In view of the illustrations which have been given, we repeat what was stated in the beginning of this paper, that Indian place-names are not proper names, that is unmeaning marks, but significant appellatives, each conveying a description of the locality to which it belongs. In those parts of the country where Indian languages are still spoken, the analysis of such names is comparatively easy. Chippewa, Cree, or (in another family) Sioux-Dakota geographical names may generally be translated with as little difficulty as other words or syntheses in the same languages. In New England, and especially in our part of New England, the case is different. We can hardly expect to ascertain the meaning of all the names which have come down to us from dead languages of aboriginal tribes. Some of the obstacles to accurate analysis have been pointed out. Nearly every geographical name has been mutilated or has suffered change. It would indeed be strange if Indian polysyntheses, with their frequent gutturals and nasals, adopted from unwritten languages and by those who were ignorant of their meanings, had been exempted from the phonetic change to which all language is subject, as a result of the universal disposition "to put more facile in the stead of more difficult sounds or combination of sounds, and to get rid altogether of what is unnecessary in the words we use."[89] What Professor Haldeman calls otosis, 'that error of the ear by which words are perverted to a more familiar form,'[90] has effected some curious transformations. Swatara,[91] the name of a stream in Pennsylvania, becomes 'Sweet Arrow;' the Potopaco of John Smith's map (p[oo]tuppâg, a bay or cove; Eliot,) on a bend of the Potomac, is naturalized as 'Port Tobacco.' Nama'auke, 'the place of fish' in East Windsor, passes through Namerack and Namalake to the modern 'May Luck.' Moskitu-auke, 'grass land,' in Scituate, R.I., gives the name of 'Mosquito Hawk' to the brook which crosses it.[92]
In Connecticut and Rhode Island special causes operated to corrupt and transform almost beyond possibility of recognition, many of the Indian place names. Five different dialects at least were spoken between Narragansett Bay and the Housatonic River, at the time of the first coming of the English. In early deeds and conveyances in the colonial and in local records, we find the same river, lake, tract of land or bound-mark named sometimes in the Muhhekan, sometimes in the Narragansett, or Niantic, or Nipmuck, or Connecticut valley, or Quinnipiac (Quiripee) dialect. The adopted name is often extra-limitary to the tribe by which it was given. Often, it is a mixture of, or a sort of compromise between, two dialects; half Muhhekan, half Narragansett or Nipmuck. In the form in which it comes to us, we can only guess from what language or languages it has been corrupted.
The analysis of those names even whose composition appears to be most obvious must be accepted as provisional merely. The recovery of a lost syllable or of a lost guttural or nasal, the correction of a false accent even, may give to the synthesis another and hitherto unsuspected meaning. It would be surprising if some of the translations which have been hazarded in this paper do not prove to be wide of their mark. Even English etymology is not reckoned among the exact sciences yet,—and in Algonkin, there is the additional disadvantage of having no Sanskrit verbs "to go," to fall back on as a last resort.
Recent manifestations of an increasing interest in Indian onomatology, or at least of awakened curiosity to discover the meanings of Indian names, may perhaps justify the writer in offering, at the close of this paper, a few suggestions, as to the method of analysis which appears most likely to give correct results, and as to the tests by which to judge of the probability that a supposed translation of any name is the true one.
1. The earliest recorded form of the name should be sought for, and every variation from it should be noted. These should be taken so far as possible from original manuscripts, not from printed copies.
2. Where the difference of forms is considerable, knowledge of the character and opportunities of the writer may sometimes determine the preference of one form to others, as probably the most accurate. A Massachusetts or Connecticut name written by John Eliot or Experience Mayhew—or by the famous interpreter, Thomas Stanton—may safely be assumed to represent the original combination of sounds more exactly than the form given it by some town-recorder, ignorant of the Indian language and who perhaps did not always write or spell his own correctly.
3. The name should be considered with some reference to the topographical features of the region to which it belongs. These may sometimes determine the true meaning when the analysis is doubtful, or may suggest the meaning which would otherwise have been unsuspected under the modern form.
4. Remembering that every letter or sound had its value,—if, in the analysis of a name, it becomes necessary to get rid of a troublesome consonant or vowel by assuming it to have been introduced 'for the sake of euphony,'—it is probable that the interpretation so arrived at is not the right one.
5. The components of every place-name—or to speak more generally, the elements of every Indian synthesis are significant roots, not mere fractions of words arbitrarily selected for new combinations. There has been no more prolific source of error in dealings with the etymology and the grammatical structure of the American languages than that one-sided view of the truth which was given by Duponceau[93] in the statement that "one or more syllables of each simple word are generally chosen and combined together, in one compound locution, often leaving out the harsh consonants for the sake of euphony,"—and repeated by Heckewelder,[94] when he wrote, that "in the Delaware and other American languages, parts or parcels of different words, sometimes a single sound or letter, are compounded together in an artificial manner so as to avoid the meeting of harsh or disagreeable sounds," &c. The "single sound or letter" the "one or more syllables," were chosen not as "part or parcel" of a word but because of their inherent significance. The Delaware "Pilape, a youth," is not—as Heckewelder and Duponceau represented it to be[95]—"formed from pilsit, chaste, innocent, and lenape, a man," but from pil- (Mass. pen-, Abn. pir-,) strange, novel, unused (and hence) pure,—and -anpe (Mass. -omp, Abn. anbé) a male, vir. It is true that the same roots are found in the two words pil-sit (a participle of the verb-adjective pil-esu, 'he is pure,') and len-anpe, 'common man:' but the statement that "one or more syllables" are taken from these words to form Pilape is inaccurate and misleading. It might with as much truth be said that the English word boyhood is formed from selected syllables of boy-ish and man-hood; or that purity 'compounds together in an artificial manner' fractions of purify and quality.