User:Dghmonwiskos/reconstruction accuracy: h₃nóh₁mn̥
This is an attempt at a methodical approach to reconstruction of PIE lemmas. I decided to use *h₃nóh₁mn̥ as its reconstruction remains contentious, and because it has a large amount of descendants. The methodology I have decided to use is representing the "likelihood" of each reconstruction being the correct one as a percentage. This was done by analysing individual phonemes and working out what percentage of the total pool of descendants show (or allow for, if the phoneme regularly disappears) each given phoneme. In some cases it is not possible to analyse phonemes individually: in this lemma, for example, we normally cannot reconstruct 2 full grades (except for the ending *-mō); and thus the initial vowel will be analysed on its own, but it will be assumed that if a language shows *ó or *é, it will not have the ending *-mén or *-mḗn. Percentages are worked out for all possible reconstructions based upon the phonemes/sequences analysed, and a list of reconstructions by accuracy is produced. It is worth noting some possible flaws with this method:
- It is still subject to personal ideas on a reconstruction; for example, with this lemma I have opted for h₃ as the initial laryngeal (others may differ, affecting the outcome of the analysis);
- It is only as accurate as the number of daughter languages in which a reflex of a given lemma is present;
- This method ignores internal reconstruction and only works well for nominative forms (as they are generally preserved well). Sound changes that may have occurred by analogy with oblique forms are not explained, and it is left up to you to explain the results obtained.
- Proto-Albanian: *enmen < *h₃n̥h₁mén
- Proto-Anatolian: *lṓʔmn̥ < *h₃lóh₁mn̥
- Old Armenian: անուն (anun), *անումն (*anumn) < *h₃nóh₁mn̥ or *h₃nómn̥
- Proto-Balto-Slavic: *inˀmen (see there for further descendants) *h₃n̥h₁mḗn
- Proto-Celtic: *anman (see there for further descendants) < *h₃n̥h₁mén
- Proto-Germanic: *namô (see there for further descendants) (< collective) < *h₃nómō
- Proto-Hellenic: *ónomə < *h₃nómn̥ (or maybe *h₃ń̥h₁mn̥)
- Proto-Indo-Iranian: *Hnā́ma (see there for further descendants) < *h₃nóh₁mn̥ or *h₃nómn̥ or *h₃néh₁mn̥
- Proto-Italic: *nōmən < *h₃nóh₁mn̥
- Phrygian: ονομαν (onoman) (accusative singular) < *h₃nómn̥ (or maybe *h₃ń̥h₁mn̥)
- Proto-Tocharian: *ñemä < *h₃néh₁mn̥
Individual phonemes (or sequences if sound laws and/or ablaut do not allow for analysis of individual phonemes):Edit
- *h₃ - 100% chance
- *n - 90.9% chance; 9.1% chance it is *l
- *ó - 63.6% chance of *o; 27.3% chance of *Ø; 9.1% chance of *e; almost all languages that preserve traces of the pitch accent show initial stress.
- *h₁ - 63.6% chance of *H; 36.4% chance of *Ø; the fact that in one daughter language, *e survives, implies that the laryngeal must be *h₁
- *mn̥ - 63.6% chance of this sequence; 18.2% chance of *mén; 9.1% chance of *mḗn; 9.1% chance of *mō
All possible reconstructions in order of likelihoodEdit
Analysis of this table implies that both the current wiktionary reconstruction for *h₁nómn̥ and my own (the two "most accurate" reconstructions) are largely inaccurate. However, the table can be restructured so as to discount many unlikely reconstructions.
- All forms with *l are based solely on Anatolian evidence, and can be explained by dissimilation with the following *m, presumably occurring only in that language family. To eliminate the *l forms, we can add the percentages of each form to an *n form with otherwise exact composition. Thus,
|Old reconstructions||Respective accuracy||New reconstruction||Accuracy|
|*h₃nóh₁mn̥, *h₃lóh₁mn̥||23.4%, 2.34%||*h₃nóh₁mn̥||25.7%|
- Given that I reconstruct a laryngeal, I can try to provide an explanation for the fact that about a 3rd of the daughter languages show no trace of one. I have explained this on another User:Dghmonwiskos/*h₃nóh₁mn̥ and so I can similarly merge forms with and without a laryngeal. I will thus represent the laryngeal in brackets.
- Finally, the reconstructions that differ from *-mn̥ to *-mō may also be merged, as the latter can be explained as a collective form of the former which presumably later replaced it in some daughter languages.
The final table is thus:
The other forms can be further explained as generalisations based on earlier ablauting forms. Hence