How many people speak a North Halmahera language? (and why we don’t know exactly)

After I published my last blog post, I was asked how so many languages, belonging to two language families, fit on the rather small island of Halmahera. I will answer this question in this post. A warning: The answer is more complex than one may think.

How many people speak a North Halmahera language?

Halmahera is not that big. 17,780 km² (6,860 sq mi) to be precise. For comparison: Connecticut is 14,357km² (5,543 sq mi) and Kuwait is 17,818 km² (6,880 sq mi). Halmahera is not densely populated. Around 500,000 people live there at the moment. So there is no danger of running into each other all the time. If you take a look at Halmahera on Google Earth, one thing will spring to your eye first: green. Halmahera is densely covered with rainforest. A look at a topographic map also shows that the center of the peninsula is quite mountainous. Therefore, most people live along the coasts and travelling to the next village used to require a boat. This explains the relatively sparse population. 100 years ago it was even thinner. Due to data collected by the colonial administration, we know quite precisely how many people lived on Halmahera in 1922, namely 102,340 people, only one fifth of today’s population.

But how many of these people speak a Northern Halmahera language? To find out, a trained linguist will check out … Wikipedia. In fact, the English Wikipedia is a good starting point to find out about speaker numbers. I’ll explain why later.

Wikipedia gives us the following speaker numbers for the Northern Halmahera languages, I have added numbers from the literature where they differ greatly from the Wikipedia numbers.

Wikipedia spuckt die folgenden Sprecherzahlen für die Nordhalmahera-Sprachen aus, ich habe noch Zahlen aus der Literatur ergänzt, wo sie stark von den Wikipediazahlen abweichen.

	Wikipedia	Literature
Galela	79,000 in 1990	35,000 (Ipol et al. 1989: v)
Loloda	17,000 in 1991
Tobelo	30,000 in 2000	15,000 (Holton 2003: 1)
Pagu	3,700 in 2000
Modole	2,000 in 1983
Tabaru	15,000 in 1991
Sahu	7,500 in 1987	12,000 (Visser & Voorhoeve 1987: 3)
Ternate	42,000 in 1981
Tidore	26,000 in 1981
West Makian	12,000 in 1977	< 8,000 (my own estimation)

Galela, Tobelo, Ternate and Tidore are by far the most spoken languages, but let’s be careful: some of the numbers are over 40 years old. Due to the decline in speakers, which I adressed in my last blog post, the numbers may have decreased rather than increased since then.

How do we get data on speaker numbers (and are is it even accurate)?

Why do we see these big discrepancies between the figures from Wikipedia and those from the literature, for example 79,000 vs 35,000 for Galela? First of all, let’s take a look at where Wikipedia actually gets the numbers from. If you click on the reference in the respective articles on the languages, you will almost always be led to Ethnologue (25th ed., 2022). Ethnologue is a language atlas whose goal is to list information about all the documented languages of the world. It is published by SIL Internation (formerly Summer Insitute of Linguistics). SIL is a Christian evangelical missionary organization based in the U.S. and has been criticized in the past. Nevertheless, the data from Ethnologue is often cited and generally accepted (taking into account the problems discussed below).

For Ethnologue, SIL collects data from the literature, official statistics (e.g., censuses), and self-collected data. How does such data come about? You may think that we can just ask a person which languages they speak and that’s it. Unfortunately, it is not that simple. In the following I will list some problems which commonly occur during the collection of speaker numbers.

(Note: I have searched relatively long in vain for a source that lists such reasons and provides comprehensive references. Unfortunately, I have not found one – if you know one, please send it to me! For this reason, the following points are mainly based on my personal experience).

Problems with estimations by linguists

Many speaker numbers found in the literature (e.g. in Ethnologue) are based on estimations by linguists (or ethnologists). A linguist who has studied a particular language estimates, based on personal experience, how many speakers there are. This is how it may happen: A linguist visits the speaker community for language documentation. She finds out in how many villages the language is spoken and may get access to official figures. Based on her experience, she knows that children no longer acquire the language and that such and such a percentage of the population are children and after subtraction of this percentage from the total population: bang, we have a speaker number. Of course, this estimation will be super rough and dependent on how well our linguist can assess the situation. I would say that most of the speaker numbers cited in the literature are based on such rough estimations.

