for the first time a spoken language, should have a thorough knowledge of the physiological alphabet, and have practised it beforehand on his own language or on other dialects the pronunciation of which he knows.
He should put from recollection, as much as possible, the historical orthography of German, English, French, or whatever his language may be, and accustom himself to write down every spoken sound under the nearest physiological category to which it seems to belong. He should first of all endeavour to recognise the principal sounds, guttural, dental, and labial, in the language he desires to dissect and to delineate; and where doubtful whether he hears a simple or a modified secondary sound, such as have been described in our alphabet, he should always incline to the simple as the more original and general.
He should never be guided by etymological impressions. This is a great temptation, but it should be resisted. If we had to write the French word for knee, we should feel inclined, knowing that it sounds ginokyo in Italian and genu in Latin, to write it gěnu. But in French the initial palatal sound is no longer produced by contact, but by a sibilant flatus, and we should therefore have to write zěnu, If we had to write down the English sound of knee, we should probably, for the same reason, be willing to persuade ourselves that we still perceived, in the pronunciation of the n the former presence of the initial k. Still no one but an etymologist could detect it, and its sound should be represented in the Missionary alphabet by "ni."
Those who know the difficulty of determining the spelling of words according to their etymology, even in French or English, although we can follow the history of these languages for centuries, and although the most eminent grammarians have been engaged in analysing their structure, will feel how essential it is, in a first attempt to fix a spoken language, that the writer should not be swayed by any hasty etymo-