Shall we start with systematic ear-training and articulation exercises?

No. Both are of doubtful value under the best of conditions. The majority of students manage eventually to understand and to make themselves understood without such adventitious and fanciful aids. Certainly. Unless the teaching rests on this foundation all the subsequent work will be distorted and false.
The young child does not have to undergo such processes when learning his native tongue, and yet he succeeds in hearing and in articulating correctly. The young child at the cradle age does little else than go through a course of such exercises. He listens and imitates, at first imperfectly, but later with great expertness, recognizing and reproducing isolated sounds and complex combinations of these.
Such exercises are extremely monotonous and dull; they are likely to kill interest and to cause the students to dislike the whole process of language-learning. Such exercises are always found extremely interesting, and tend to constitute an additional attraction to the study of the language.
Few language-teachers know how to make the foreign sounds correctly, and therefore few can give such exercises without causing the students to acquire bad habits. No teacher should be allowed to do language-work who is not proficient in the sounds of the foreign language, for those who are incapable of making the sounds cannot be good language-teachers.
It is useless to attempt to teach systematically the sounds of the language, seeing that these vary from one region to another and from one person to another. Any form of normal speech will serve as a model, provided that the speech is that of educated natives. In the absence of any model at all, the student will speak the foreign language with the sounds of his mother-tongue!

Shall we admit or reject the use of phonetic transcription?

Reject it certainly, for various reasons. Accept it certainly for various reasons.
It is extremely difficult; those who have been learning languages for years, even languages with strange alphabets, find phonetic symbols so puzzling that they are forced to discontinue their efforts. It is extremely easy; young children learn to use it readily and accurately. Those who experience any difficulty are those who are unable or unwilling to form new habits. A language is such a difficult thing that we must utilize every means of making our work easier.
It would take weeks or even months to learn the strange and unnatural symbols. The half a dozen strange symbols usually required in addition to those of the ordinary alphabet can usually be learnt at sight without any special practice. Even a strange ‘orthographic alphabet’ such as the Russian one can be mastered in a few days.
The whole proceeding is an unnatural one, contrary to all the laws of language. All writing is an unnatural process in the sense that it is not performed by instinct, but has to be learnt as an art. Of all systems of writing, however, the phonetic system is the one most in accordance with logic and natural law.
It is trying to the eyes. Most phonetic alphabets are clearer than those used in German and Russian, for instance.
It is a waste of valuable time to learn an artificial alphabet. The learning of a perfectly natural alphabet is in itself of educative value; it inculcates the idea of phonetic writing and serves once for all as an essential preparation for the study of any number of foreign languages.
It is evident that the use of a phonetic alphabet will make havoc of the ordinary spelling to be learnt subsequently. It has been ascertained experimentally that those who have been taught to read and to write a language phonetically become quite as efficient spellers as those not so trained. In many cases the phonetically trained student becomes the better speller.
To learn phonetic writing means learning two languages instead of one. In all cases where the traditional orthography is not in agreement with the native pronunciation the student is necessarily forced to learn the two things. The use of a phonetic alphabet is the only way to perform this double work rapidly, rationally, and with the minimum of confusion.
Phonetic texts always give slovenly and incorrect manners of pronouncing words. Authors of phonetic texts always strive to give an accurate rendering of the language as really and effectively spoken by educated natives; they rarely attempt to teach forms that have no existence in the language as actually used in ordinary speech.

Should we teach intonation in the early stages?

No. It is a fancy subject of little or no importance and certainly forms no integral part of language-study. Yes. It is a subject of great importance and forms an integral part of language-study. In many languages speech without the correct tones is only half intelligible; in Chinese and other languages it is perfectly unintelligible.
In any case it can be left to the very final stage of the programme. If it is not taught in the very earliest stage correct intonation will be very difficult to acquire. Language-study is a habit-forming process, and the habit of speaking with wrong tones is a bad habit.

Word or sentence first?

The word is the unit of language. Whatever the unit of language is, it is not the word.
Words are definite entities and constitute the component parts of sentences. Sentences may be reduced to component parts; sometimes these are words, but quite as often they are word-groups (such as compounds and phrases) or units less than words (such as affixes).
The word, not the sentence, is the basis of translation. Since a word has a definite meaning and conveys a definite idea it is easy to find the foreign equivalent. A sentence has generally, if not always, a definite foreign equivalent. A word is so unstable that it may entirely change its meaning when used with other words.
It is easy to memorize words and difficult to memorize sentences. It is as easy to memorize a six-word sentence as six words.
We speak in words. We express our thoughts in sentences.
If we learn a few dozen words we can build up thousands of sentences from these by the synthetic process. If we learn a few dozen sentences we can construct thousands of others from these by disintegration and substitution, and, what is more, we can recognize them and use them even in rapid speech.
Words are the basis of grammar. Sentences are the basis of syntax.
The collection of word-families is a valuable way of enriching one’s vocabulary. The enriching of one’s vocabulary should be left to a comparatively late stage in the study of language, especially in the study of most derivatives and compounds.
Words constitute the ‘primary matter’ (i.e. matter to be memorized integrally without analysis or synthesis). Sentences constitute the ‘secondary matter’ (i.e. matter to be derived synthetically from primary matter). It is precisely because sentences are so rarely considered as ‘memorized matter’ that so few people manage to understand the foreign language when spoken or to express themselves correctly in it.
Take care of the words and the sentences will take care of themselves. Take care of the sentences and the words will take care of themselves.

Should irregularities be included or excluded during the earlier stages?

The regular is easy, the irregular is difficult; in the interest of gradation let us therefore exclude temporarily the irregular. Irregular forms are generally more used and more useful than regular ones; in the interest of gradation let us therefore include all necessary irregularities even in the earlier stages.
Irregular forms make it difficult to formulate precise rules. Rules with numerous exceptions are not worth formulating at all.
The normal and logical should precede the abnormal and illogical. Then, as natural languages are full of abnormalities and bad logic, let the student start with an artificial language!