Problems with figures collected through surveys

Then why not just do an official survey by the government and ask people, “What language do you speak?” Again, this sounds easier than it is (even if we can assume that such a survey is logistically possible), here is a list of problems (probably incomplete):

Participants do not understand the question

It is often not clear what is actually meant by “language”. Let’s take an example: Some of you may be speakers of a (so-called) dialect of your first language – does that count as a “language” in our fictitious survey? For example, would you list both the dialect and the standard language (instead of just the standard language (Kölsch and German, Sicilian and Italian, Cantonese and Mandarin)? Sometimes languages don’t even have a name that is commonly used within the speaker community. People may speak of “our language” or the language may be called by the name of the village in which they live. In such situations, it is unclear what counts as one language (I will write a separate post about this problem in the future).

Participant don’t know if they count as speakers

Okay, but then why not rephrase our question and ask “Do you speak language xy?”. This can also lead to problems. Again, an example: Would you state your language skills from French class in high school? Can one speak Latin? People can answer this question differently, even if they took the same French or Latin class.

There is one more issue, which can be summarized under the term ‘language ownership’: Who counts as a speaker is not clear regardless of how well a person speaks a language. For example, language ownership is inherited in some speaker communities. For example, children of a speaker of a certain language are automatically considered speakers as well, even if they prefer to speak another language. At the same time, they are not considered speakers of that language because they have not “inherited” it. The concept is not so strange to people in Western countries. For example, people with a migration background who grew up in a Western country are told that they “already speak the language well” (definitely a thing in Germany). Also in Western countries language ownership is often associated with belonging to an ethnic group. Many people speak several languages and are unsure how many of them they should list in the survey. In the literature, one often encounters the term ‘native speaker’. This usually refers to a person who has acquired a language from birth. This term is also not without problems (see this article).

People sometimes tend to overestimate or underestimate their language abilities for other reasons as well. Many people can communicate quite ok in a language, but still would not claim to know it (for example, because they cannot write it).  On the other hand, it can sometimes be practical to get the highest possible number of speakers in a survey, for example, if you would like to have schooling financed for your children or to be recognized by the state as a distinct ethnic group. In this situation, all members of an ethnic group may declare that they speak its language, regardless of how proficient they are. (And depending on the idea of language ownership this is not “wrong”).

Participants do not dare to answer honestly

Even in (supposedly) anonymous surveys, speakers sometimes have reservations about admitting that they speak a certain language. For example, languages may be socially rückstigmatized. In countries with one dominant official language and many small local languages, the latter are sometimes stigmatized as “backward”. In this case, speakers tend to conceal their language abilities in surveys, preferring to indicate the prestigious official language. Sometimes speaking a language may be a threat to your life, for example if it reveals that you belong to a opressed or persecuted ethnic group.

So much for now about speaker numbers. I could continue this list of problematic points. But I think the above mentioned points are enough to give you an idea how complex question “How many people speak a certain language? may get. Quoted speaker numbers are often inaccurate and do not tell the whole truth. That is why they can vary so much between different sources. To return to Halmahera: Regardless of exactly how many speakers the languages have, there is still enough room for them.

Literature

https://en.wikipedia.org/wiki/Halmahera

colonial reports about Indonesia (in Dutch): https://kitlv-docs.library.leiden.edu/open/Metamorfoze/Kol.%20Verslag/koloniaal%20verslag.html (the numbers for Halmahera can be found in the report for 1923, p. 10 of Appendix A (p. 150 in the PDF) under „Ternate en onderhoorigheden“)

Ipol, Robinson, Howard Shelden, Deidre B. Shelden & Yosafat Etha. 1989. O Galelaka yomatekebobicara = Percakapan-percakapan bahasa Galela = Galela conversations. Ambon: Universitas Pattimura and Summer Institute of Linguistics.

Holton, Gary. 2003. Tobelo (Languages of the World/Materials 328). München: Lincom.

Visser, Leontine E. & C. L. Voorhoeve. 1987. Sahu-Indonesian-English dictionary and Sahu grammar sketch (Verhandelingen van het Koninklijk Instituut voor Taal-, Land- en Volkenkunde 126). Dordrecht, Providence: Foris Publications.

Evans, Nicholas. 2001. „The last speaker is dead – long live the last speaker“. In P. Newman & M. Ratliff (Hrsg.), Linguistic Fieldwork.  Cambridge, Cambridge University Press. 250-281.

Rumsay, Alan. 1993. “Language and Territoriality in Aboriginal Australia.” In Michael Walsh & Colin Yallop (Hrsg.), Language and Culture in Aboriginal Australia. Canberra: Aboriginal Studies Press. 

Speas, Margaret. 2013. “Language ownership and language ideologies.” In Laetitia La Follette (Hrsg.), Negotiating culture: heritage, ownership, and intellectual property. 101-121.

Attribution

Sadalmelik, CC BY-SA 3.0 https://creativecommons.org/licenses/by-sa/3.0, via Wikimedia Commons

---

---